chunk_uid
stringlengths
40
40
chunk_type
stringclasses
2 values
chunk_index
int64
0
6.71k
total_chunks
int64
1
6.71k
section_title
stringlengths
1
157
embed_text
stringlengths
1
83.3k
spans
dict
paper_doi
stringlengths
0
63
paper_id_arxiv
stringlengths
9
16
title
stringlengths
7
245
authors
listlengths
1
768
categories
listlengths
1
7
year
int64
2k
2.02k
language
stringclasses
2 values
discipline
stringclasses
8 values
sparse_indices
listlengths
1
1.02k
sparse_values
listlengths
1
1.02k
42a7341fe75822d745ddec94f91a0a15c565485f
subsection
9
63
Approximate Theory and Method
This motivates our approximate methods.We note that given {\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2,{y\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }}, the i are independent with \begin{equation} \pi (\mu _i \mid {\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2,{y)\crcr \vbox to....
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 3293, 167729, 90, 2446, 35707, 5134, 15829, 150624, 12137, 20537, 34475, 59865, 6, 23150, 334, 11728, 47, 304, 3355, 24854, 41872, 127, 4369, 3675, 112, 8152, 18043, 7, 47391, 1743, 102, 13331, 53, 5, 70, 17, 621, 41371, 678, 372, 6820,...
[ 0.04058837890625, 0.2064208984375, 0.068603515625, 0.070068359375, 0.139404296875, 0.19091796875, 0.145751953125, 0.1728515625, 0.005279541015625, 0.126953125, 0.07733154296875, 0.2666015625, 0.007659912109375, 0.119384765625, 0.057647705078125, 0.153564453125, 0.092529296875, 0.12...
adb0877f4b08d339b98ae5e7e244a523678d1277
subsection
10
63
Approximation to the Posterior Density
In this section we discuss the approximation to the joint posterior density in (REF ).Let f({\tau )\crcr \vbox to.2ex{\hbox{$\tau \tilde{}$}\vss }} = e^{h({\tau )\crcr \vbox to.2ex{\hbox{$\tau \tilde{}$}\vss }}} denote the density of a vector of parameters {\tau \crcr \vbox to.2ex{\hbox{$\tau \tilde{}$}\vss }}. Let g\c...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 903, 40059, 642, 45252, 35707, 53950, 2320, 47, 124416, 39225, 168, 7, 2481, 23, 11766, 919, 124480, 1238, 50104, 23150, 334, 11728, 304, 3355, 3675, 112, 18043, 28, 127, 8, 48345, 173, 18770, 171859, 10842, 706, 177, 35358, 2517, 136, ...
[ 0.0548095703125, 0.1182861328125, 0.00250244140625, 0.1328125, 0.1544189453125, 0.2320556640625, 0.0221710205078125, 0.099853515625, 0.259765625, 0.307373046875, 0.261474609375, 0.1605224609375, 0.1304931640625, 0.06103515625, 0.139404296875, 0.2020263671875, 0.098388671875, 0.1833...
63b86f6be8d350be96c640c39c0627ca5c80ebff
subsection
11
63
Approximation to the Posterior Density
Then, \tau \crcr \vbox to.2ex{\hbox{$\tau \tilde{}$}\vss } approximately has a multivariate normal distribution,{\tau \stackrel{}{\crcr }\vbox to.2ex{\hbox{$\tau \tilde{}$}\vss }}{}{\sim }\mbox{Normal} ({\tau ^\crcr \vbox to.2ex{\hbox{$\tau \tilde{}$}\vss }}*-H^{-1}{g,\crcr \vbox to.2ex{\hbox{$g\tilde{}$}\vss }}-H^{-1}...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 47009, 41872, 50104, 23150, 334, 11728, 47, 304, 3355, 127, 4369, 6, 3675, 112, 8152, 18043, 51912, 189275, 1556, 10, 6024, 124640, 67, 3638, 113068, 590, 7962, 24854, 5, 7, 47391, 5072, 50645, 2749, 15, 9, 841, 8353, 5759, 177, 4, 62...
[ 0.024505615234375, 0.0667724609375, 0.338623046875, 0.187255859375, 0.1102294921875, 0.2218017578125, 0.152587890625, 0.10760498046875, 0.18017578125, 0.03826904296875, 0.004669189453125, 0.00457763671875, 0.2115478515625, 0.261962890625, 0.004730224609375, 0.1905517578125, 0.0042114...
b991df9707015e96fef4cd7ea95dfd08b1d01235
subsection
12
63
Approximation to the Posterior Density
That is,y_{ij}|\mu _i,{\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)} \stackrel{ind}{\sim } \mbox{Bernoulli} \left\lbrace \frac{e^{{x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij}^{\prime }{\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}+\mu _i}}{1+e^{{x_\crcr \vbox to.2ex{\hbox{$x\tilde...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 9925, 83, 53, 454, 13786, 58745, 561, 101, 14, 59865, 23150, 6, 334, 11728, 47, 5, 304, 3355, 24854, 41872, 127, 4369, 3675, 112, 8152, 18043, 7, 47391, 177609, 590, 2594, 7962, 7250, 5072, 51912, 39, 33189, 3165, 7532, 133, 2480, 994...
[ 0.08270263671875, 0.06671142578125, 0.1514892578125, 0.0445556640625, 0.22900390625, 0.046295166015625, 0.180419921875, 0.0362548828125, 0.08453369140625, 0.255859375, 0.136962890625, 0.036041259765625, 0.1065673828125, 0.2010498046875, 0.156494140625, 0.03875732421875, 0.11169433593...
34c2c34f69bc523ac9eb580223effbd4cadc6003
subsection
13
63
Approximation to the Posterior Density
First, we find a convenient point to expand the log-likelihood in a multivariate Taylor's series expansion. In Appendix B, we show how to obtain quasi-modes for {\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)} and \mu _i, i=1,\ldots ,\ell , of the log-likelihood function.First, we use the empirical logist...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 23972, 642, 7413, 142267, 6275, 47, 71062, 12684, 9, 5062, 150, 44462, 6024, 124640, 67, 63569, 25, 36549, 14700, 66, 6889, 360, 5659, 17215, 425, 335, 7639, 3642, 113054, 12404, 61170, 59865, 101, 23150, 6, 334, 11728, 304, 3355, 24854, ...
[ 0.05535888671875, 0.015625, 0.03839111328125, 0.1597900390625, 0.08978271484375, 0.1182861328125, 0.2125244140625, 0.2130126953125, 0.10205078125, 0.2119140625, 0.202880859375, 0.1444091796875, 0.1285400390625, 0.1798095703125, 0.021270751953125, 0.247802734375, 0.078857421875, 0.1...
d1f1497862b9b35c9432c0f8b3ff30aeae92c1d6
subsection
14
63
Approximation to the Posterior Density
We use the log-likelihood of the \mu _i with the regression coefficients replaced by {\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}^\ast , and solve its first derivative for zeros using a first-order Taylor's series expansion to get{\mu _i}^* = \log \left[ \frac{\frac{1}{n_i}\sum _{j=1}^{n_i}e^{-{x_\crc...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1401, 4527, 12684, 9, 5062, 150, 44462, 111, 70, 6, 41872, 561, 101, 14, 678, 456, 148448, 552, 4240, 11044, 35066, 91995, 71, 10666, 59865, 23150, 334, 11728, 47, 5, 304, 3355, 24854, 127, 4369, 3675, 112, 8152, 18043, 7, 47391, 1776...
[ 0.0911865234375, 0.12060546875, 0.190673828125, 0.05108642578125, 0.2130126953125, 0.19140625, 0.143310546875, 0.00732421875, 0.051727294921875, 0.051971435546875, 0.08343505859375, 0.2529296875, 0.1378173828125, 0.18115234375, 0.06500244140625, 0.095458984375, 0.274169921875, 0.08...
71b7c46b1318716cd77ef17ba36b4af265564427
subsection
15
63
Approximation to the Posterior Density
The partial derivatives can be expressed in terms of response yij and covariates x_x\tilde{}ij as \frac{\partial \Delta }{\partial {\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}}=\sum _{i=1}^\ell \sum _{j=1}^{n_i} ( {x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij} y_{ij}-\frac{{x_\crcr \vbox to.2ex{...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 581, 2878, 289, 30057, 42991, 831, 186, 36510, 297, 23, 69407, 111, 57553, 113, 13786, 136, 552, 124640, 1636, 1022, 454, 425, 3675, 112, 8152, 237, 6, 41872, 132076, 24854, 15866, 58598, 102, 51912, 10666, 59865, 101, 23150, 334, 11728, ...
[ 0.07513427734375, 0.25634765625, 0.183349609375, 0.281005859375, 0.2176513671875, 0.1668701171875, 0.028472900390625, 0.252685546875, 0.1524658203125, 0.028411865234375, 0.11376953125, 0.02874755859375, 0.253173828125, 0.076171875, 0.2032470703125, 0.0958251953125, 0.11328125, 0.23...
3d1b7f26f882042f90efdc342a4036073c581c25
subsection
16
63
Approximation to the Posterior Density
For the convenience of computation, denote {g=\crcr \vbox to.2ex{\hbox{$g\tilde{}$}\vss }}\left(\begin{array}{c} {g_\crcr \vbox to.2ex{\hbox{$g\tilde{}$}\vss }}1\\{g_\crcr \vbox to.2ex{\hbox{$g\tilde{}$}\vss }}2\end{array}\right) and H=-\left(\begin{array}{c} D\quad C^{\prime }\\C \quad B\end{array}\right), where{g_\cr...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1326, 50080, 6620, 181135, 1363, 8, 48345, 177, 1369, 23150, 6, 334, 11728, 47, 304, 3355, 41872, 127, 3675, 112, 24854, 8152, 4369, 18043, 7, 47391, 133, 2480, 132, 372, 6820, 19305, 53, 238, 10666, 5, 418, 13273, 454, 3611, 54969, 1...
[ 0.040191650390625, 0.1224365234375, 0.032867431640625, 0.1826171875, 0.020294189453125, 0.1815185546875, 0.2491455078125, 0.172607421875, 0.032562255859375, 0.154541015625, 0.0186004638671875, 0.1256103515625, 0.23046875, 0.177734375, 0.0770263671875, 0.1593017578125, 0.0164337158203...
f0b0295510663d886b6980929a4dc90d3e2c0aa0
subsection
17
63
Approximation to the Posterior Density
Then, A, B and C of the negative Hessian matrix can be written as,B = \frac{\partial ^2\Delta }{\partial {\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}^2}=\sum _{i=1}^\ell \sum _{j=1}^{n_i} p_{ij}(1-p_{ij}){x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij}{x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 62, 4, 335, 136, 313, 40907, 1529, 7, 39209, 50944, 425, 831, 186, 59121, 237, 571, 2203, 132076, 15866, 289, 304, 58598, 102, 59865, 23150, 334, 11728, 47, 3355, 127, 3675, 112, 18043, 177609, 11832, 6796, 170, 19, 915, 13786, 254, 1...
[ 0.09423828125, 0.02850341796875, 0.1151123046875, 0.085693359375, 0.12890625, 0.25048828125, 0.1651611328125, 0.1160888671875, 0.1810302734375, 0.2005615234375, 0.1043701171875, 0.056243896484375, 0.0204010009765625, 0.197998046875, 0.087890625, 0.1719970703125, 0.067626953125, 0.1...
d9cbca8dee8fefef90cb221a85e8bfcaf9e4b2d4
subsection
18
63
Approximation to the Posterior Density
Then, the Schur complement isS = \sum _{i=1}^\ell \sum _{j=1}^{n_i} p_{ij}(1-p_{ij}) \sum _{j=1}^{n_i} \omega _{ij} {x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij} {x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij}^\prime - \sum _{i}^\ell \sum _{j=1}^{n_i} p_{ij}(1-p_{ij}) \sum _{j=1}^{n_i} \omega _{ij} {x_\crcr \...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 47009, 70, 42763, 42, 110633, 83, 294, 2203, 11832, 24854, 14, 33000, 8152, 8353, 6796, 6, 170, 19, 915, 454, 13786, 41600, 9, 254, 16, 101, 41872, 306, 2765, 10666, 23150, 11728, 47, 5, 304, 3355, 127, 4369, 425, 3675, 112, 18043, ...
[ 0.06365966796875, 0.0810546875, 0.2293701171875, 0.23681640625, 0.309814453125, 0.0986328125, 0.2083740234375, 0.126220703125, 0.261474609375, 0.015716552734375, 0.045196533203125, 0.1849365234375, 0.0157470703125, 0.015411376953125, 0.2137451171875, 0.015228271484375, 0.135009765625...
13cf8d89d4f42735cb969a5379d1545b7866b69e
subsection
19
63
Approximation to the Posterior Density
By Lemma 2.1 the posterior density is approximately a multivariate normal density.
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 3311, 636, 18023, 34513, 70, 39225, 168, 7, 2481, 83, 189275, 10, 6024, 124640, 67, 3638, 5 ]
[ 0.0859375, 0.0684814453125, 0.2071533203125, 0.2174072265625, 0.06549072265625, 0.34326171875, 0.244140625, 0.132080078125, 0.1318359375, 0.1317138671875, 0.217529296875, 0.0496826171875, 0.14306640625, 0.1845703125, 0.07440185546875, 0.18212890625, 0.006805419921875 ]
3220d395713cf957729168c95c6cf8a6fba4c248
subsection
20
63
Approximation to the Posterior Density
We provide the approximate mean and variance to completely specify the multivariate normal density.By Lemma 2.1, evaluating all appropriate quantities at {\tau ^\crcr \vbox to.2ex{\hbox{$\tau \tilde{}$}\vss }}\ast , the posterior mean is\left(\begin{array}{c}{\mu _\crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }}{\mu }\\...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1401, 22691, 35707, 5134, 15829, 29459, 136, 16757, 3956, 64557, 40140, 18929, 70, 6024, 124640, 67, 3638, 168, 7, 2481, 75358, 18023, 34513, 151575, 756, 95307, 102134, 2449, 99, 50104, 23150, 6, 334, 11728, 47, 5, 304, 3355, 24854, 4187...
[ 0.035186767578125, 0.110107421875, 0.1375732421875, 0.1748046875, 0.1092529296875, 0.242919921875, 0.097412109375, 0.261474609375, 0.10498046875, 0.1348876953125, 0.1689453125, 0.06597900390625, 0.040252685546875, 0.203369140625, 0.26513671875, 0.09814453125, 0.2359619140625, 0.254...
266fd0090121def361b639d0e391f969c2a82cfa
subsection
21
63
Integrated Nested Normal Approximation
In this section, we obtain the integrated nested normal approximation (INNA). INNA, which does not require posterior modes, is competitive to the integrated nested Laplace approximation (INLA) that requires posterior modes.Next, using the normal priors for the \mu _i and Theorem 1, we have the following approximate hie...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 360, 903, 40059, 642, 113054, 70, 78779, 297, 8512, 71, 3638, 35707, 53950, 2320, 15, 89882, 284, 194, 5881, 6691, 14602, 959, 64209, 39225, 13736, 7, 83, 168296, 47, 239, 23935, 8852, 8356, 144570, 81280, 17368, 41928, 6, 41872, 561, 1...
[ 0.03515625, 0.0843505859375, 0.1138916015625, 0.05657958984375, 0.1497802734375, 0.033416748046875, 0.2125244140625, 0.10491943359375, 0.2337646484375, 0.119873046875, 0.2152099609375, 0.1312255859375, 0.2003173828125, 0.0731201171875, 0.033355712890625, 0.2393798828125, 0.2790527343...
fc75a39848e21ed4e966ae783dff0f364b29ddd1
subsection
22
63
Integrated Nested Normal Approximation
Then, using Bayes' theorem again, the approximate joint posterior density for the parameters {\mu ,\crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }}{\beta \crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }} and 2 is \pi _a({\mu ,\crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }}{\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 47009, 17368, 9631, 90, 25, 70, 58391, 13438, 35707, 5134, 15829, 124416, 39225, 168, 7, 2481, 100, 171859, 561, 6, 23150, 334, 11728, 47, 304, 3355, 24854, 41872, 127, 4369, 3675, 112, 8152, 18043, 47391, 59865, 5, 136, 116, 83, 1434, ...
[ 0.057098388671875, 0.0938720703125, 0.212890625, 0.1986083984375, 0.0616455078125, 0.0572509765625, 0.20361328125, 0.1165771484375, 0.1470947265625, 0.1846923828125, 0.14501953125, 0.299072265625, 0.331298828125, 0.271484375, 0.1673583984375, 0.197265625, 0.061126708984375, 0.27270...
ae5eca1173bc4a2583d7df591058821ee7648b67
subsection
23
63
Integrated Nested Normal Approximation
\begin{equation} \times \frac{|D|^{1/2}}{{|\delta ^2I|}^{1/2}|G|^{1/2}}\times \frac{1}{(1+\delta ^2)^2}\times e^{-\frac{1}{2}\left[{\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}-{\mu _\crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }}{\beta }\right]^{\prime }G^{-1}\left[{\beta _\crcr \vbox to.2ex{\hbox{$\...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 372, 6820, 13, 5490, 2320, 41872, 70141, 132076, 58745, 397, 8353, 118551, 172162, 1743, 102, 13331, 304, 568, 724, 418, 41600, 1328, 10461, 28, 2480, 59865, 23150, 334, 11728, 47, 3355, 3675, 112, 18043, 6, 177609, 561, 54969, 114654, 57...
[ 0.006378173828125, 0.100830078125, 0.0465087890625, 0.2149658203125, 0.10711669921875, 0.049774169921875, 0.2401123046875, 0.18896484375, 0.053863525390625, 0.167724609375, 0.018646240234375, 0.21240234375, 0.009857177734375, 0.137451171875, 0.166259765625, 0.00799560546875, 0.182250...
bfba56b5aa2ead0ad75f1a7e6f37ce005589505e
subsection
24
63
Integrated Nested Normal Approximation
\end{equation} }Next, we state the main result of the paper in Theorem 2.2.Theorem 3.2 Using the multiplication rule, the joint posterior density, \pi ({\mu ,\crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }}{\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2 \mid {y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vs...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 3611, 13, 5490, 2320, 81280, 642, 11341, 5201, 16750, 15122, 581, 58391, 41187, 3957, 55054, 345, 6953, 127664, 1363, 79986, 70, 124416, 39225, 168, 7, 2481, 6, 1434, 561, 41872, 23150, 334, 11728, 47, 304, 3355, 24854, 127, 4369, 3675, ...
[ 0.0460205078125, 0.0211334228515625, 0.1864013671875, 0.025970458984375, 0.071044921875, 0.0377197265625, 0.1260986328125, 0.15625, 0.200927734375, 0.1829833984375, 0.0863037109375, 0.2073974609375, 0.173095703125, 0.06475830078125, 0.2174072265625, 0.045166015625, 0.0257720947265625...
1d1d0ac357103422b02adf417cdbea682b17223e
subsection
25
63
Integrated Nested Normal Approximation
Because (D+\frac{1}{\delta ^2}I) is diagonal, given {\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2,{y\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }}, the i are independent. This is an important result because parallel computation can be done for i, which accommodates time-consuming and massive storage ...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 88949, 397, 1328, 132076, 418, 1743, 102, 13331, 304, 568, 83, 207997, 34475, 59865, 23150, 334, 11728, 47, 3355, 24854, 3675, 112, 18043, 53, 4, 70, 17, 621, 41371, 5526, 16750, 6637, 95103, 181135, 831, 16940, 211196, 1733, 106750, 1320...
[ 0.1064453125, 0.1787109375, 0.262451171875, 0.1795654296875, 0.1015625, 0.140380859375, 0.184814453125, 0.034210205078125, 0.1790771484375, 0.226318359375, 0.1097412109375, 0.279541015625, 0.073486328125, 0.2125244140625, 0.090087890625, 0.038848876953125, 0.16162109375, 0.10070800...
3f5381d8c91a0c0a0b3ad9c27fa8c73547c1580b
subsection
26
63
Integrated Nested Normal Approximation
Second, because {\mu \crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }} has a multivariate normal distribution, we can integrate out \mu \crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss } from the joint posterior density \pi _a({\mu ,\crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }} {\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tild...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 77648, 6637, 41872, 561, 6, 23150, 334, 11728, 47, 304, 3355, 127, 3675, 112, 8152, 4369, 18043, 7, 47391, 1556, 10, 6024, 124640, 67, 3638, 113068, 642, 831, 78779, 13, 1810, 24854, 51912, 1295, 70, 124416, 39225, 168, 2481, 1434, 11, ...
[ 0.1884765625, 0.1339111328125, 0.0241851806640625, 0.2352294921875, 0.023468017578125, 0.1663818359375, 0.1138916015625, 0.22412109375, 0.16455078125, 0.0955810546875, 0.171630859375, 0.038421630859375, 0.2296142578125, 0.271240234375, 0.0242156982421875, 0.0240631103515625, 0.219970...
403148f2e1f30147749335dfd6710bafab5dc2fd
subsection
27
63
Integrated Nested Normal Approximation
\right. and } \pi ({\beta |\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2,{y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }} \propto e^{-\frac{1}{2} \left[{\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}-{\mu _\crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }}{\beta }\right]^{\prime }G^{-1}\left[{\b...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 54969, 136, 51912, 41872, 1434, 15, 59865, 58745, 23150, 334, 11728, 47, 304, 3355, 127, 3675, 112, 8152, 4369, 18043, 47391, 1743, 102, 53, 16, 5, 24854, 36290, 188, 28, 8353, 132076, 418, 133, 2480, 6, 7, 177609, 561, 114654, 724, 5...
[ 0.146484375, 0.21044921875, 0.0159912109375, 0.03509521484375, 0.314453125, 0.039276123046875, 0.28759765625, 0.050994873046875, 0.13232421875, 0.1014404296875, 0.1954345703125, 0.1365966796875, 0.1339111328125, 0.1552734375, 0.0340576171875, 0.2137451171875, 0.2403564453125, 0.000...
ebfaf77424d85065ecf7228ebafa5931af5c650c
subsection
28
63
Integrated Nested Normal Approximation
\right. }\end{equation}Let \Delta _{(0)}=CD^{-1}(D^{-1}+\delta ^2I)^{-1}D^{-1}C^{\prime }+G^{-1}, \delta ^2_0={j^{\prime }\crcr \vbox to.2ex{\hbox{$j\tilde{}$}\vss }}(D^{-1}+\delta ^2I)^{-1}{j,\crcr \vbox to.2ex{\hbox{$j\tilde{}$}\vss }} {\gamma =\crcr \vbox to.2ex{\hbox{$\gamma \tilde{}$}\vss }}CD^{-1}(D^{-1}+\de...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 41872, 54969, 3611, 13, 5490, 2320, 8152, 124480, 58598, 102, 101, 24854, 177609, 1369, 16069, 8353, 5759, 132, 397, 1328, 1743, 304, 568, 16, 441, 114654, 51912, 724, 6, 13331, 454, 2389, 170, 23150, 334, 11728, 47, 3355, 127, 4369, 36...
[ 0.021331787109375, 0.1441650390625, 0.074462890625, 0.11474609375, 0.26416015625, 0.09893798828125, 0.0350341796875, 0.1553955078125, 0.279296875, 0.2264404296875, 0.0745849609375, 0.016357421875, 0.2218017578125, 0.07568359375, 0.1959228515625, 0.09478759765625, 0.197265625, 0.015...
c0e451968f1755051d2c6c0a730b511dcb44e457
subsection
29
63
Integrated Nested Normal Approximation
After extensive algebraic manipulation, we can show that {\beta |\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2,{y\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }} has an approximate multivariate normal density, \begin{equation} \left(\begin{array}{c} \beta _0\\{\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 24372, 1119, 41745, 272, 144, 429, 2844, 1771, 45258, 1363, 642, 831, 7639, 450, 59865, 6, 58745, 41872, 23150, 334, 11728, 47, 304, 3355, 24854, 127, 4369, 3675, 112, 8152, 18043, 7, 47391, 1743, 102, 13331, 4, 53, 5, 1556, 142, 3570...
[ 0.089599609375, 0.0231475830078125, 0.0994873046875, 0.04266357421875, 0.04827880859375, 0.1578369140625, 0.14013671875, 0.0416259765625, 0.2178955078125, 0.0662841796875, 0.06475830078125, 0.09600830078125, 0.1826171875, 0.011474609375, 0.30908203125, 0.042083740234375, 0.0422668457...
1c23f4e5e0db8a8027e1e6e8a15b7ee82dba64cf
subsection
30
63
Integrated Nested Normal Approximation
Samples are obtained by first drawing from \pi _a(\delta ^2|{y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }}, then \pi _a({\beta \mid \crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2, {y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }} and finally \pi _a({\mu \mid \crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }}{\b...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 3362, 99726, 621, 113054, 297, 390, 5117, 188304, 1295, 1434, 101, 11, 1743, 102, 304, 53, 23150, 334, 11728, 47, 3355, 127, 3675, 112, 18043, 47391, 4, 7068, 59865, 22000, 136, 77681, 561, 6626, 168, 7, 31075, 4527, 5570, 123840, 50960...
[ 0.284423828125, 0.29443359375, 0.1156005859375, 0.2130126953125, 0.130859375, 0.107666015625, 0.1083984375, 0.2393798828125, 0.0748291015625, 0.22509765625, 0.027923583984375, 0.1392822265625, 0.1617431640625, 0.155029296875, 0.1451416015625, 0.1080322265625, 0.0980224609375, 0.073...
36b432ab020c5172b4e1f255a3841cd320a17fad
subsection
31
63
Integrated Nested Normal Approximation
Then, posterior density, \pi (\eta |{y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }}, is\pi (\eta |{y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }}\propto \left\lbrace \frac{1}{|\delta ^2D+I|^{1/2}}\left|\left(\begin{array}{c} \delta ^2_0 \quad {\gamma ^{\prime }\crcr \vbox to.2ex{\hbox{$\gamma \tilde{}$}\vss }}\\ {\gamm...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 47009, 39225, 168, 7, 2481, 41872, 1434, 4241, 6, 58745, 53, 16, 23150, 334, 11728, 47, 304, 3355, 24854, 127, 3675, 112, 8152, 4369, 18043, 47391, 83, 36290, 188, 133, 2480, 48543, 99407, 132076, 418, 1743, 102, 13331, 397, 1328, 8353,...
[ 0.0921630859375, 0.34521484375, 0.281982421875, 0.194580078125, 0.185546875, 0.04095458984375, 0.261474609375, 0.2059326171875, 0.040679931640625, 0.039886474609375, 0.112548828125, 0.0406494140625, 0.1494140625, 0.10467529296875, 0.2305908203125, 0.12109375, 0.1241455078125, 0.167...
50b62e34487b666cf7987f0f7f02898417d2029a
subsection
32
63
Integrated Nested Normal Approximation
Equation (REF ) is the basis of our INNA approximation. This is simply the multiplication rule of probability; simply draw \eta from (REF ) and retransform to \delta ^2 to get \pi _a(\delta ^2 \mid {y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }}, draw {\beta \crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }} from a(\bet...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 241, 5490, 2320, 11766, 919, 1388, 83, 70, 18231, 111, 2446, 5881, 6691, 35707, 53950, 5, 3293, 42856, 127664, 1363, 79986, 37242, 2481, 74, 79442, 6, 4241, 1295, 456, 215175, 47, 1743, 102, 304, 2046, 1434, 101, 11, 132, 41872, 22000, ...
[ 0.066162109375, 0.257568359375, 0.1541748046875, 0.17578125, 0.2469482421875, 0.024139404296875, 0.08001708984375, 0.056304931640625, 0.213134765625, 0.08160400390625, 0.12255859375, 0.1978759765625, 0.2880859375, 0.17138671875, 0.2393798828125, 0.024749755859375, 0.0721435546875, ...
3ebcab434a8d09cd9636004a17535c4c4eefa6b6
subsection
33
63
Comparison of the Two Methods
As a summary, we compare the approximate and exact methods. The exact method is given in Appendix A.First, we note that the exact method actually uses the approximate method. We use a Metropolis step with \pi _a({\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2|{y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1301, 10, 177074, 4, 642, 69101, 70, 35707, 5134, 15829, 136, 24763, 150624, 5, 55300, 83, 34475, 23, 5659, 17215, 425, 62, 207263, 20537, 450, 20653, 4527, 7, 1401, 117366, 164, 29954, 678, 6, 41872, 1434, 101, 11, 132, 24854, 59865, ...
[ 0.0163726806640625, 0.050933837890625, 0.132568359375, 0.0511474609375, 0.06732177734375, 0.234375, 0.06475830078125, 0.213623046875, 0.257568359375, 0.281494140625, 0.1314697265625, 0.3408203125, 0.2919921875, 0.051116943359375, 0.275146484375, 0.05108642578125, 0.087158203125, 0....
e9e59ee2c2dce780a9c06da6aeca3cb304ed7f6c
subsection
34
63
Comparison of the Two Methods
Third, for prediction two Bayesian bootstraps are used to get the nonsampled household sizes and the nonsampled covariates (\approx two million people). This is done within wards.
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 198387, 4, 100, 92054, 1830, 6626, 9631, 90, 3378, 49935, 2816, 6423, 621, 11814, 47, 2046, 351, 7, 25133, 6259, 197540, 13267, 136, 552, 124640, 1636, 41872, 64101, 19879, 3395, 194, 3293, 83, 16940, 28032, 19364 ]
[ 0.1883544921875, 0.018402099609375, 0.0841064453125, 0.2415771484375, 0.14501953125, 0.1195068359375, 0.173828125, 0.1370849609375, 0.15234375, 0.1453857421875, 0.1756591796875, 0.152099609375, 0.0540771484375, 0.129150390625, 0.038848876953125, 0.1148681640625, 0.1527099609375, 0....
bfd43dcf8861ed5b181ac314d1e4d8dd67980bd8
subsection
35
63
Illustrative Example
In Section 3.1, we briefly describe the Nepal Living Standards Survey (NLSS II) and in Section 3.2, we use the health status data with five covariates to compare our approximate Bayesian logistic regression method with the exact one.
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 360, 140978, 45151, 642, 59335, 98363, 56250, 43722, 20744, 7, 181842, 39102, 15268, 1995, 23, 55054, 4527, 16227, 10778, 2053, 678, 43606, 552, 124640, 1636, 47, 69101, 2446, 35707, 5134, 15829, 9631, 90, 3378, 150535, 456, 148448, 55300, ...
[ 0.0193939208984375, 0.1177978515625, 0.1876220703125, 0.024383544921875, 0.08721923828125, 0.113037109375, 0.25927734375, 0.1790771484375, 0.1754150390625, 0.00677490234375, 0.1856689453125, 0.1669921875, 0.22509765625, 0.1932373046875, 0.009246826171875, 0.1961669921875, 0.085327148...
cd62c9d4b1609bdf5f892900f39b0e257606f902
subsection
36
63
Nepal Living Standards Survey
We use data from the Nepal Living Standards Survey (NLSS II, Central Bureau of Statistics, 2003-2004) to illustrate INNA with logistic regression. NLSS is a national household survey in Nepal, actually population based (i.e., interviews are done for individual household members). NLSS follows the World Bank's Living St...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1401, 4527, 2053, 1295, 56250, 43722, 20744, 7, 181842, 39102, 15268, 1995, 15881, 86907, 111, 126124, 28021, 4, 6052, 30416, 16, 58755, 67, 5881, 6691, 678, 150535, 456, 148448, 83, 10, 15889, 197540, 110297, 23, 20653, 43904, 35509, 15, ...
[ 0.023284912109375, 0.1011962890625, 0.147705078125, 0.0382080078125, 0.298095703125, 0.1900634765625, 0.1790771484375, 0.03826904296875, 0.203125, 0.243896484375, 0.28759765625, 0.17236328125, 0.0256195068359375, 0.09674072265625, 0.038238525390625, 0.094970703125, 0.03814697265625, ...
cdd08e6e739d58f9035e7d9e10feba07ab3ec8e5
subsection
37
63
Nepal Living Standards Survey
We use Bayesian bootstraps (Rubin 1981) for unknown household sizes and nonsampled covariates; the bootstrapping is done within wards. The 2001 Census can potentially provide these two pieces of information, but there is a mis-match between the households in the census and the NLSS (a record linkage can be performed). ...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1401, 4527, 9631, 90, 3378, 49935, 2816, 6423, 26951, 7568, 26771, 100, 51, 69723, 197540, 13267, 136, 351, 25133, 6259, 552, 124640, 1636, 26783, 16940, 28032, 19364, 6789, 51292, 7921, 831, 38516, 22691, 6626, 126371, 4677, 1284, 83, 1239...
[ 0.0738525390625, 0.1378173828125, 0.171630859375, 0.1416015625, 0.13525390625, 0.1881103515625, 0.208740234375, 0.158935546875, 0.0579833984375, 0.1295166015625, 0.1514892578125, 0.051910400390625, 0.06427001953125, 0.177734375, 0.1719970703125, 0.1510009765625, 0.02423095703125, 0...
da8e0d0f3b74d559f93617f5a741da0a92c42409
subsection
38
63
Numerical Comparisons
We predict the household proportions of members in good health for 60,221 households. This analysis is based on 3,912 sample households from 326 wards (PSUs). Our primary purpose is to compare the approximate method with the exact method when there are random effects at the household level. We want to show that one can...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1401, 92054, 197540, 123875, 7, 111, 43032, 23, 4127, 16227, 1496, 4, 304, 3117, 114137, 83, 35509, 98, 6, 125157, 1530, 121413, 138, 4046, 19364, 9059, 1062, 22929, 158978, 60042, 69101, 70, 35707, 5134, 15829, 55300, 678, 24763, 621, 96...
[ 0.0309906005859375, 0.26220703125, 0.23046875, 0.2265625, 0.068603515625, 0.008453369140625, 0.202880859375, 0.009246826171875, 0.160400390625, 0.23291015625, 0.15380859375, 0.035125732421875, 0.1104736328125, 0.1683349609375, 0.1417236328125, 0.00848388671875, 0.039794921875, 0.00...
0edc5c1fdd13324b7695bc2669bb2507efcc542f
subsection
39
63
Concluding Remarks
We make three statistical comments. First, the approximate method is necessary when there are a large number (millions) of households (clusters or areas). Second, it is difficult to use the census data effectively but it is desirable (matching problem). Third, it is possible to obtain similar approximations for spatial...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1401, 3249, 17262, 80835, 289, 24626, 4, 70, 35707, 5134, 15829, 55300, 83, 63559, 3229, 621, 10, 21334, 14012, 123342, 5245, 16, 111, 197540, 7, 15, 29786, 37499, 707, 58555, 77648, 442, 34844, 4527, 149039, 2053, 191984, 1284, 224, 481,...
[ 0.0294036865234375, 0.058258056640625, 0.1658935546875, 0.161865234375, 0.0662841796875, 0.197021484375, 0.011138916015625, 0.0108642578125, 0.139892578125, 0.1646728515625, 0.121826171875, 0.179931640625, 0.010894775390625, 0.16650390625, 0.029571533203125, 0.01104736328125, 0.01089...
95b92df183a40d00607251fdd2a08bf98c1e271e
subsection
40
63
Concluding Remarks
Letn \ge n_e = (\sum _{i=1}^n \omega _i)^2/(\sum _{i=1}^n \omega _i^2), ~~\tilde{\omega }_i = n_e \frac{\omega _i}{\sum _{i=1}^n \omega _i}, i=1,\ldots ,n;see Potthof, Woodbury and Manton (1992) for pioneering work on equivalent sample sizes.For (y_i, \tilde{\omega }_i,{x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}i),...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 10842, 19, 6, 41872, 429, 653, 454, 13, 2203, 11832, 101, 14, 33000, 8353, 306, 2765, 45792, 10461, 4, 26236, 3675, 112, 51912, 132076, 17, 30591, 21231, 13762, 18, 21676, 39076, 87180, 136, 1572, 1507, 186348, 53918, 4209, 4488, 98, 18...
[ 0.2408447265625, 0.1766357421875, 0.03411865234375, 0.1221923828125, 0.25537109375, 0.16796875, 0.1767578125, 0.2418212890625, 0.1644287109375, 0.252685546875, 0.05999755859375, 0.1337890625, 0.1986083984375, 0.0589599609375, 0.12158203125, 0.2259521484375, 0.255859375, 0.182373046...
8063471da440862d790c96b413ae19e13f2a8723
subsection
41
63
Concluding Remarks
For small areas, with (y_{ij}, \tilde{\omega }_{ij},{x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij}), ~j=1,\ldots ,n_i, ~i=1,\ldots ,\ell , we have p(y_{ij} \mid {\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }} {\nu }_i) = \frac{e^{y_{ij} ({\tilde{x}_\crcr \vbox to.2ex{\hbox{$\tilde{x}\tilde{}$}\vss }}{ij...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1326, 19336, 58555, 678, 53, 13786, 8152, 6, 3675, 112, 306, 2765, 425, 454, 23150, 41872, 334, 11728, 47, 304, 3355, 24854, 127, 4369, 18043, 7, 170, 33000, 30591, 19, 14, 4, 6796, 642, 765, 915, 22000, 10666, 59865, 5, 47391, 539, ...
[ 0.0836181640625, 0.21142578125, 0.23046875, 0.011322021484375, 0.09710693359375, 0.2149658203125, 0.00128173828125, 0.00079345703125, 0.1929931640625, 0.216796875, 0.04913330078125, 0.162841796875, 0.05908203125, 0.000732421875, 0.1356201171875, 0.0013427734375, 0.09979248046875, 0...
5b953f502497d34e72ef6c7db4a55c769534d3b7
subsection
42
63
Concluding Remarks
We assume that y_{ijk}|{\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\nu _i, \mu _{ij} \stackrel{ind}{\sim } \mbox{Bernoulli} \left\lbrace \frac{e^{{x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ijk}^{\prime }{\beta +\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\nu _i + \mu _{ij}}}{1+e^{{x_\crcr \vbox ...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1401, 41591, 450, 113, 454, 134362, 58745, 59865, 4, 23150, 334, 11728, 47, 304, 3355, 127, 3675, 112, 8152, 18043, 47391, 539, 101, 14, 561, 13786, 590, 2594, 7962, 7250, 24854, 5072, 33189, 3165, 7532, 2480, 99407, 132076, 13, 425, 43...
[ 0.0999755859375, 0.20947265625, 0.0419921875, 0.205810546875, 0.06524658203125, 0.250732421875, 0.0460205078125, 0.2890625, 0.003082275390625, 0.1802978515625, 0.1300048828125, 0.193603515625, 0.15234375, 0.1517333984375, 0.1650390625, 0.0341796875, 0.2237548828125, 0.248291015625,...
567bb2484025eab98cd67de26c48a9480d6e67e8
subsection
43
63
Concluding Remarks
For logistic regression, this research is currently in progress. }{}}\begin{} \section {Exact Method for Logistic Regression} \end{}Recall the Bayesian logistic model with covariates that we worked on with INNA method y_{ij}|\mu _i,{\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)} \stackrel{ind}{\sim } ...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1326, 150535, 456, 148448, 903, 25188, 82424, 42658, 372, 6820, 8152, 6, 58994, 19279, 47013, 74644, 100, 13146, 48242, 853, 3611, 85763, 70, 9631, 90, 3378, 3299, 678, 552, 124640, 1636, 642, 79786, 98, 5881, 6691, 55300, 113, 454, 13786...
[ 0.1014404296875, 0.2880859375, 0.1153564453125, 0.30322265625, 0.0245819091796875, 0.1416015625, 0.09942626953125, 0.09405517578125, 0.0067138671875, 0.0498046875, 0.0074462890625, 0.006866455078125, 0.06719970703125, 0.09759521484375, 0.1669921875, 0.1942138671875, 0.0555419921875, ...
1a665a871fc4bd8e425b54c6b4bd303178263fe4
subsection
44
63
Concluding Remarks
\end{equation} According to Bayes^{\prime } theorem, the joint posterior density of the parameters ({\mu ,\crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }}{\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2|{y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }} is \pi ({\mu ,\crcr \vbox to.2ex{\hbox{$\mu \tilde{...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 3611, 5490, 129551, 47, 9631, 90, 8353, 114654, 51912, 70, 58391, 124416, 39225, 168, 7, 2481, 111, 171859, 561, 6, 41872, 23150, 334, 11728, 304, 3355, 24854, 127, 4369, 3675, 112, 8152, 18043, 47391, 59865, 4, 5, 1743, 102, 53, 16, ...
[ 0.01361083984375, 0.1458740234375, 0.0648193359375, 0.1265869140625, 0.221435546875, 0.19677734375, 0.06695556640625, 0.2435302734375, 0.0148773193359375, 0.110107421875, 0.214111328125, 0.31787109375, 0.350830078125, 0.302734375, 0.214111328125, 0.1849365234375, 0.01239013671875, ...
b70265866a39c4fbcab05ecaf55afa6cf826a214
subsection
45
63
Concluding Remarks
Putting a logconcave prior on the i does not change the logconavity of (\beta ,\beta \tilde{}\mu \mid \mu \tilde{}2,y)y\tilde{} because the product of two logconcave densities is another logconcave density. In addition, logconcave densities have sub-exponential tails and their moment generating functions exist (see Dha...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 19577, 1916, 12684, 2271, 408, 272, 41928, 98, 17, 14602, 959, 15549, 61630, 53, 59865, 3675, 112, 561, 22000, 304, 6637, 12996, 6626, 168, 7, 31075, 15700, 2481, 765, 1614, 157034, 19, 80973, 308, 7870, 3095, 12663, 32354, 32316, 174122,...
[ 0.07769775390625, 0.063232421875, 0.1890869140625, 0.16357421875, 0.1983642578125, 0.2088623046875, 0.232177734375, 0.0821533203125, 0.13671875, 0.028411865234375, 0.0859375, 0.1673583984375, 0.1923828125, 0.09051513671875, 0.210205078125, 0.173583984375, 0.2109375, 0.1473388671875...
4c73135cefbdebad69cf6677281da7919ad128cf
subsection
46
63
Concluding Remarks
We apply Metropolis Hastings sampler to draw samples for parameters {\beta \crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}, 2 and \mu \crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }.The idea of exact method is to get full conditional posterior distributions for all of the parameters in the model, and then get a large ...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1401, 59911, 117366, 164, 149240, 5180, 121413, 42, 79442, 7, 100, 171859, 59865, 6, 23150, 334, 11728, 47, 5, 304, 3355, 24854, 41872, 127, 4369, 3675, 112, 8152, 18043, 47391, 4, 116, 561, 51912, 6528, 111, 24763, 55300, 83, 2046, 439...
[ 0.035858154296875, 0.145751953125, 0.1705322265625, 0.1534423828125, 0.1854248046875, 0.2421875, 0.269287109375, 0.130126953125, 0.17236328125, 0.08203125, 0.03570556640625, 0.25, 0.25341796875, 0.032562255859375, 0.1383056640625, 0.083984375, 0.1885986328125, 0.0863037109375, 0....
f0838e11f14c2a0467be358b9cfe0b25fc36a630
subsection
47
63
Concluding Remarks
Divide the integration domain to m equal intervals [t_{k-1},t_k], k=1,...,m. Let z_i=\frac{\mu _i-\beta _0}{\delta } with standard normal distribution.
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 76278, 112, 70, 157353, 77758, 47, 347, 105950, 51514, 7, 18, 92, 5759, 454, 472, 33000, 39, 5, 10842, 97, 14, 1369, 132076, 561, 9, 59865, 2389, 1743, 102, 5570, 3638, 113068 ]
[ 0.1646728515625, 0.1090087890625, 0.0258331298828125, 0.273193359375, 0.1920166015625, 0.062042236328125, 0.15087890625, 0.1734619140625, 0.2213134765625, 0.01080322265625, 0.09991455078125, 0.1173095703125, 0.114990234375, 0.0775146484375, 0.1226806640625, 0.0819091796875, 0.1280517...
74893babb36660dfa5473db386d61b22f4d8f7a1
subsection
48
63
Concluding Remarks
We get an approximate density (very accurate though),\pi ({\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2|{y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }}\propto \frac{1}{(1+\delta ^2)^2}\left(\frac{1}{\sqrt{\delta ^2}}\right)^\ell \prod _{i=1}^\ell \left\lbrace \sum _{k=1}^{m}\int _{t_{k-1}}^{t_k}\fr...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1401, 2046, 142, 35707, 5134, 15829, 168, 7, 2481, 15, 104643, 152018, 21208, 1434, 59865, 6, 23150, 334, 11728, 47, 5, 304, 3355, 24854, 41872, 4369, 3675, 112, 8152, 18043, 47391, 1743, 102, 53, 16, 36290, 188, 132076, 418, 41600, 132...
[ 0.081298828125, 0.120361328125, 0.0216064453125, 0.1463623046875, 0.1756591796875, 0.144287109375, 0.271240234375, 0.173095703125, 0.1895751953125, 0.012847900390625, 0.1102294921875, 0.2080078125, 0.0924072265625, 0.26171875, 0.259521484375, 0.0230865478515625, 0.09442138671875, 0...
6fefc64134e0692b84a46f31fbb94f56c0d86ac6
subsection
49
63
Concluding Remarks
We have the following deduction\pi ({\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2|{y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }}\approx \frac{1}{(1+\delta ^2)^2} \prod _{i=1}^\ell \left\lbrace \sum _{k=1}^{m}\frac{e^{\sum \limits _{j=1}^{n_i}({x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij}^{\p...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1401, 765, 25632, 8, 77391, 1434, 59865, 23150, 334, 11728, 47, 304, 3355, 3675, 112, 18043, 1743, 102, 53, 64101, 132076, 418, 1328, 10461, 112348, 6796, 2480, 99407, 11832, 92, 33000, 39, 13, 93343, 13786, 114654, 177609, 2943, 4288, 18...
[ 0.05206298828125, 0.0684814453125, 0.1453857421875, 0.2249755859375, 0.259033203125, 0.2452392578125, 0.2364501953125, 0.120361328125, 0.0201873779296875, 0.1480712890625, 0.0924072265625, 0.087158203125, 0.1229248046875, 0.17333984375, 0.233154296875, 0.1185302734375, 0.108154296875...
ac1f1e68462649dcc8e5fcbdd2e0438e175bba48
subsection
50
63
Concluding Remarks
Then the joint posterior density for {\beta \crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }} and 2 can be expressed as \begin{equation} \pi ({\beta ,\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\delta ^2|{y)\crcr \vbox to.2ex{\hbox{$y\tilde{}$}\vss }}\approx \frac{1}{(1+\delta ^2)^2} \prod _{i=1}^\ell \left\lbrace...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 47009, 124416, 39225, 168, 7, 2481, 59865, 6, 23150, 334, 11728, 47, 304, 3355, 41872, 127, 4369, 3675, 112, 24854, 8152, 18043, 47391, 136, 116, 831, 186, 36510, 297, 237, 372, 6820, 5490, 2320, 1434, 5, 1743, 102, 53, 16, 2631, 6410...
[ 0.0035400390625, 0.267578125, 0.319580078125, 0.263916015625, 0.17578125, 0.1781005859375, 0.2509765625, 0.0152130126953125, 0.1357421875, 0.07177734375, 0.1378173828125, 0.0948486328125, 0.0811767578125, 0.1422119140625, 0.015289306640625, 0.0308837890625, 0.0154571533203125, 0.16...
e59c4d8cc1f8bc2284295e7da20cf972366497a9
subsection
51
63
Concluding Remarks
So we take {\beta \sim \crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}\mbox{Normal}({\hat{\beta },\crcr \vbox to.2ex{\hbox{$\hat{\beta }\tilde{}$}\vss }} \sigma ^2 \hat{\Sigma }), \eta /\sigma ^2 \sim \mbox{Gamma}(\eta /2, 1/2). Tuning of the Metropolis sampler is obtained by varying \eta (e.g., \eta =8 corresponds t...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1061, 642, 5646, 59865, 5072, 23150, 6, 334, 11728, 47, 304, 3355, 41872, 127, 4369, 3675, 112, 24854, 8152, 18043, 7, 47391, 50645, 2749, 132, 2943, 51912, 4, 5, 20561, 192, 294, 872, 247, 4241, 248, 36557, 12477, 22128, 1371, 592, 1...
[ 0.0223541259765625, 0.0533447265625, 0.077880859375, 0.199462890625, 0.1922607421875, 0.093017578125, 0.005126953125, 0.0631103515625, 0.1448974609375, 0.07965087890625, 0.0653076171875, 0.1156005859375, 0.00628662109375, 0.007171630859375, 0.00537109375, 0.1551513671875, 0.216674804...
32dd8b1c1164994602f5d1497abb0a9b27a41b4a
subsection
52
63
Quasi-Modes for Logistic Regression
Now we have to specify {\beta ^\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}*_{(0)}, {\mu ^\crcr \vbox to.2ex{\hbox{$\mu \tilde{}$}\vss }}*, {g\crcr \vbox to.2ex{\hbox{$g\tilde{}$}\vss }} and H. Consider the log likelihood function f({\tau )\crcr \vbox to.2ex{\hbox{$\tau \tilde{}$}\vss }} =\log h({\tau )\crcr \vbo...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 14240, 642, 765, 47, 40140, 18929, 59865, 13331, 23150, 6, 334, 11728, 304, 3355, 41872, 127, 4369, 3675, 112, 24854, 8152, 18043, 7, 47391, 454, 177609, 561, 5, 1639, 4, 177, 136, 572, 137399, 12684, 1884, 150, 44462, 32354, 1238, 5010...
[ 0.0673828125, 0.06463623046875, 0.0792236328125, 0.1162109375, 0.202880859375, 0.099365234375, 0.276123046875, 0.0455322265625, 0.1448974609375, 0.0059814453125, 0.12744140625, 0.212646484375, 0.087158203125, 0.177734375, 0.084228515625, 0.06463623046875, 0.006317138671875, 0.18603...
2eb9aa6824fffec07a57a0eab08313e078a20f10
subsection
53
63
Quasi-Modes for Logistic Regression
Plug {\hat{\mu }_i}^* in the log likelihood function () and consider it as a function of {\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)} only,g({\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}) =\sum _{i=1}^\ell \sum _{j=1}^{n_i} \left\lbrace ({x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 194351, 41872, 2943, 561, 454, 14, 1639, 23, 12684, 1884, 150, 44462, 32354, 16, 136, 16916, 442, 237, 59865, 101, 23150, 6, 334, 11728, 47, 304, 3355, 24854, 127, 4369, 3675, 112, 8152, 18043, 7, 47391, 177609, 4734, 177, 132, 5, 220...
[ 0.1990966796875, 0.027435302734375, 0.1964111328125, 0.1947021484375, 0.026702880859375, 0.1168212890625, 0.038330078125, 0.011993408203125, 0.1883544921875, 0.1932373046875, 0.1868896484375, 0.1435546875, 0.209228515625, 0.0267181396484375, 0.0298004150390625, 0.1024169921875, 0.001...
ff8344cc02b5096b99bf4f94af720e2127d80c34
subsection
54
63
Quasi-Modes for Logistic Regression
We use first order Taylor series approximation to simplify the above function. Since the first order Taylor expansion of (1+e^{-({x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij}^{\prime }{\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}+\hat{\mu }_i^*)} )^{-1} equals (1-e^{-({x_\crcr \vbox to.2ex{\hbox{...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 1401, 4527, 5117, 12989, 63569, 36549, 35707, 53950, 2320, 112892, 18929, 36917, 32354, 14700, 66, 6889, 111, 2858, 1328, 13, 8353, 24854, 9, 132, 425, 454, 23150, 6, 334, 11728, 47, 304, 3355, 41872, 127, 4369, 3675, 112, 8152, 18043, ...
[ 0.050567626953125, 0.087158203125, 0.108154296875, 0.19580078125, 0.3017578125, 0.21240234375, 0.1180419921875, 0.212890625, 0.047393798828125, 0.2120361328125, 0.153076171875, 0.10455322265625, 0.2008056640625, 0.1422119140625, 0.2144775390625, 0.1429443359375, 0.011444091796875, ...
81f331799ea639456821d919e739f812d51fde5a
subsection
55
63
Quasi-Modes for Logistic Regression
Thus (REF ) approximately equals\sum _{i=1}^\ell \sum _{j=1}^{n_i} \left\lbrace {x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij} y_{ij}-{x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij} (1-(1-({x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij}^{\prime }{\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 12613, 7, 15, 11766, 919, 1388, 189275, 105950, 41872, 11832, 14, 33000, 8152, 6796, 6, 24854, 170, 8353, 19, 454, 2480, 48543, 99407, 425, 23150, 334, 11728, 47, 304, 3355, 127, 3675, 112, 4369, 18043, 47391, 13786, 113, 9, 5, 2858, ...
[ 0.1422119140625, 0.08984375, 0.0819091796875, 0.231689453125, 0.3251953125, 0.100341796875, 0.291748046875, 0.25048828125, 0.018035888671875, 0.281982421875, 0.117919921875, 0.18212890625, 0.0188446044921875, 0.260009765625, 0.060791015625, 0.01898193359375, 0.17333984375, 0.018234...
405c62fdb9d1e5246e0136161ac8a72e1ba69496
subsection
56
63
Quasi-Modes for Logistic Regression
Solve for g^{\prime }({\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)})=0, and we can get the approximate posterior mode of {\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}{\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}^*= \left[ \sum _{i=1}^\ell \sum _{j=1}^{n_i}{x_\crcr \vbo...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 6678, 272, 100, 706, 8353, 24854, 114654, 59865, 101, 23150, 6, 334, 11728, 47, 304, 3355, 41872, 127, 4369, 3675, 112, 8152, 18043, 7, 47391, 177609, 16, 145407, 831, 2046, 70, 35707, 5134, 15829, 39225, 13736, 5, 1639, 1369, 133, 1065...
[ 0.2086181640625, 0.2047119140625, 0.1552734375, 0.1998291015625, 0.11328125, 0.0242767333984375, 0.26025390625, 0.264892578125, 0.0205841064453125, 0.19189453125, 0.019561767578125, 0.1431884765625, 0.2298583984375, 0.1597900390625, 0.10040283203125, 0.18896484375, 0.0196990966796875...
65cc02ae04e1476c0117c9271a50f89688b6ccf8
subsection
57
63
Quasi-Modes for Logistic Regression
The first derivative function of q(i) over i is \begin{equation} q^{\prime }(\mu _i)=\sum _{j=1}^{n_i} \left\lbrace y_{ij}-\frac{e^{({x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij}^{\prime }{\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}^*+\mu _i)}}{1+e^{{x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss ...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 581, 5117, 30057, 4935, 32354, 111, 8096, 132, 14, 16, 645, 17, 83, 6820, 5490, 114654, 561, 101, 41872, 11832, 24854, 170, 33000, 8152, 8353, 19, 454, 6, 2480, 99407, 113, 13786, 9, 132076, 425, 23150, 334, 11728, 47, 5, 304, 3355, ...
[ 0.052886962890625, 0.2200927734375, 0.246826171875, 0.186767578125, 0.2247314453125, 0.038116455078125, 0.210693359375, 0.09326171875, 0.1683349609375, 0.0183563232421875, 0.2171630859375, 0.1629638671875, 0.1094970703125, 0.04058837890625, 0.1473388671875, 0.1939697265625, 0.1767578...
c5b3c2ff3f620cd1bc579f611122fcda4e5d5d03
subsection
58
63
Quasi-Modes for Logistic Regression
Solve for q'(i)=0 , then the approximate posterior mode (quasi-mode) of i can be obtained as {\mu _i}^*=\log \left[ \frac{\sum _{j=1}^{n_i}e^{-{x_\crcr \vbox to.2ex{\hbox{$x\tilde{}$}\vss }}{ij}^{\prime }{\beta _\crcr \vbox to.2ex{\hbox{$\beta \tilde{}$}\vss }}{(0)}^*}}{n_i(1-\bar{y}_i)} \right] . Notice that the term ...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 6678, 272, 100, 8096, 25, 14, 145407, 7068, 35707, 5134, 15829, 39225, 13736, 13722, 172, 9, 61170, 111, 17, 831, 113054, 237, 561, 101, 1639, 4867, 2480, 132076, 11832, 170, 33000, 19, 425, 23150, 334, 11728, 47, 304, 3355, 3675, 112, ...
[ 0.176025390625, 0.1922607421875, 0.1103515625, 0.1798095703125, 0.1771240234375, 0.1756591796875, 0.243408203125, 0.006195068359375, 0.12353515625, 0.166748046875, 0.1185302734375, 0.310791015625, 0.24267578125, 0.1634521484375, 0.212646484375, 0.05389404296875, 0.230712890625, 0.0...
a9e5c9945b7b2ea46ed99e75bfd71de22a2ca89c
subsection
59
63
Empirical Logistic Transform (ELT)
We consider the empirical logistic transform (ELT) without covariates for binary data. See Cox and Snell (1972) for the empirical logistic transform (ELT) that accommodates binary data. Letting y denote a binomial random variable with success probability p, the empirical logistic transform, Z, isZ=\mbox{log} (\frac{Y+\...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 16916, 156002, 6827, 150535, 27198, 162408, 15490, 552, 124640, 1636, 2394, 6635, 2053, 1311, 425, 136, 69527, 123599, 16, 70, 211196, 113, 8, 48345, 10, 333, 1687, 15403, 96759, 77336, 678, 36272, 37242, 2481, 915, 567, 4, 83, 1511, 1369...
[ 0.083251953125, 0.262451171875, 0.0684814453125, 0.291015625, 0.2939453125, 0.28515625, 0.15234375, 0.099853515625, 0.1912841796875, 0.04718017578125, 0.2333984375, 0.1793212890625, 0.174560546875, 0.0687255859375, 0.12548828125, 0.0638427734375, 0.1746826171875, 0.0950927734375, ...
4079d02261f7972932ea3f50ad96f0a7a51d2f07
subsection
60
63
Empirical Logistic Transform (ELT)
Faes, C., Ormerod, J. T. and Wand, M. P. (2011). Variational Bayesian Inference for Parametric and Nonparametric Regression With Missing Data. Journal of the American Statistical Association, 106, 959-971. Fay, R.E. and Herriot, R.A. (1979). Estimates of Income for Small Places: an Application of James-Stein Procedures...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 3036, 90, 4, 3347, 282, 5914, 821, 5, 384, 136, 110798, 81757, 111477, 43315, 9631, 3378, 360, 69988, 100, 1720, 186518, 3775, 6276, 853, 148448, 17106, 16771, 214, 11809, 34419, 111, 70, 15672, 126124, 21533, 56187, 38264, 483, 15770, 11...
[ 0.1221923828125, 0.1318359375, 0.052978515625, 0.059234619140625, 0.06842041015625, 0.16552734375, 0.0357666015625, 0.049041748046875, 0.0072021484375, 0.0927734375, 0.161865234375, 0.195556640625, 0.244873046875, 0.07098388671875, 0.197021484375, 0.1602783203125, 0.09918212890625, ...
32306fc6ec28f9ab60ff47e862ece8ac2845de2b
subsection
61
63
Empirical Logistic Transform (ELT)
Journal of the American Statistical Association, 105, 120-135. Nandram, B. and Erhardt, E. (2005). Fitting Bayesian Two-Stage Generalized Linear Models Using Random Samples via the SIR Algorithm. Sankhya, 66, 733-755. Ormerod, J. T., and Wand, M. P. (2010). Explaining Variational Approximations. The American Statistici...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 34419, 126124, 21533, 56187, 21213, 5390, 88622, 5, 353, 14519, 39, 136, 1004, 25612, 18, 241, 117593, 21736, 9631, 90, 3378, 32964, 9, 294, 16852, 9082, 29367, 16777, 147, 13293, 7, 345, 39643, 306, 3362, 99726, 1829, 70, 159, 14688, 1...
[ 0.128662109375, 0.158447265625, 0.037109375, 0.07470703125, 0.1029052734375, 0.0675048828125, 0.059906005859375, 0.037139892578125, 0.0274810791015625, 0.1407470703125, 0.128662109375, 0.051544189453125, 0.01031494140625, 0.1556396484375, 0.073486328125, 0.036712646484375, 0.13818359...
ac1a3b91cf32635e65434fa6eac0376eabdaa57b
subsection
62
63
Empirical Logistic Transform (ELT)
Journal of Multivariate Analysis,127, 36-55. [table]skip=0pt [Table: Categorical tables for 60,221 households by posterior coefficient of variation of model (2) with random effects at theward level projected to the households and the model (1) at household level][Figure: Comparison of the posterior means (PM) of the ho...
{ "cite_spans": [] }
1806.00446
Bayesian Logistic Regression for Small Areas with Numerous Households
[ "Balgobin Nandram", "Lu Chen", "Shuting Fu", "Binod Manandhar" ]
[ "stat.ME" ]
2,018
en
Statistics
[ 34419, 19335, 124640, 67, 114837, 164, 133249, 4039, 112930, 22819, 44024, 145407, 6328, 618, 2886, 6, 161637, 6827, 23180, 7, 100, 1496, 4, 304, 3117, 197540, 390, 39225, 552, 13, 24500, 45964, 111, 143834, 3299, 1737, 96759, 93425, 99, ...
[ 0.1002197265625, 0.1512451171875, 0.1778564453125, 0.026458740234375, 0.1258544921875, 0.0310821533203125, 0.13232421875, 0.0631103515625, 0.108154296875, 0.1466064453125, 0.1192626953125, 0.081787109375, 0.0631103515625, 0.1231689453125, 0.1708984375, 0.0169677734375, 0.168701171875...
cac981698b6ee19c3c6b5ab039a31183274fd4a6
abstract
0
14
Abstract
Self-replication is a key aspect of biological life that has been largely overlooked in Artificial Intelligence systems. Here we describe how to build and train self-replicating neural networks. The network replicates itself by learning to output its own weights. The network is designed using a loss function that can b...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 55347, 9, 107, 182867, 83, 10, 22799, 43585, 333, 109622, 6897, 450, 1556, 2809, 21334, 538, 645, 64747, 297, 231996, 157075, 76519, 5, 11853, 642, 98363, 3642, 47, 45367, 136, 25550, 15970, 52721, 1916, 108, 82451, 33120, 143126, 1636, 6...
[ 0.2509765625, 0.1107177734375, 0.2188720703125, 0.310546875, 0.07415771484375, 0.0231170654296875, 0.142822265625, 0.1279296875, 0.11376953125, 0.08404541015625, 0.16259765625, 0.03765869140625, 0.045135498046875, 0.053253173828125, 0.0894775390625, 0.004974365234375, 0.101318359375,...
565f3d9b39fd3667161d0154b2357481c6dc0a55
subsection
1
14
Introduction
The concept of an artificial self-replicating machine was first proposed by John von Neumann in the 1940s prior to the discovery of DNA's role as the physical mechanism for biological replication. Specifically, Von Neumann demonstrated a configuration of initial states and transformation rules for a cellular automaton ...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 23755, 111, 142, 66733, 15970, 9, 107, 52721, 1916, 36279, 509, 5117, 26171, 71, 390, 4939, 542, 11059, 5761, 23, 70, 27712, 7, 41928, 47, 103882, 53, 27583, 31486, 72761, 191619, 100, 333, 109622, 456, 182867, 25958, 4, 21209, 106804, ...
[ 0.20947265625, 0.037994384765625, 0.051361083984375, 0.2220458984375, 0.2396240234375, 0.06439208984375, 0.2325439453125, 0.318359375, 0.1722412109375, 0.2298583984375, 0.05322265625, 0.1278076171875, 0.18310546875, 0.03802490234375, 0.05096435546875, 0.11279296875, 0.1805419921875, ...
869b9c405fc16a403ebed09a6fab5d4d97cbd9b0
subsection
2
14
Introduction
Analogously, we can construct a self-improving mechanism for artificial intelligence via natural selection if AI agents had the ability to replicate and improve themselves without additional machinery.\item Neural networks are capable of learning powerful representations across many different domains of data \cite{repr...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 5710, 4867, 79850, 4, 642, 831, 64549, 15970, 128659, 191619, 66733, 6, 130687, 1829, 6083, 132216, 2174, 38730, 97957, 1902, 70, 81273, 47, 143126, 67, 136, 52295, 61261, 15490, 78301, 36279, 1294, 5, 41872, 217, 195, 799, 82451, 33120, ...
[ 0.0872802734375, 0.0780029296875, 0.044647216796875, 0.045166015625, 0.05303955078125, 0.1361083984375, 0.1669921875, 0.209716796875, 0.2166748046875, 0.217041015625, 0.1571044921875, 0.045074462890625, 0.1702880859375, 0.07464599609375, 0.156005859375, 0.21337890625, 0.0657958984375...
1c7328425497fd7c5e83e4b4dc729590a4755a0d
subsection
3
14
Introduction
To circumvent this, we need an indirect way of referring to $\Theta$.\subsubsection{Indirect Reference}HyperNEAT \cite{hyperneat} is a neuro-evolution method that describes a neural network by identifying every topological connection with a coordinate and a weight. We pursue the same strategy in building a quine. Inste...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 75723, 16274, 3871, 116361, 3917, 111, 15005, 2852, 41872, 3957, 102, 5, 22144, 7, 58994, 4153, 80581, 215996, 60977, 1264, 8015, 14609, 6, 47959, 3038, 86, 257, 83, 10, 37817, 13, 137089, 55300, 98363, 108, 82451, 33120, 390, 135812, 214...
[ 0.130859375, 0.1517333984375, 0.042938232421875, 0.2340087890625, 0.050384521484375, 0.0255584716796875, 0.131591796875, 0.0850830078125, 0.025421142578125, 0.1630859375, 0.208251953125, 0.0252685546875, 0.158447265625, 0.0987548828125, 0.1251220703125, 0.117919921875, 0.243896484375...
6b83f948159c7bdc4cee638275add2b77327da49
subsection
4
14
Introduction
We demonstrate a visualization of this in Figure \ref{fig:no-one-hot}: contiguous weights might be very different, but contiguous outputs cannot be very different.\begin{figure}[h]\begin{center}\includegraphics[width=0.25\textwidth]{no-one-hot.png}\end{center}\caption{Log-normalized illustration of the weights and weig...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 106804, 21176, 47691, 55412, 6, 29087, 83671, 157, 9, 3630, 21620, 8152, 22832, 12282, 10821, 57888, 7, 13648, 186, 4552, 12921, 4, 1284, 140992, 53418, 372, 6820, 24854, 59994, 13, 127, 268, 41872, 30090, 217028, 48461, 1065, 146984, 5, ...
[ 0.0748291015625, 0.142578125, 0.0362548828125, 0.1739501953125, 0.030426025390625, 0.1048583984375, 0.14990234375, 0.07427978515625, 0.030303955078125, 0.12158203125, 0.18994140625, 0.0305633544921875, 0.120361328125, 0.1864013671875, 0.030242919921875, 0.2493896484375, 0.05718994140...
53443a40b76f1f3cf4fbc66952150e38b0360d12
subsection
5
14
Introduction
Our primary aim in this paper is to demonstrate a proof of concept for a neural network quine, which makes MNIST a suitable auxiliary task as it is considered an easy problem for modern machine learning algorithms.\section{Training the Network}\subsection{Network Architecture}Before describing how the neural network qu...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 22929, 158978, 10, 464, 903, 15122, 106804, 6, 98869, 111, 23755, 108, 82451, 33120, 569, 86, 30482, 46720, 35602, 202319, 119591, 53, 66211, 442, 83, 90698, 142, 23468, 2967, 100, 5744, 36279, 52080, 234873, 7, 28542, 14653, 28997, 8152, ...
[ 0.01153564453125, 0.1396484375, 0.027801513671875, 0.1798095703125, 0.0184478759765625, 0.051849365234375, 0.07012939453125, 0.02783203125, 0.148681640625, 0.02789306640625, 0.1756591796875, 0.09027099609375, 0.2161865234375, 0.222412109375, 0.262939453125, 0.26611328125, 0.031585693...
0398507541fd981d1865564948885a948b2df313
subsection
6
14
Introduction
In our case, the loss function is a moving target, since $\Theta_c$ changes after each gradient update. Updating the loss function after every mini-batch update is expensive. To avoid that, we split the set of possible coordinates into random mini-batches of size 10, and update the loss function after every training ep...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 2446, 7225, 86669, 32354, 83, 10, 98567, 30388, 4, 3957, 102, 454, 238, 65572, 7103, 12638, 35358, 2517, 28350, 10091, 56464, 11907, 7665, 2848, 206, 135587, 71864, 65450, 70, 5423, 111, 7722, 176866, 90, 3934, 96759, 17007, 13267, 209, 2...
[ 0.0205841064453125, 0.032196044921875, 0.2442626953125, 0.21142578125, 0.068115234375, 0.0247955322265625, 0.150146484375, 0.2288818359375, 0.0190887451171875, 0.0941162109375, 0.115478515625, 0.01910400390625, 0.0728759765625, 0.15625, 0.0216217041015625, 0.09521484375, 0.158203125,...
7040c37f52150f7dc579152277765942a7850fa3
subsection
7
14
Introduction
We note that regeneration is sensitive to choices of weight initialization and activation function.\begin{program}\mbox{Pseudo-code for Regeneration:}\BEGIN \\ |Initialize set of parameters|\ \Theta_C|Initialize number of generation epochs|\ G|Initialize number of optimization epochs|\ T\FOR g:=0 \TO G \DO//\ Optimizat...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 1401, 20537, 94410, 1363, 83, 191125, 47, 44126, 7, 57888, 61475, 47691, 136, 34704, 32354, 372, 6820, 18749, 8152, 41872, 39, 11728, 683, 184, 11547, 9, 40899, 68744, 56, 2320, 12, 20090, 108526, 18991, 6, 4153, 40705, 119066, 5423, 111,...
[ 0.00390625, 0.0733642578125, 0.283447265625, 0.1307373046875, 0.056640625, 0.2154541015625, 0.031524658203125, 0.16552734375, 0.03167724609375, 0.2088623046875, 0.1566162109375, 0.03826904296875, 0.030853271484375, 0.1375732421875, 0.114013671875, 0.031524658203125, 0.03692626953125,...
b15df6a0bf0d032c24b79172f259ce15621555c6
subsection
8
14
Introduction
For example, \cite{lipson} estimate the self-replicating quotient of Penrose Tiling \cite{penrose} to be below $0.69$ and that of animals to be at least $46.05$. This framework is useful for distinguishing between trivial and non-trivial replicators. To compute this metric for our network, we need to compute the chance...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 1326, 27781, 4, 47959, 95680, 191, 25902, 70, 15970, 107, 52721, 1916, 41502, 18750, 4267, 75287, 5064, 214, 2278, 47, 186, 35064, 122887, 1126, 4369, 136, 111, 85825, 99, 19713, 9271, 5, 8194, 170846, 83, 80234, 157167, 54700, 17721, 192...
[ 0.01708984375, 0.048858642578125, 0.0182952880859375, 0.042022705078125, 0.087890625, 0.1143798828125, 0.15283203125, 0.0182952880859375, 0.197265625, 0.1561279296875, 0.28173828125, 0.07177734375, 0.199951171875, 0.1376953125, 0.114013671875, 0.2257080078125, 0.1990966796875, 0.15...
50be5a6e832cb86731a0a2fe7bf7a7a5248e8797
subsection
9
14
Introduction
This corresponds to a self-replicating quotient of $10.06$ and an average weight prediction margin of $0.0065$, which is an order of magnitude better than the best solution found previously.\begin{figure}[h]\begin{center}\includegraphics[width=0.5\textwidth]{vquine_gen.png}\end{center}\caption{Training a vanilla quine ...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 3293, 42518, 7, 47, 10, 15970, 107, 52721, 1916, 6, 41502, 18750, 111, 168404, 5, 9016, 142, 83080, 57888, 92054, 1830, 92142, 37509, 12424, 4, 3129, 83, 12989, 101668, 13, 11522, 3501, 70, 2965, 29806, 14037, 198395, 372, 6820, 24854, ...
[ 0.0270843505859375, 0.1226806640625, 0.0171051025390625, 0.0171051025390625, 0.02642822265625, 0.2142333984375, 0.1798095703125, 0.282958984375, 0.1217041015625, 0.0170440673828125, 0.1795654296875, 0.148681640625, 0.0223236083984375, 0.166015625, 0.05718994140625, 0.2113037109375, 0...
7fca1a7383e7d40d6119e8cf72b841af7a3f4d3f
subsection
10
14
Introduction
(It is not immediately obvious from Figure \ref{fig:mquine_loss}, but the first few training epochs reduce the self-replicating loss too.)There are parallels to be drawn between self-replication in the case of a neural network quine and biological reproduction in nature, as well as specialization at the auxiliary task ...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 83, 959, 109312, 162520, 1295, 55412, 13, 6, 29087, 24854, 83671, 3181, 86, 2400, 7, 8152, 4, 1284, 70, 5117, 10846, 23189, 771, 206, 34390, 15970, 107, 52721, 1916, 86669, 5792, 95103, 186, 79442, 19, 182867, 23, 7225, 108, 82451, 3312...
[ 0.03045654296875, 0.05517578125, 0.03790283203125, 0.09417724609375, 0.020965576171875, 0.1358642578125, 0.0305938720703125, 0.03057861328125, 0.0296630859375, 0.03070068359375, 0.1475830078125, 0.1837158203125, 0.199951171875, 0.1900634765625, 0.030670166015625, 0.030670166015625, 0...
4ddbbccf269e5a76fee0f971007a9363de61dafd
subsection
11
14
Introduction
8:1798--1828.\bibitem{breivik2001self}Breivik, J. (2001).\newblock Self-organization of template-replicating polymers and thespontaneous rise of genetic information.\newblock {\em Entropy}, 3(4):273--279.\bibitem{wiki:neumann}contributors, W. (2017).\newblock Self-replication --- wikipedia{,} the free encyclopedia.\new...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 382, 22950, 16665, 9, 9444, 3882, 5, 41872, 127872, 2982, 4134, 14, 5342, 38824, 39379, 8152, 9569, 4, 821, 133206, 54936, 81409, 55347, 53404, 2320, 111, 110934, 107, 52721, 1916, 35874, 31648, 136, 70, 18908, 60626, 10821, 58944, 101412, ...
[ 0.2418212890625, 0.184814453125, 0.32958984375, 0.082763671875, 0.1982421875, 0.290283203125, 0.0513916015625, 0.03851318359375, 0.183349609375, 0.265869140625, 0.08349609375, 0.05816650390625, 0.254150390625, 0.19189453125, 0.1749267578125, 0.038360595703125, 0.1258544921875, 0.03...
5cf5b8ece0f07b151d70c8c05990c3bf4b651e9b
subsection
12
14
Introduction
(1980).\newblock {\em G{\"o}del, Escher, Bach: an Eternal Golden Braid}.\newblock New York: Vintage Books.\bibitem{hornik1991approximation}Hornik, K. (1991).\newblock Approximation capabilities of multilayer feedforward networks.\newblock {\em Neural networks}, 4(2):251--257.\bibitem{huang2006extreme}Huang, G.-B., Zhu,...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 223265, 5, 41872, 54936, 81409, 195, 527, 24854, 58, 31, 8152, 1743, 1184, 5372, 35631, 12, 142, 1775, 56, 6236, 43114, 6163, 532, 2356, 5753, 128329, 83266, 127872, 2982, 5922, 1881, 76550, 2631, 64101, 464, 2320, 91658, 4, 209721, 5659,...
[ 0.22509765625, 0.0804443359375, 0.03985595703125, 0.04302978515625, 0.1717529296875, 0.1090087890625, 0.0693359375, 0.0389404296875, 0.037994384765625, 0.0369873046875, 0.04083251953125, 0.206298828125, 0.039794921875, 0.188232421875, 0.195556640625, 0.0631103515625, 0.060302734375, ...
d2f2e8a16a68278dae8881438e673f851b0906da
subsection
13
14
Introduction
Springer.\bibitem{shen2017natural}Shen, J., Pang, R., Weiss, R.~J., Schuster, M., Jaitly, N., Yang, Z., Chen, Z.,Zhang, Y., Wang, Y., Skerry-Ryan, R., et~al. (2017).\newblock Natural tts synthesis by conditioning wavenet on mel spectrogrampredictions.\newblock {\em arXiv preprint arXiv:1712.05884}.\bibitem{hyperneat}St...
{ "cite_spans": [] }
1803.05859
Neural Network Quine
[ "Oscar Chang", "Hod Lipson" ]
[ "cs.AI", "cs.NE" ]
2,018
en
Computer Science
[ 38026, 56, 5, 41872, 127872, 2982, 65726, 3259, 84832, 8152, 294, 3220, 4, 821, 43531, 627, 215228, 2306, 1375, 42763, 1515, 276, 823, 217, 538, 4954, 567, 123378, 1511, 8467, 990, 59246, 56901, 56216, 9, 1643, 82, 289, 52977, 54936, 81...
[ 0.18701171875, 0.1927490234375, 0.0521240234375, 0.048004150390625, 0.1502685546875, 0.2183837890625, 0.1773681640625, 0.220703125, 0.214111328125, 0.06463623046875, 0.04644775390625, 0.1923828125, 0.046844482421875, 0.1163330078125, 0.2176513671875, 0.045440673828125, 0.1748046875, ...
6e5b3a24bed4d9648510f0d678968f6ae19a5303
abstract
0
34
Abstract
In this paper we use the framework of algebraic effects from programming language theory to analyze the Beta-Bernoulli process, a standard building block in Bayesian models. Our analysis reveals the importance of abstract data types, and two types of program equations, called commutativity and discardability. We develo...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 15122, 4527, 170846, 144, 429, 2844, 93425, 1295, 56037, 449, 46876, 154453, 7968, 53, 48941, 9, 33189, 3165, 7532, 9433, 5570, 33976, 46389, 9631, 90, 3378, 115774, 114137, 122273, 131011, 164789, 2053, 52895, 6626, 1528, 13722, 375, 68754, ...
[ 0.04071044921875, 0.0560302734375, 0.13671875, 0.012603759765625, 0.0892333984375, 0.08349609375, 0.1904296875, 0.003875732421875, 0.1861572265625, 0.079345703125, 0.1759033203125, 0.1685791015625, 0.0938720703125, 0.0143280029296875, 0.24560546875, 0.046600341796875, 0.1510009765625...
f10bf79d6c6f0f9852319a04acd84b479137d9f7
subsection
1
34
Introduction
From the perspective of programming, a family of Boolean random processes is implemented by a module that supports the following interface: module type ProcessFactory = sig type process val new : H -> process val get : process -> bool end where H is some type of hyperparameters. Thus one can initialize a new process,...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 800, "openalex_id": "", "raw": "Gordon Plotkin and John Power. Algebraic operations and generic effects. Appl. Categ. Structures, 11(1):69–94, 2003.", "source_ref_id": "8f406e2797ccea47f827d9d370634f7e953a8bf3", "start...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 70, 80280, 56037, 449, 14449, 73783, 34677, 96759, 9433, 90, 29479, 297, 88996, 8060, 7, 25632, 101758, 10644, 123996, 135518, 31667, 1144, 3502, 3525, 572, 2046, 6, 142484, 3564, 83, 111, 59058, 6276, 29089, 831, 61475, 20650, 10, 4, 136...
[ 0.00994873046875, 0.06988525390625, 0.1748046875, 0.05487060546875, 0.1063232421875, 0.1275634765625, 0.208740234375, 0.1900634765625, 0.2039794921875, 0.09442138671875, 0.1768798828125, 0.01055908203125, 0.2646484375, 0.1368408203125, 0.01019287109375, 0.028350830078125, 0.182006835...
b2b30324919ff1c4677452d9b15a11ade2419458
subsection
2
34
Introduction
Discardability is the requirement that when |x| is not free in |u|, \Big (\text{|let x = t in u|}\Big ) = \Big (\text{|u|}\Big ).Together, these properties say that data flow, rather than the control flow, is what matters. For example, in a standard programming language, the purely functional total expressions are comm...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1214, "openalex_id": "", "raw": "Nathanael L. Ackerman, Cameron E. Freer, and Daniel M. Roy. Exchangeable random primitives. Workshop on Probabilistic Programming Semantics (PPS 2016), 2016. URL: http://pps2016.soic.indiana.edu/fi...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 6206, 38931, 41159, 83, 64209, 674, 450, 3229, 6, 425, 959, 4092, 23, 34, 58745, 4, 129933, 15, 41872, 22829, 24854, 1974, 1022, 2203, 808, 75, 8152, 1388, 7763, 9319, 183871, 2053, 86608, 43257, 3501, 6226, 2367, 26866, 7, 1326, 27781,...
[ 0.183837890625, 0.3203125, 0.28369140625, 0.0849609375, 0.205078125, 0.042572021484375, 0.01171875, 0.0819091796875, 0.011627197265625, 0.10089111328125, 0.1435546875, 0.1744384765625, 0.0682373046875, 0.1153564453125, 0.014129638671875, 0.01177978515625, 0.0992431640625, 0.0119018...
a6b228a4f368973d18e7aa06ae46ee28d87c0b67
subsection
3
34
Introduction
REF ).We argue that these results open up a new method for analyzing Bayesian models, based on algebraic effects (see § and This paper formalizes and proves a conjecture from , which is an unpublished abstract.).
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 212, "openalex_id": "", "raw": "Sam Staton, Hongseok Yang, Nathanael Ackerman, Cameron Freer, and Daniel M. Roy. Exchangeable random processes and data abstraction. Workshop on Probabilistic Programming Semantics (PPS 2017), 2017....
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 9069, 919, 12137, 187, 6261, 6097, 50339, 9803, 1257, 3525, 55300, 7968, 53, 35401, 9631, 90, 3378, 115774, 35509, 98, 144, 429, 2844, 1771, 93425, 21231, 5360, 3293, 15122, 1483, 119066, 23534, 158, 75161, 3408, 1295, 6, 51, 179193, 1647...
[ 0.037322998046875, 0.08544921875, 0.040435791015625, 0.1163330078125, 0.0750732421875, 0.07781982421875, 0.1500244140625, 0.09320068359375, 0.042205810546875, 0.0916748046875, 0.15673828125, 0.0921630859375, 0.0782470703125, 0.046600341796875, 0.1649169921875, 0.090087890625, 0.05212...
8900f27158c2c1f7af607465ea6043e325dcd619
subsection
4
34
An algebraic presentation of the Beta-Bernoulli process
In this section, we present syntactic rules for well-formed client programs of the Beta-Bernoulli module, and axioms for deriving equations on those programs.
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 360, 903, 40059, 642, 13379, 6002, 102, 49086, 91736, 100, 5299, 5037, 297, 23282, 103391, 48941, 9, 33189, 3165, 7532, 88996, 136, 10, 5134, 28016, 16406, 6496, 28, 13722, 5256, 98, 8382 ]
[ 0.035888671875, 0.1080322265625, 0.1353759765625, 0.012115478515625, 0.1181640625, 0.1400146484375, 0.19140625, 0.1624755859375, 0.188720703125, 0.042755126953125, 0.0809326171875, 0.1365966796875, 0.006072998046875, 0.2384033203125, 0.1827392578125, 0.2191162109375, 0.02372741699218...
51dc8d781fa3805c75dcd6541c3326b142d6bc50
subsection
5
34
An algebraic presentation of finite probability
Recall the module |Bernoulli| from the introduction which provides a method of sampling with odds (i:j). We will axiomatize its equational properties. Algebraic effects provide a way to axiomatize the specific features of this module while putting aside the general properties of programming languages, such as \beta /\e...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1016/j.entcs.2013.09.007", "end": 1072, "openalex_id": "https://openalex.org/W1983128619", "raw": "Danel Ahman and Sam Staton. Normalization by evaluation and algebraic effects. In Proc. MFPS 2013, volume 298 of Electron. Notes Theor. Co...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 853, 85763, 88996, 33189, 3165, 7532, 1295, 39864, 87344, 55300, 57, 2676, 2069, 678, 103044, 170, 1401, 1221, 10, 78697, 16476, 731, 13722, 43315, 183871, 884, 429, 2844, 1771, 93425, 22691, 3917, 70, 29458, 66139, 903, 8752, 4537, 56037, ...
[ 0.01678466796875, 0.137451171875, 0.239990234375, 0.1280517578125, 0.2242431640625, 0.2509765625, 0.01080322265625, 0.0908203125, 0.0633544921875, 0.12890625, 0.08740234375, 0.1829833984375, 0.0994873046875, 0.0909423828125, 0.238037109375, 0.1763916015625, 0.000640869140625, 0.021...
59a1249cafb046bcbbd94094e5f7454ce059c5bc
subsection
6
34
An algebraic presentation of finite probability
The theory of rational convexity is the first-order algebraic theory with binary operations \mathop {{}_{i}\!{?}\!_{j}} for all i,j\in \mathbb {N} such that i+j>0, subject to the axiom schemesw,x,y,z\vdash &(w \mathop {{}_{i}\!{?}\!_{j}} x) \mathop {{}_{i+j}\!{?}\!_{k+l}}(y \mathop {{}_{k}\!{?}\!_{l}} z) = (w \mathop {...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1172, "openalex_id": "", "raw": "F. E. J. Linton. Autonomous equational categories. J. Math. Mech., 15:637–642, 1966.", "source_ref_id": "202f8be46407e0d36d321f6be59e0495e86cdebb", "start": 1001 }, { "arx...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 581, 154453, 111, 168487, 158, 272, 425, 2481, 83, 70, 5117, 80596, 144, 429, 2844, 1771, 678, 2394, 6635, 41018, 7, 41872, 2227, 58052, 8152, 454, 24854, 14, 38, 32, 170, 47391, 756, 17, 73, 6, 5125, 839, 6044, 1328, 2740, 2389, 28...
[ 0.086669921875, 0.302001953125, 0.1334228515625, 0.33447265625, 0.2034912109375, 0.3076171875, 0.212158203125, 0.2109375, 0.155029296875, 0.0523681640625, 0.124267578125, 0.181884765625, 0.048675537109375, 0.1162109375, 0.0948486328125, 0.0313720703125, 0.109130859375, 0.2114257812...
2b2ff3de467df12e93c5fce5a89335a326d170ab
subsection
7
34
A parameterized algebraic signature for Beta-Bernoulli
In the theory of convex sets, the parameters i,j for |get| range over the integers. These integers are not a first class concept in our equational presentation: we did not axiomatize integer arithmetic. However, in the Beta-Bernoulli process, or any module |M| for the |ProcessFactory| interface, it is helpful to unders...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1219, "openalex_id": "", "raw": "Gordon Plotkin and John Power. Algebraic operations and generic effects. Appl. Categ. Structures, 11(1):69–94, 2003.", "source_ref_id": "8f406e2797ccea47f827d9d370634f7e953a8bf3", "star...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 360, 154453, 111, 158, 272, 425, 5423, 7, 171859, 17, 4, 170, 100, 6, 3794, 58745, 37457, 645, 892, 26255, 32255, 621, 959, 5117, 18507, 23755, 2446, 13722, 43315, 92252, 78697, 1505, 8962, 48941, 9, 33189, 3165, 7532, 9433, 88996, 594,...
[ 0.004241943359375, 0.20361328125, 0.0193328857421875, 0.1224365234375, 0.2161865234375, 0.1376953125, 0.22265625, 0.043731689453125, 0.258544921875, 0.12939453125, 0.054351806640625, 0.1966552734375, 0.06817626953125, 0.002960205078125, 0.1929931640625, 0.013031005859375, 0.181640625...
05fc0b84a23ba6cf15a7c71318eea81b01449546
subsection
8
34
Axioms for Beta-Bernoulli
The axioms for the Beta-Bernoulli theory comprise the axioms for rational convexity (Def. REF ) together with the following axiom schemes.Commutativity. All the operations commute with each other: p,q\mathop |w,x,y,z:0\vdash \;&(w\mathop {?_{q}} x)\mathop {?_{p}}(y\mathop {?_{q}} z)=(w\mathop {?_{p}} y)\mathop {?_{q}}...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 10, 5134, 28016, 100, 70, 48941, 9, 33189, 3165, 7532, 154453, 48402, 13, 168487, 158, 272, 425, 2481, 187423, 9069, 919, 1388, 25842, 678, 25632, 306, 150370, 7, 10770, 68754, 54613, 53, 3164, 41018, 375, 561, 67, 12638, 3789, 915, 864...
[ 0.0665283203125, 0.1923828125, 0.1871337890625, 0.07073974609375, 0.01629638671875, 0.27392578125, 0.0643310546875, 0.1337890625, 0.224365234375, 0.2376708984375, 0.2392578125, 0.11572265625, 0.03997802734375, 0.2327880859375, 0.1182861328125, 0.2340087890625, 0.1302490234375, 0.11...
ccf6a750845be90bbe469c1af2f00a3ec05b8ee3
subsection
9
34
Axioms for Beta-Bernoulli
It immediately follows from conjugacy and discardability that x \mathop {{}_{i}\!{?}\!_{j}} y is definable as \nu _{i,j}p.(x\mathop {?_{p}} y) for i,j>0.As an example, consider t(r) = (r \mathop {?_{p}} x) \mathop {?_{p}} (y \mathop {?_{p}} r) that represents tossing a coin with bias p twice, continuing with x or y if ...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 626, "openalex_id": "", "raw": "John von Neumann. Various techniques used in connection with random digits. Nat. Bur. Stand. Appl. Math. Series, 12:36–38, 1951.", "source_ref_id": "c3ed2edddf93d0abfe276563458006b55cd5387d", ...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 109312, 28960, 1295, 228186, 2408, 136, 17116, 5861, 41159, 1022, 2227, 58052, 14, 38, 170, 113, 83, 45154, 2886, 237, 539, 254, 425, 32, 100, 17, 2740, 2389, 27781, 16916, 808, 132, 42, 53, 1690, 33636, 47, 38001, 43780, 678, 333, 16...
[ 0.1314697265625, 0.109375, 0.00439453125, 0.262939453125, 0.099365234375, 0.03265380859375, 0.1009521484375, 0.1793212890625, 0.134033203125, 0.1470947265625, 0.1568603515625, 0.251708984375, 0.08563232421875, 0.06390380859375, 0.168701171875, 0.2015380859375, 0.08056640625, 0.2427...
a2454a94ffa01d6b45a504fe0b8500ac796f96c4
subsection
10
34
A complete interpretation in measure theory
In this section we give an interpretation of terms using measures and integration operators, the standard formalism for probability theory (e.g. , ), and we show that this interpretation is complete (Thm. REF ). Even if the reader is not interested in measure theory, they may still find value in the syntactical results...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 204, "openalex_id": "", "raw": "David Pollard. A user's guide to measure theoretic probability. Cambridge University Press, 2001.", "source_ref_id": "aac73d52a5546ec1460493ff434401ddd05e3e31", "start": 0 }, { ...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 903, 40059, 8337, 206019, 69407, 17368, 72350, 136, 157353, 39933, 5570, 23113, 8780, 37242, 2481, 154453, 7639, 28484, 9069, 919, 155255, 959, 60892, 1543, 7464, 7413, 34292, 6002, 102, 49086, 50339, 5360, 23534, 7432, 16750, 41419, 1330, 73...
[ 0.0621337890625, 0.123779296875, 0.051910400390625, 0.2296142578125, 0.19873046875, 0.0816650390625, 0.2386474609375, 0.0260009765625, 0.216552734375, 0.215576171875, 0.1856689453125, 0.1934814453125, 0.1038818359375, 0.24365234375, 0.1036376953125, 0.1888427734375, 0.06646728515625,...
2d658d15231c63414fc2eb9b2cac064d29706eee
subsection
11
34
Programs as probability kernels.
Forgetting about abstract types for a moment, terms in the |BetaBern| module are first-order probabilistic programs. So we have a standard denotational semantics due to where terms are interpreted as probability kernels and \nu  as integration. Let I= [0,1] denote the unit interval. We write \beta _{i,j} for the \textr...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1016/0022-0000(81)90036-2", "end": 245, "openalex_id": "https://openalex.org/W2024355065", "raw": "Dexter Kozen. Semantics of probabilistic programs. J. Comput. System Sci., 22(3):328–350, 1981.", "source_ref_id": "6bbbdec185ba6c08...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 1326, 429, 19514, 164789, 52895, 3095, 69407, 23, 6766, 102, 33189, 19, 58745, 88996, 621, 5117, 80596, 37242, 48242, 103391, 765, 5570, 8, 58358, 43315, 484, 109109, 29481, 2481, 77924, 14794, 41872, 539, 157353, 10842, 87, 1369, 48345, 25...
[ 0.02545166015625, 0.140869140625, 0.06365966796875, 0.2034912109375, 0.1778564453125, 0.11767578125, 0.2413330078125, 0.008575439453125, 0.17626953125, 0.2060546875, 0.25634765625, 0.1737060546875, 0.030059814453125, 0.2298583984375, 0.0599365234375, 0.05572509765625, 0.1380615234375...
1f25043c452691db308270127138ba858d45d287
subsection
12
34
Programs as probability kernels.
We interpret terms \Gamma \mathop |\Delta \vdash t as probability kernels \llbracket t\rrbracket : I^\ell \times \Sigma ( \llbracket \Delta \rrbracket )\rightarrow [0,1] inductively, for \vec{p}\in I^\ell and U\in \Sigma ( \llbracket \Delta \rrbracket ) :& \llbracket x_i(p_{j_1},\ldots ,p_{j_m})\rrbracket (\vec{p}, U) ...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 1401, 29481, 69407, 724, 36557, 2227, 58052, 6, 58598, 102, 334, 155892, 808, 237, 37242, 2481, 77924, 14794, 1181, 2844, 27853, 29161, 87, 6796, 70141, 294, 872, 192, 41872, 1388, 54969, 2389, 76172, 135989, 4935, 538, 4, 100, 35259, 254...
[ 0.06085205078125, 0.23046875, 0.1961669921875, 0.0653076171875, 0.255615234375, 0.136962890625, 0.251220703125, 0.00750732421875, 0.21240234375, 0.1842041015625, 0.1246337890625, 0.254150390625, 0.205322265625, 0.1087646484375, 0.3046875, 0.1640625, 0.2410888671875, 0.18408203125, ...
48b98de7639d72220b6f2425bde449776f75b9d6
subsection
13
34
Interpretation as functionals
We write \mathbb {R}^{I^m} for the vector space of continuous functions I^m \rightarrow \mathbb {R}, endowed with the supremum norm. Given a probability kernel \kappa :I^\ell \times \Sigma \big (\sum _{j=1}^kI^{m_j}\big )\rightarrow [0,1] and \vec{p}\in I^\ell , we define a linear map \phi _{\vec{p}}: \mathbb {R}^{I^{m...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.2168/lmcs-11(2:5)2015", "end": 956, "openalex_id": "https://openalex.org/W2763556082", "raw": "Robert Furber and Bart Jacobs. From Kleisli categories to commutative C*-algebras: Probabilistic Gelfand duality. Log. Methods Comput. Sci., 1...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 1401, 33022, 125458, 5125, 1052, 8152, 568, 8353, 39, 100, 173, 18770, 32628, 62005, 223, 32354, 7, 87, 54969, 118201, 6, 22, 246, 678, 111780, 316, 22599, 77878, 37242, 2481, 77924, 583, 161, 7495, 6796, 70141, 872, 41872, 32976, 15, 1...
[ 0.06719970703125, 0.202880859375, 0.0885009765625, 0.1800537109375, 0.182373046875, 0.00909423828125, 0.1343994140625, 0.157470703125, 0.21044921875, 0.06536865234375, 0.1766357421875, 0.17578125, 0.2015380859375, 0.19287109375, 0.0675048828125, 0.19091796875, 0.007293701171875, 0....
f3c9d358fd7bb33db9939d0ff04c074f79f1360a
subsection
14
34
Interpretation as functionals
It is informative to spell out the interpretation of terms {p_1,\ldots ,p_\ell \mathop |x_1 : m_1, \ldots , x_k : m_k\vdash t} as maps \llbracket t\rrbracket : \mathbb {R}^{I^{m_1}} \times \ldots \times \mathbb {R}^{I^{m_k}} \rightarrow \mathbb {R}^{I^\ell } since it fits the algebraic notation: we may think of the va...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 1650, 83, 149584, 47, 160093, 1810, 206019, 69407, 254, 115187, 30591, 6, 454, 6796, 2227, 58052, 425, 152, 347, 4, 41872, 1022, 92, 334, 155892, 808, 8152, 237, 22288, 7, 1181, 2844, 27853, 29161, 125458, 5125, 10666, 8353, 24854, 568, ...
[ 0.0177459716796875, 0.070556640625, 0.1826171875, 0.00006103515625, 0.1163330078125, 0.0816650390625, 0.30712890625, 0.207763671875, 0.09979248046875, 0.1693115234375, 0.138671875, 0.012908935546875, 0.033355712890625, 0.162353515625, 0.09869384765625, 0.1861572265625, 0.065124511718...
805b806c0ce193fffbca9dc569d5ca4aef2a954d
subsection
15
34
Technical background on Bernstein polynomials
[Bernstein polynomials] For i=0,\ldots ,k, we define the i-th basis Bernstein polynomial b_{i,k} of degree k as b_{i,k}(p) = \binom{k}{i} p^{k-i}(1-p)^i. For a multi-index I = (i_1,\ldots ,i_\ell ) with 0 \le i_j \le k, we let b_{I,k}(\vec{p}) = b_{i_1,k}(p_1)\cdots b_{i_\ell ,k}(p_\ell ). A Bernstein polynomial is a l...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 33189, 19, 18055, 35874, 1687, 15403, 7, 268, 1326, 17, 145407, 41872, 30591, 92, 642, 61924, 9, 927, 18231, 71342, 876, 454, 14, 4, 111, 79385, 472, 237, 8152, 132, 254, 964, 915, 8353, 41600, 6024, 4711, 87, 2203, 115187, 6796, 757,...
[ 0.18017578125, 0.1573486328125, 0.2841796875, 0.266357421875, 0.2423095703125, 0.2203369140625, 0.0845947265625, 0.009124755859375, 0.0615234375, 0.2188720703125, 0.1827392578125, 0.037506103515625, 0.1143798828125, 0.182373046875, 0.03936767578125, 0.2022705078125, 0.0938720703125, ...
b5f5b3590e7047edb27189b5017afadce03c1172
subsection
16
34
Normal forms and completeness
For the completeness proof of the measure-theoretic model, we proceed as follows: To decide \Gamma \mathop |\Delta \vdash t = u for two terms t,u, we transform them into a common normal form whose interpretations can be given explicitly. We then use a series of linear independence results to show that if the interpreta...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 1326, 28484, 7432, 98869, 72350, 9, 2347, 4524, 9523, 3299, 642, 172337, 28960, 717, 28288, 724, 36557, 2227, 58052, 58598, 102, 334, 155892, 808, 2203, 75, 6626, 69407, 34, 27198, 39210, 3638, 3173, 206019, 143726, 192617, 215131, 50339, 7...
[ 0.031005859375, 0.1904296875, 0.1297607421875, 0.2235107421875, 0.267822265625, 0.01025390625, 0.0765380859375, 0.120849609375, 0.020294189453125, 0.2469482421875, 0.026519775390625, 0.1151123046875, 0.0751953125, 0.024810791015625, 0.1759033203125, 0.0283660888671875, 0.23193359375,...
5f90014500a6afaa9c9fb30d4ca0bcb26324d2d0
subsection
17
34
Stone's normal forms for rational convex sets
Normal forms for the theory of rational convex sets have been described by Stone . We note that if -\mathop |x_1\ldots x_k : 0\vdash t is a term in the theory of rational convex sets (Def. REF ) then \llbracket t\rrbracket :\mathbb {R}^k\rightarrow \mathbb {R} is a unital positive linear map that takes rationals to rat...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/bf02413910", "end": 82, "openalex_id": "https://openalex.org/W2094414831", "raw": "M. H. Stone. Postulates for the barycentric calculus. Ann. Mat. Pura Appl. (4), 29:25–30, 1949.", "source_ref_id": "cd31282310b322fe76a4c4d6222...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 28723, 3173, 7, 100, 154453, 168487, 158, 272, 425, 5423, 151552, 47002, 2174, 2227, 58052, 115187, 30591, 454, 92, 757, 334, 155892, 808, 13579, 187423, 9069, 919, 1181, 2844, 27853, 29161, 125458, 8353, 1052, 83, 25072, 289, 24491, 192617...
[ 0.2158203125, 0.2413330078125, 0.0340576171875, 0.0054931640625, 0.1927490234375, 0.22265625, 0.0966796875, 0.2158203125, 0.1226806640625, 0.2176513671875, 0.1104736328125, 0.19287109375, 0.0198516845703125, 0.05804443359375, 0.1778564453125, 0.1182861328125, 0.061767578125, 0.0137...
52f874ec93727cc9a62f8ef9ddf0fc94498951f9
subsection
18
34
Characterization and completeness for
This section concerns the normalization of terms using free parameters but no \nu . Consider a single parameter p. If we think of a term t as a syntactic tree, commutativity and discardability can be used to move all occurrences of \mathop {?_{p}} to the root of the tree, making it a tree diagram of some depth k. Let u...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 3293, 40059, 50509, 3638, 47691, 111, 69407, 17368, 4092, 171859, 7, 1284, 110, 6, 539, 137399, 11001, 915, 4263, 5351, 13579, 808, 237, 6002, 102, 49086, 53201, 375, 68754, 54613, 53, 136, 17116, 5861, 41159, 831, 186, 11814, 25813, 756,...
[ 0.053955078125, 0.11572265625, 0.070556640625, 0.259033203125, 0.220703125, 0.013275146484375, 0.312255859375, 0.1275634765625, 0.2093505859375, 0.28857421875, 0.06524658203125, 0.09527587890625, 0.09808349609375, 0.0067138671875, 0.18408203125, 0.0921630859375, 0.146484375, 0.1330...
5d2376237163184ca59973a849756a7f62102f73
subsection
19
34
Characterization and completeness for
\then \then   For example, normalizing (v\mathop {?_{p}} x){\mathop {?_{p}}}(y\mathop {?_{p}} v) gives (v\mathop {?_{p}}(x \mathop {{}_{1}\!{?}\!_{1}} y)){\mathop {?_{p}}} ((x\mathop {{}_{1}\!{?}\!_{1}} y)\mathop {?_{p}} v)=C_2(v,x{\mathop {{}_{1}\!{?}\!_{1}}}y,v).From this we obtain the following completeness result:P...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1090/s0002-9947-00-02595-2", "end": 1264, "openalex_id": "https://openalex.org/W1535868076", "raw": "Victoria Powers and Bruce Reznick. Polynomials that are positive on an interval. Trans. Amer. Math. Soc., 352(10):4677–4692, 2000.", ...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 92733, 27781, 3638, 84382, 334, 2227, 58052, 254, 1022, 53, 81, 76199, 425, 113, 304, 418, 113054, 25632, 28484, 7432, 16750, 10752, 40322, 8622, 1582, 100034, 42518, 17721, 224743, 3956, 61112, 69407, 915, 115187, 6796, 53861, 155892, 808, ...
[ 0.184814453125, 0.12744140625, 0.2064208984375, 0.185546875, 0.1552734375, 0.09228515625, 0.1641845703125, 0.07635498046875, 0.0955810546875, 0.0775146484375, 0.180908203125, 0.112060546875, 0.0496826171875, 0.13427734375, 0.07647705078125, 0.02276611328125, 0.033203125, 0.06079101...
38b922d477aee0f58cd7665c86c2029267749e10
subsection
20
34
Normalization of Beta-Bernoulli
For arbitrary terms {p_1 \ldots p_\ell \mathop |x_1\colon m_1, \ldots , x_s\colon m_s\vdash t}, we employ the following normalization procedure. Using conjugacy and the commutativity axioms (–), we can push all uses of \nu towards the leaves of the tree, until we end up with a tree of ratios and free parameter choices ...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 1326, 61799, 1294, 69407, 254, 115187, 30591, 915, 454, 6796, 2227, 58052, 425, 22796, 347, 1022, 7, 334, 155892, 808, 642, 187016, 25632, 3638, 47691, 50491, 345, 6953, 228186, 2408, 136, 375, 68754, 54613, 5134, 28016, 1104, 831, 25944, ...
[ 0.0833740234375, 0.1751708984375, 0.04815673828125, 0.26318359375, 0.05572509765625, 0.15283203125, 0.139404296875, 0.05535888671875, 0.0259552001953125, 0.1669921875, 0.056640625, 0.1702880859375, 0.05517578125, 0.0254364013671875, 0.0740966796875, 0.0293731689453125, 0.07275390625,...
c996ee7520186567a9ac6ddbe98220610e5261d7
subsection
21
34
Proof of completeness
Consider a chain c = \nu _{i_1,j_1} p_{\ell +1}.\ldots \nu _{i_d,j_d} p_{\ell + d}.\,x(p_{\tau (1)},\ldots ,p_{\tau (m)}). Its measure-theoretic interpretation \llbracket c\rrbracket (p_1,\ldots ,p_\ell ) is a pushforward of a product of d beta distributions, supported on a hyperplane segment that is parameterized by t...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 137399, 10, 121293, 501, 2203, 6, 539, 14, 115187, 170, 8152, 915, 454, 24854, 6796, 57157, 5, 30591, 71, 4, 41872, 997, 104, 425, 132, 254, 50104, 798, 7344, 15, 39, 16, 194, 1650, 7, 72350, 9, 2347, 4524, 9523, 206019, 1181, 2844,...
[ 0.125244140625, 0.0222625732421875, 0.287841796875, 0.1776123046875, 0.0225677490234375, 0.0223236083984375, 0.177978515625, 0.1441650390625, 0.184326171875, 0.137451171875, 0.022674560546875, 0.0926513671875, 0.0701904296875, 0.0227203369140625, 0.1822509765625, 0.1702880859375, 0.0...
24ead67b749c685a4b07df1da02e780635d67c3d
subsection
22
34
Proof of completeness
Let \sum a_i h_{i*}(\mu _i) = 0 as a signed measure. We show by induction over the dimension of the chains that all a_i vanish. Assume that a_i = 0 whenever the dimension of c_i is less than d, and consider an arbitrary subspace \tau _j of dimension d. We can define a signed Borel measure on I^d by restriction  nullfon...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/978-1-4612-4190-4", "end": 503, "openalex_id": "https://openalex.org/W1539382740", "raw": "Alexander Kechris. Classical Descriptive Set Theory. Springer, 1995.", "source_ref_id": "5dbe8cdfd4e4e4352b801568c885ae932b2504ee", ...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 10842, 6, 11832, 10, 454, 14, 1096, 1639, 8152, 561, 101, 16, 2203, 757, 237, 181141, 72350, 1401, 7639, 390, 23, 77391, 645, 70, 91403, 121293, 7, 450, 756, 131, 4745, 62, 66596, 13, 217684, 501, 83, 40715, 3501, 104, 4, 136, 16916...
[ 0.07354736328125, 0.03076171875, 0.2120361328125, 0.13671875, 0.0989990234375, 0.1436767578125, 0.10089111328125, 0.025421142578125, 0.031341552734375, 0.150390625, 0.033447265625, 0.03131103515625, 0.053436279296875, 0.1240234375, 0.0556640625, 0.254638671875, 0.251708984375, 0.01...
b63b8cdb29310046fa33c498b3f8133a9037d754
subsection
23
34
Proof of completeness
The interpretations of these normal forms are given explicitly by  nullfont\displaystyle \llbracket t\rrbracket (\vec{f})(\vec{p}) = \sum _{j} \frac{w_{Ij}}{w_I} \cdot b_{I,k}(\vec{p}) \cdot \llbracket c_j\rrbracket (\vec{f})(\vec{p}) \text{ where } w_I = \sum _j w_{Ij} \then \then   and analogously for t^{\prime }. T...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 206019, 7, 6097, 3638, 3173, 621, 34475, 143726, 390, 56574, 75635, 1181, 2844, 27853, 808, 29161, 35259, 420, 254, 11832, 170, 132076, 434, 568, 454, 15464, 876, 92, 501, 22829, 148, 92733, 60223, 8353, 114654, 5, 2203, 35388, 756, 757, ...
[ 0.24609375, 0.04241943359375, 0.0936279296875, 0.219970703125, 0.2255859375, 0.000732421875, 0.0870361328125, 0.1312255859375, 0.0081787109375, 0.1746826171875, 0.218994140625, 0.1121826171875, 0.129638671875, 0.1280517578125, 0.14697265625, 0.0772705078125, 0.20654296875, 0.138183...
f90d1d103c1b4f47e26b9719bff9eab8732a55d5
subsection
24
34
Extensionality and syntactical completeness
In this section we use the model completeness of the previous section to establish some syntactical results about the theory of Beta-Bernoulli. Although the model is helpful in informing the proofs, the statements of the results in this section are purely syntactical.The ultimate result of this section is equational sy...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 903, 40059, 4527, 3299, 28484, 7432, 96362, 137633, 6002, 102, 49086, 50339, 154453, 111, 48941, 9, 33189, 3165, 7532, 98893, 8180, 98869, 63805, 20654, 16750, 13722, 43315, 50886, 9069, 919, 831, 110, 53333, 28, 15490, 25553, 6892, 81604, ...
[ 0.04296875, 0.117431640625, 0.07281494140625, 0.2105712890625, 0.252685546875, 0.122314453125, 0.000701904296875, 0.08551025390625, 0.13525390625, 0.187744140625, 0.08111572265625, 0.1793212890625, 0.1900634765625, 0.050018310546875, 0.2498779296875, 0.03948974609375, 0.1138916015625...
c944cdaf8b64a8d27b9b7f8e261a57b4133d77fe
subsection
25
34
Extensionality
Proposition (Extensionality for closed terms) Suppose \Gamma ,q\mathop |\Delta \vdash t and \Gamma ,q\mathop |\Delta \vdash u. If \Gamma \mathop |\Delta \vdash \nu _{i,j}q.t=\nu _{i,j}q.u for all i,j, then also \Gamma \mathop |\Delta \vdash t=u.We show the contrapositive. By the model completeness theorem (Thm. REF )...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 1250, 40322, 19279, 128872, 134393, 155738, 69407, 121691, 8364, 724, 36557, 864, 2227, 58052, 58598, 334, 155892, 808, 136, 75, 4263, 539, 170, 14, 34, 756, 17, 1369, 7639, 2304, 67890, 4935, 3311, 3299, 28484, 7432, 70, 58391, 9069, 919...
[ 0.1900634765625, 0.2279052734375, 0.08941650390625, 0.227783203125, 0.1319580078125, 0.2130126953125, 0.2314453125, 0.0897216796875, 0.0037841796875, 0.05291748046875, 0.2353515625, 0.1806640625, 0.1029052734375, 0.19189453125, 0.187744140625, 0.019073486328125, 0.119140625, 0.1751...
180b9f59b1856833caddb411cddf5c1a97dc896d
subsection
26
34
Extensionality
So, \llbracket \nu _{i_n,j_n}q.t\rrbracket \ne \llbracket \nu _{i_n,j_n}q.u\rrbracket .Proposition (Extensionality for ground terms) In brief: If {t[^{v_1 \dots v_k}\!/\!_{x_1\dots x_k}] = u[^{v_1 \dots v_k}\!/\!_{x_1\dots x_k}]} for all suitable ground v_1\dots v_k, then t =u.In detail: Consider t and u with -\matho...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 1061, 41872, 1181, 2844, 27853, 6, 539, 101, 14, 454, 19, 4, 170, 864, 18, 29161, 86, 24854, 8152, 34, 5, 10752, 40322, 19279, 128872, 134393, 100, 61585, 69407, 360, 59335, 4263, 334, 115187, 15464, 7, 81, 38, 64, 425, 1022, 92, 26...
[ 0.069091796875, 0.04241943359375, 0.1522216796875, 0.24658203125, 0.2362060546875, 0.0123291015625, 0.19140625, 0.00067138671875, 0.055877685546875, 0.051513671875, 0.0958251953125, 0.011871337890625, 0.165283203125, 0.123046875, 0.135986328125, 0.115478515625, 0.0673828125, 0.0125...
68484d6a98fcf9ab450c4d1de6d941b7b4ae3def
subsection
27
34
Extensionality
Then  nullfont\displaystyle \llbracket t[^{v_1\dots v_k}\!/\!_{x_1\dots x_k}]\rrbracket (1,0) = \llbracket t\rrbracket (0, {\dots },b_{I,k},{\dots }, 0) \ne \llbracket u\rrbracket (0,{\dots }, b_{I,k},{\dots }, 0) = \llbracket u[^{v_1\dots v_k}\!/\!_{x_1\dots x_k}]\rrbracket (1,0). \then \then   The required {\lnot \b...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 47009, 56574, 75635, 186341, 41872, 1181, 2844, 27853, 808, 334, 115187, 15464, 81, 454, 92, 38, 425, 1022, 29161, 206808, 2203, 22085, 275, 568, 757, 75, 876, 92733, 56065, 45621, 18, 32976, 2227, 58052, 53, 169, 53861, 155892, 28960, 36...
[ 0.036163330078125, 0.280517578125, 0.30078125, 0.00341796875, 0.003265380859375, 0.138916015625, 0.1409912109375, 0.158447265625, 0.1925048828125, 0.1900634765625, 0.1708984375, 0.1590576171875, 0.2015380859375, 0.0262298583984375, 0.1038818359375, 0.051483154296875, 0.06524658203125...
1cfffad08246efbc5392883ecfa6acff0af8b6bf
subsection
28
34
Relative syntactical completeness
Proposition (Neumann, ) If t,u are terms in the theory of rational convexity (Def. REF ), then either t=u is derivable or it implies x\mathop {{}_{i}\!{?}\!_{j}} y=x\mathop {{}_{i^{\prime }}\!{?}\!_{j^{\prime }}} y for all nonzero i,i^{\prime },j,j^{\prime }.The theory of Beta-Bernoulli is syntactically complete rela...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/bf01220869", "end": 85, "openalex_id": "https://openalex.org/W2315523708", "raw": "Walter D. Neumann. On the quasivariety of convex subsets of affine space. Arch. Math., 21:11–16, 1970.", "source_ref_id": "a988d7521c0da83d7425...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 1250, 40322, 6433, 34, 5761, 4263, 808, 4, 621, 69407, 154453, 168487, 158, 272, 425, 2481, 187423, 9069, 919, 40101, 1369, 83, 30057, 2661, 35388, 1022, 2227, 58052, 14, 38, 170, 113, 114654, 756, 351, 80510, 17, 111, 48941, 9, 33189, ...
[ 0.192138671875, 0.2091064453125, 0.0731201171875, 0.1826171875, 0.2066650390625, 0.03436279296875, 0.17529296875, 0.0276031494140625, 0.0311126708984375, 0.232177734375, 0.198974609375, 0.2169189453125, 0.1163330078125, 0.2271728515625, 0.1187744140625, 0.06793212890625, 0.1169433593...
b855084cae7a3fe6e061c349a7081d0b70f8b08d
subsection
29
34
Remark about stateful implementations
In the introduction we recalled the idea of using Pólya's urn to implement a Beta-Bernoulli process using local (hidden) state.Our equational presentation gives a recipe for understanding the correctness of the stateful implementation. First, one would give an operational semantics, and then a basic notion of observati...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/978-3-662-46678-0_18", "end": 531, "openalex_id": "https://openalex.org/W3106280144", "raw": "Ales Bizjak and Lars Birkedal. Step-indexed logical relations for probability. In Proc. FOSSACS 2015, pages 279–294, 2015.", "source...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 39864, 1363, 189232, 6528, 17368, 107859, 538, 11, 25, 2780, 19, 29479, 48941, 9, 33189, 3165, 7532, 9433, 4000, 15080, 555, 11341, 13722, 43315, 92252, 48564, 100094, 26785, 7432, 7844, 208124, 41018, 484, 109109, 62822, 110, 150556, 224743,...
[ 0.10040283203125, 0.00262451171875, 0.069580078125, 0.1033935546875, 0.081298828125, 0.0982666015625, 0.09454345703125, 0.1116943359375, 0.008819580078125, 0.1845703125, 0.1046142578125, 0.1585693359375, 0.1951904296875, 0.0157623291015625, 0.0838623046875, 0.1920166015625, 0.1899414...
a6a5fcd10ab865df620f4451fccfac219642ccf2
subsection
30
34
Conclusion
Exchangeable random processes are central to many Bayesian models. The general message of this paper is that the analysis of exchangeable random processes, based on basic concepts from programming language theory, depends on three crucial ingredients: commutativity, discardability, and abstract types. We have illustrat...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1198/016214506000000302", "end": 1163, "openalex_id": "https://openalex.org/W2158266063", "raw": "Yee Whye Teh, Michael I. Jordan, Matthew J. Beal, and David M. Blei. Hierarchical Dirichlet processes. J. Amer. Statist. Assoc., 101(476):1...
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 104758, 2886, 96759, 9433, 9879, 5941, 9631, 90, 3378, 115774, 4537, 26008, 15122, 114137, 121122, 4, 35509, 98, 62822, 23755, 7, 56037, 449, 46876, 154453, 56566, 17262, 106157, 145170, 375, 68754, 54613, 53, 17116, 5861, 41159, 164789, 5289...
[ 0.259521484375, 0.211669921875, 0.194091796875, 0.17626953125, 0.12255859375, 0.047943115234375, 0.1712646484375, 0.1370849609375, 0.116455078125, 0.21337890625, 0.0758056640625, 0.1107177734375, 0.045074462890625, 0.1171875, 0.269775390625, 0.03521728515625, 0.00787353515625, 0.03...
65a21fcffa03376c3940e4b3306f48e8d92bb5fd
subsection
31
34
Example derivations
In this appendix, we derive equations mentioned in the main text of the paper. The first equation, as found in §REF , is  nullfont\displaystyle x \mathop {{}_{1}\!{?}\!_{1}} y = ((x \mathop {{}_{1}\!{?}\!_{1}} y) \mathop {?_{p}} x) \mathop {?_{p}} (y \mathop {?_{p}} (x \mathop {{}_{1}\!{?}\!_{1}} y)) \then \then   whic...
{ "cite_spans": [] }
10.4230/LIPIcs.ICALP.2018.141
1802.09598
The Beta-Bernoulli process and algebraic effects
[ "Sam Staton", "Dario Stein", "Hongseok Yang", "Nathanael L. Ackerman", "Cameron E. Freer", "Daniel M. Roy" ]
[ "cs.PL" ]
2,018
en
Computer Science
[ 360, 903, 114689, 428, 425, 642, 122, 5844, 28, 13722, 5256, 119056, 5201, 7986, 15122, 5117, 5490, 2320, 237, 14037, 5360, 11766, 919, 83, 56574, 75635, 186341, 7786, 1022, 41872, 2227, 58052, 454, 418, 38, 32, 113, 2203, 60494, 254, 6...
[ 0.0294189453125, 0.0780029296875, 0.1959228515625, 0.1947021484375, 0.1192626953125, 0.046356201171875, 0.1776123046875, 0.14892578125, 0.148193359375, 0.2568359375, 0.136474609375, 0.135009765625, 0.099365234375, 0.134033203125, 0.161376953125, 0.205322265625, 0.2841796875, 0.1757...