plateform stringclasses 1
value | repo_name stringlengths 13 113 | name stringlengths 3 74 | ext stringclasses 1
value | path stringlengths 12 229 | size int64 23 843k | source_encoding stringclasses 9
values | md5 stringlengths 32 32 | text stringlengths 23 843k |
|---|---|---|---|---|---|---|---|---|
github | bsxfan/meta-embeddings-master | tracer.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_library/test/tracer.m | 1,081 | utf_8 | 5e8d7ea9aefc9d1c1cc8161546bd9483 | function [w,deriv] = tracer(w,vstring,gstring,jstring)
% This is an MV2DF. See MV2DF_API_DEFINITION.readme.
%
% Applies linear transform y = map(w). It needs the transpose of map,
% transmap for computing the gradient. map and transmap are function
% handles.
if nargin==0
test_this();
return;
end
if nargin<2... |
github | bsxfan/meta-embeddings-master | test_MV2DF_noHess.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_library/test/test_MV2DF_noHess.m | 1,125 | utf_8 | 2af48174c2441c011dffbf316b93612d | function test_MV2DF_noHess(f,x0)
%id_in = identity_trans([]);
%id_out = identity_trans([]);
%f = f(id_in);
%f = id_out(f);
x0 = x0(:);
Jc = cstepJacobian(f,x0);
Jr = rstepJacobian(f,x0);
[y0,deriv] = f(x0);
m = length(y0);
n = length(x0);
J2 = zeros(size(Jr));
for i=1:m;
y = zeros(m,1);
y(i) = 1;
J2(i,... |
github | bsxfan/meta-embeddings-master | inv_lu2.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/utils/inv_lu2.m | 1,044 | utf_8 | 8fa16d13c5b1ad8e681b3f2ba0f9b2c9 | function [inv_map,bi_inv_map,logdet,invA] = inv_lu2(A)
% INV_LU2
% Does a LU decomposition on A and returns logdet, inverse and
% two function handles that respectively map X to A\X and A\X/A.
%
if nargin==0
test_this();
return;
end
[L,T,p] = lu(A,'vector');
P = sparse(p,1:length(p),1);
% P... |
github | bsxfan/meta-embeddings-master | invchol2.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/utils/invchol2.m | 968 | utf_8 | 936256e3c3a28ed65ad0c15d9fbb04cd | function [inv_map,bi_inv_map,logdet,invA] = invchol2(A)
% INVCHOL2
% Does a Cholesky decomposition on A and returns logdet, inverse and
% two function handles that respectively map X to A\X and A\X/A.
%
if nargin==0
test_this();
return;
end
if isreal(A)
R = chol(A); %R'*R = A
inv_map = @(X) R\(R'\X);... |
github | bsxfan/meta-embeddings-master | invchol_or_lu.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/utils/invchol_or_lu.m | 1,418 | utf_8 | 468d08dd52dd54bb77a471a5e2d0a856 | function [inv_map,bi_inv_map,logdet,invA] = invchol_or_lu(A)
% INVCHOL_OR_LU
% Does a Cholesky decomposition on A and returns logdet, inverse and
% two function handles that respectively map X to A\X and A\X/A.
%
if nargin==0
test_this();
return;
end
if isreal(A)
R = chol(A); %R'*R = A
inv_map = @(X)... |
github | bsxfan/meta-embeddings-master | invchol_taylor.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/utils/invchol_taylor.m | 1,241 | utf_8 | 0f52f57c84dc1ae326e32c031541c496 | function [inv_map,logdet] = invchol_taylor(A)
% Does a Cholesky decomposition on A and returns:
% inv_map: a function handle to solve for X in AX = B
% logdet (of A)
%
% This code is designed to work correctly if A has a small complex
% perturbation, such as used in complex step differentiation, even though
% the ... |
github | bsxfan/meta-embeddings-master | train_system.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/discrim_training/train_system.m | 4,969 | utf_8 | 69b1726b853595599d4a79414b256c8b | function [w,mce,divergence,w_pen,c_pen,optimizerState,converged] = train_system(classf,system,penalizer,W0,lambda,confusion,maxiters,maxCG,prior,optimizerState)
%
% Supervised training of a regularized K-class linear logistic
% regression. Allows regularization via weight penalties and via
% label confusion prob... |
github | bsxfan/meta-embeddings-master | sum_of_functions.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/sum_of_functions.m | 1,094 | utf_8 | af1885792c3ce587c098ffc61a10cc06 | function [y,deriv] = sum_of_functions(w,weights,f,g)
% This is an MV2DF (see MV2DF_API_DEFINITION.readme) which
% represents the new function, s(w), obtained by summing the
% weighted outputs of the given functions:
% s(w) = sum_i weights(i)*functions{i}(w)
%
% Usage examples:
%
% s = @(w) sum_of_functions(w,[1,... |
github | bsxfan/meta-embeddings-master | scale_function.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/scale_function.m | 856 | utf_8 | fae26a24cbea0fcc7ae35cf1642b18e4 | function [y,deriv] = scale_function(w,scale,f)
% This is an MV2DF (see MV2DF_API_DEFINITION.readme) which
% represents the new function,
%
% g(w) = scale(w)*f(w),
%
% where scale is scalar-valued and f is matrix-valued.
%
%
% Here scale and f are function handles to MV2DF's.
if nargin==0
test_this();
r... |
github | bsxfan/meta-embeddings-master | outerprod_of_functions.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/outerprod_of_functions.m | 1,085 | utf_8 | 731782f761b675bb6d3567ddb560c950 | function [y,deriv] = outerprod_of_functions(w,f,g,m,n)
% This is an MV2DF (see MV2DF_API_DEFINITION.readme) which
% represents the new function,
%
% g(w) = f(w)g(w)'
%
% where f(w) and g(w) are column vectors of sizes m and n respectively.
%
% Here f,g are function handles to MV2DF's.
if nargin==0
test_thi... |
github | bsxfan/meta-embeddings-master | interleave.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/interleave.m | 2,028 | utf_8 | 0cdd5849311559d9813888914f7530cd | function [y,deriv] = interleave(w,functions)
% interleave is an MV2DF (see MV2DF_API_DEFINITION.readme) which
% represents the new function, s(w), obtained by interleaving the outputs of
% f() and g() thus:
%
% S(w) = [f(w)';g(w)'];
% s(w) = S(:);
if nargin==0
test_this();
return;
end
if isempty(w)... |
github | bsxfan/meta-embeddings-master | shift_function.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/shift_function.m | 896 | utf_8 | 82abefaa89d6e02403f0b543a3c69a0b | function [y,deriv] = shift_function(w,shift,f)
% This is an MV2DF (see MV2DF_API_DEFINITION.readme) which
% represents the new function,
%
% g(w) = shift(w)+f(w),
%
% where shift is scalar-valued and f is matrix-valued.
%
%
% Here shift and f are function handles to MV2DF's.
if nargin==0
test_this();
r... |
github | bsxfan/meta-embeddings-master | dotprod_of_functions.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/dotprod_of_functions.m | 952 | utf_8 | 2999899143500736ecbc06d5afc09df0 | function [y,deriv] = dotprod_of_functions(w,f,g)
% This is an MV2DF (see MV2DF_API_DEFINITION.readme) which
% represents the new function,
%
% g(w) = f(w)'g(w)
%
% where f(w) and g(w) are column vectors of the same size.
%
% Here f,g are function handles to MV2DF's.
if nargin==0
test_this();
return;
en... |
github | bsxfan/meta-embeddings-master | dottimes_of_functions.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/dottimes_of_functions.m | 641 | utf_8 | 45195e5dddb789f3431e2f84495b06c7 | function [y,deriv] = dottimes_of_functions(w,A,B)
% This is an MV2DF (see MV2DF_API_DEFINITION.readme)
%
% w --> A(w) .* B(w)
%
% Here A and B are function handles to MV2DF's.
if nargin==0
test_this();
return;
end
if isempty(w)
s = stack(w,A,B);
y = dottimes(s);
return;
end
if isa(w,'fun... |
github | bsxfan/meta-embeddings-master | replace_hessian.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/replace_hessian.m | 1,399 | utf_8 | 1e948d50795df5102876278fd5022da8 | function [y,deriv] = replace_hessian(w,f,cstep)
% This is an MV2DF. See MV2DF_API_DEFINITION.readme.
%
if nargin==0
test_this();
return;
end
if isempty(w)
y = @(w)replace_hessian(w,f,cstep);
return;
end
if isa(w,'function_handle')
outer = replace_hessian([],f,cstep);
y = compose_mv(outer,w,[... |
github | bsxfan/meta-embeddings-master | product_of_functions.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/product_of_functions.m | 745 | utf_8 | ae86bc0a8429bacd6044704a6a8a0e06 | function [y,deriv] = product_of_functions(w,A,B,m,k,n)
% This is an MV2DF (see MV2DF_API_DEFINITION.readme)
%
% w --> vec ( reshape(A(w),m,k) * reshape(B(w),k,n) )
%
% Here A and B are function handles to MV2DF's.
if nargin==0
test_this();
return;
end
if isempty(w)
s = stack(w,A,B);
y = gemm(s,... |
github | bsxfan/meta-embeddings-master | stack.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/stack.m | 3,136 | utf_8 | cbfe5ccd3255b5021692c3eb13e1798f | function [y,deriv] = stack(w,f,g,eqlen)
% STACK is an MV2DF (see MV2DF_API_DEFINITION.readme) which
% represents the new function, s(w), obtained by stacking the outputs of
% f() and g() thus:
% s(w) = [f(w);g(w)]
if nargin==0
test_this();
return;
end
if ~exist('eqlen','var')
eqlen = false;
end
if i... |
github | bsxfan/meta-embeddings-master | scale_and_translate.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/scale_and_translate.m | 1,341 | utf_8 | 121b1cd2e23a3d7111f310db8e3b6a05 | function [y,deriv] = scale_and_translate(w,vectors,params,m,n)
% This is an MV2DF (see MV2DF_API_DEFINITION.readme) which
% represents the new function, obtained by scaling and translating the
% column vectors of the output matrix of the function vectors(w). The
% scaling and translation parameters, params(w) is also... |
github | bsxfan/meta-embeddings-master | compose_mv.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/utility_funcs/Optimization_Toolkit/MV2DF/function_combination/compose_mv.m | 2,958 | utf_8 | 108f7eb78b4ff77e907d369fc9ae14db | function [y,deriv] = compose_mv(outer,inner,x)
% COMPOSE_MV is an MV2DF (see MV2DF_API_DEFINITION.readme) which represents
% the combination of two functions. If 'outer' is an MV2DF for a function
% g() and 'inner' for a function f(), then this MV2DF represents g(f(x)).
%feature scopedaccelenablement off
if nargin==0... |
github | bsxfan/meta-embeddings-master | pav_calibration.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/calibration/pav_calibration.m | 2,716 | utf_8 | 2a9298835be5d7757fb4d660a5a2d7b3 | function [pav_trans,score_bounds,llr_bounds] = pav_calibration(tar,non,small_val)
% Creates a calibration transformation function using the PAV algorithm.
% Inputs:
% tar: A vector of target scores.
% non: A vector of non-target scores.
% small_val: An offset to make the transformation function
% invertible. ... |
github | bsxfan/meta-embeddings-master | align_with_ndx.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/classes/@Scores/align_with_ndx.m | 2,628 | utf_8 | 5899b5e5bd43dea8280d84cea8fdf0ec | function aligned_scr = align_with_ndx(scr,ndx)
% The ordering in the output Scores object corresponds to ndx, so
% aligning several Scores objects with the same ndx will result in
% them being comparable with each other.
% Inputs:
% scr: a Scores object
% ndx: a Key or Ndx object
% Outputs:
% aligned_scr: scr res... |
github | bsxfan/meta-embeddings-master | filter.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/classes/@Scores/filter.m | 2,622 | utf_8 | ebaa2297b42e23384ffa07c12bdcc005 | function outscr = filter(inscr,modlist,seglist,keep)
% Removes some of the information in a Scores object. Useful for
% creating a gender specific score set from a pooled gender score
% set. Depending on the value of 'keep', the two input lists
% indicate the models and test segments (and their associated
% scores) t... |
github | bsxfan/meta-embeddings-master | filter.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/classes/@Key/filter.m | 3,047 | utf_8 | 9274e13ab0bf80ca9a90fd6f46da8ff0 | function outkey = filter(inkey,modlist,seglist,keep)
% Removes some of the information in a key. Useful for creating a
% gender specific key from a pooled gender key. Depending on the
% value of 'keep', the two input lists indicate the strings to
% retain or the strings to discard.
% Inputs:
% inkey: A Key object.
... |
github | bsxfan/meta-embeddings-master | read_hdf5.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/classes/@Key/read_hdf5.m | 1,196 | utf_8 | 4057278a996259de22fed6ee29c5d3b2 | function key = read_hdf5(infilename)
% Reads a Key object from an hdf5 file.
% Inputs:
% infilename: The name for the hdf5 file to read.
% Outputs:
% key: A Key object created from the information in the hdf5
% file.
assert(nargin==1)
assert(isa(infilename,'char'))
key = Key();
key.modelset = h5strings_to_ce... |
github | bsxfan/meta-embeddings-master | filter.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/classes/@Ndx/filter.m | 2,788 | utf_8 | 6d39760ecafc786f43259d1adb98a810 | function outndx = filter(inndx,modlist,seglist,keep)
% Removes some of the information in an Ndx. Useful for creating a
% gender specific Ndx from a pooled gender Ndx. Depending on the
% value of 'keep', the two input lists indicate the strings to
% retain or the strings to discard.
% Inputs:
% inndx: An Ndx object... |
github | bsxfan/meta-embeddings-master | read_hdf5.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/classes/@Ndx/read_hdf5.m | 838 | utf_8 | 424ae971c22eb22cf8c27af6130b9698 | function ndx = read_hdf5(infilename)
% Creates an Ndx object from the information in an hdf5 file.
% Inputs:
% infilename: The name of the hdf5 file contain the information
% necessary to construct an Ndx object.
% Outputs:
% ndx: An Ndx object containing the information in the input
% file.
assert(nargin=... |
github | bsxfan/meta-embeddings-master | filter_on_right.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/classes/@Id_Map/filter_on_right.m | 1,885 | utf_8 | deb124220c828ae065475bc93957d53f | function out_idmap = filter_on_right(in_idmap,idlist,keep)
% Removes some of the information in an idmap. Depending on the
% value of 'keep', the idlist indicates the strings to retain or
% the strings to discard.
% Inputs:
% in_idmap: An Id_Map object to be pruned.
% idlist: A cell array of strings which will be ... |
github | bsxfan/meta-embeddings-master | read_hdf5.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/classes/@Id_Map/read_hdf5.m | 777 | utf_8 | 47581f23817e49ffc325aed95a088106 | function idmap = read_hdf5(infilename)
% Creates an Id_Map object from the information in an hdf5 file.
% Inputs:
% infilename: The name of the hdf5 file containing the information
% necessary to construct an Id_Map object.
% Outputs:
% idmap: An Id_Map object containing the information in the input
% file.... |
github | bsxfan/meta-embeddings-master | filter_on_left.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/bosaris_toolkit/classes/@Id_Map/filter_on_left.m | 1,871 | utf_8 | 27abe0c92b6ff488892e389fee1fb5e9 | function out_idmap = filter_on_left(in_idmap,idlist,keep)
% Removes some of the information in an idmap. Depending on the
% value of 'keep', the idlist indicates the strings to retain or
% the strings to discard.
% Inputs:
% in_idmap: An Id_Map object to be pruned.
% idlist: A cell array of strings which will be c... |
github | bsxfan/meta-embeddings-master | L_BFGS.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/optimization/L_BFGS.m | 5,021 | utf_8 | ccfcc8e580c4dfa191a42bfcbaf055cb | function [w,y,mem,logs] = L_BFGS(obj,w,maxiters,timeout,mem,stpsz0,callback)
% L-BFGS Quasi-Newton unconstrained optimizer.
% -- This has a small interface change from LBFGS.m --
%
% Inputs:
% obj: optimization objective, with interface: [y,grad] = obj(w),
% where w is the parameter vector, y is the scalar o... |
github | bsxfan/meta-embeddings-master | create_PYCRP.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/CRP/create_PYCRP.m | 13,438 | utf_8 | 97417297bf1d08ceaa6be2bbfe82c2bb | function PYCRP = create_PYCRP(alpha,beta,e,n)
% alpha: alpha>=0, concentration
% beta: 0<= beta <=1, discount
if nargin==0
%test_this2();
test_Gibbsmatrix()
return;
end
if nargin==4
PYCRP = create_PYCRP(1,0);
PYCRP.set_expected_number_tables(e,n,alpha,be... |
github | bsxfan/meta-embeddings-master | rand_fake_Dirichlet.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/language_recognition/synth_data/rand_fake_Dirichlet.m | 810 | utf_8 | 3070cfe2d8487d8ae48c7740fdc24c2d | function R = rand_fake_Dirichlet(alpha,m,n)
% This is no longer Dirichlet. I replaced it with a faster ad-hoc
% distribution.
%
% Generates m-by-n matrix of n samples from m-category Dirichlet, with
% concentration parameter: alpha > 0.
if nargin==0
test_this();
return;
end
%R = reshape(ra... |
github | bsxfan/meta-embeddings-master | create_T_backend.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/language_recognition/T_backend/create_T_backend.m | 6,482 | utf_8 | ee013ab7facb848049ba586dfb7b6f33 | function TBE = create_T_backend(nu,dim,K)
% Create a (multivariate) T-distribution generative backend for multiclass classification.
% The classes have different means, but the scatter matrix and degrees of
% freedom are common to all clases.
%
% This object provides a method for supervised ML training (EM algorithm),
... |
github | bsxfan/meta-embeddings-master | train_TLDIvector.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/language_recognition/TLDIvector/train_TLDIvector.m | 4,374 | utf_8 | 6029b83917586ea3e2977c8fad616093 | function [W,Mu,TT] = train_TLDIvector(stats_or_ivectors,N,T,TT,nu,labels,niters,W,Mu)
% Inputs:
% stats_or_ivectors: can be either F, or ivectors
% F: dm-by-n first-order stats (m: UBM size; d: feature dim; n: no segments)
% ivectors: k-by-n, classical i-vector point-estimates
% N: m-by-n zero order st... |
github | bsxfan/meta-embeddings-master | create_diagonalized_C.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/language_recognition/TLDIvector/create_diagonalized_C.m | 4,986 | utf_8 | c1a642f817fa817097bf89b070af4c04 | function C = create_diagonalized_C(B,R,RM,Ra,W,M,a)
% Creates object to represent: C = inv(lambda W + B),
%
% Inputs:
% B: positive definite matrix (i-vector dimension)
% R: chol(W), so that R'R=W (i-vector dimension)
% RM: R*M, where M has language means in columns
% Ra: (R')\a, vector (i-vector dime... |
github | bsxfan/meta-embeddings-master | create_augmenting_backend.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/language_recognition/augmentation/create_augmenting_backend.m | 3,599 | utf_8 | 852abc76e3f7dfa8796f7015880bc788 | function ABE = create_augmenting_backend(nu,dim,T,K,L)
% Inputs:
% nu: scalar nu>0, t-distribution degrees of freedom
% dim: ivector dimension
% T: i-vector extr\zctor T-matrix
% K: UBM size
% L: number of languages
if nargin==0
test_this();
return;
end
assert(dim==size(T,2));
... |
github | bsxfan/meta-embeddings-master | testBackprop_rs.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/test/testBackprop_rs.m | 2,198 | utf_8 | 9d88acaf06002d97bf8eb2fdc07bf7b8 | function total = testBackprop_rs(block,X,delta,mask)
%same as testFBblock, but with real step
if ~iscell(X)
X = {X};
end
dX = cellrndn(X);
if exist('mask','var')
assert(length(mask)==length(X));
dX = cellmask(dX,mask);
end
cX1 = cellstep(X,dX,delta);
cX2 = cell... |
github | bsxfan/meta-embeddings-master | testBackprop_multi.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/test/testBackprop_multi.m | 2,361 | utf_8 | ced4a5b7e925d5c00a2d991fb34d012c | function total = testBackprop_multi(block,nout,X,mask)
% same as testBackprop, but handles multiple outputs
if ~iscell(X)
X = {X};
end
dX = cellrndn(X);
if exist('mask','var')
assert(length(mask)==length(X));
dX = cellmask(dX,mask);
end
cX = cellcomplex(X,dX);
D... |
github | bsxfan/meta-embeddings-master | testBackprop.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/test/testBackprop.m | 2,015 | utf_8 | fc73fa404f5441c097eb63f249106078 | function total = testBackprop(block,X,mask)
if ~iscell(X)
X = {X};
end
dX = cellrndn(X);
if exist('mask','var')
assert(length(mask)==length(X));
dX = cellmask(dX,mask);
end
cX = cellcomplex(X,dX);
DX = cell(size(X));
[Y,back] = block(X{:});
... |
github | bsxfan/meta-embeddings-master | create_univariate_Laplace_node.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/create_univariate_Laplace_node.m | 2,432 | utf_8 | 723752e00cd25de77113d0e0c21ac71e | function node = create_univariate_Laplace_node(prior)
if nargin==0
test_this();
return;
end
value = prior.sample();
%posterior stuff
post_mu = []; %approximate posterior mean
post_h = []; %approximate posterior hessian (- precision)
%logpost = []; %unnormalized true lo... |
github | bsxfan/meta-embeddings-master | create_symmetricDirichlet_node.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/create_symmetricDirichlet_node.m | 5,749 | utf_8 | 29d843a52a81df9405bb868555c339bf | function node = create_symmetricDirichlet_node(logalpha,sz)
% Creates symmetric Dirichlet node (part of a Bayesian network), that is
% equippped to do alternating Gibbs sampling. This node expects a
% non-conjugate (upstream) hyper-prior to supply the Dirichlet parameter
% alpha. It also expects (downstream) one or mor... |
github | bsxfan/meta-embeddings-master | createDPMM.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/createDPMM.m | 1,065 | utf_8 | 6c6b4270dc5b4c5c841b1f8101fad8d6 | function model = createDPMM(W,B,crp)
if nargin==0
test_this();
return;
end
cholW = chol(W);
cholB = chol(B);
dim = size(W,1);
model.sample = @sample;
function [X,Means,hlabels,counts] = sample(n)
[labels,counts] = crp.sample(n);
m = length(counts)... |
github | bsxfan/meta-embeddings-master | create_truncGMM.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/create_truncGMM.m | 11,488 | utf_8 | 60dcde7a91887483186b7ebd904866fc | function model = create_truncGMM(W,F,alpha,m)
% This is a truncated version of DP micture model, with a specified maximum number of
% components. The observations are realted to the hidden cluster variables
% like in an SPLDA model. The hidden variable for cluster i is z_i in R^d.
% The observations, x_j are in R^D, wh... |
github | bsxfan/meta-embeddings-master | test_GMM_Gibbs.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/test_GMM_Gibbs.m | 3,062 | utf_8 | 23d78374c004d789decbe42ff4a4e0a9 | function test_GMM_Gibbs()
close all;
dim = 500;
tame = 10;
sep = 0.6e-3; %increase to move clusters further apart in smulated data
%sep = sep*200;
%sep = 1;
alpha0 = 200; %increase to get more clusters
n = 1000;
m = 300;
alpha = alpha0/m;
... |
github | bsxfan/meta-embeddings-master | create_GibbsGMM.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/create_GibbsGMM.m | 11,427 | utf_8 | 39c2d8aee27ef2a790747bef2aae64c2 | function model = create_GibbsGMM(W,F,alphaPrior,m)
% This is a truncated version of DP micture model, with a specified maximum number of
% components. The observations are realted to the hidden cluster variables
% like in an SPLDA model. The hidden variable for cluster i is z_i in R^d.
% The observations, x_j are in R^... |
github | bsxfan/meta-embeddings-master | randg.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/randg.m | 2,448 | utf_8 | bf1edfa92d41e108e3e1e2ef55461799 | function G = randg(alpha,m,n)
% Generates an m-by-n matrix of random gamma variates, having scale = 1
% and shape alpha.
% Inputs:
% alpha: scalar, vector or matrix
% m,n: [optional] size of output matrix. If not given, the size is the
% same as that of alpha. If given, then alpha should be m-by-n, or an
% m... |
github | bsxfan/meta-embeddings-master | test_GMM_Gibbs_Balerion.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/test_GMM_Gibbs_Balerion.m | 3,075 | utf_8 | e0afdf1916c50ecfbb2769db53b73d08 | function test_GMM_Gibbs_Balerion()
close all;
dim = 500;
tame = 10;
sep = 0.6e-3; %increase to move clusters further apart in smulated data
%sep = sep*200;
%sep = 1;
alpha0 = 1000; %increase to get more clusters
n = 100000;
m = 2000;
alpha = al... |
github | bsxfan/meta-embeddings-master | dualAveraging.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/NUTS-matlab-master/dualAveraging.m | 3,700 | utf_8 | 95a22ca6047c33503568b405eb9968ef | function [theta, epsilonbar, epsilon_seq, epsilonbar_seq] = dualAveraging(f, theta0, delta, n_warmup, n_updates)
% function [theta, epsilonbar, epsilon_seq, epsilonbar_seq] = dualAveraging(f, n_warmup, theta0, delta, n_updates)
%
% Adjusts the step-size of NUTS by the dual-averaging (stochastic
% optimization) algorit... |
github | bsxfan/meta-embeddings-master | ESS.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/NUTS-matlab-master/ESS.m | 2,793 | utf_8 | 2f52219f2530dd16c3f9b2a395217f66 | function [ess, auto_cor] = ESS(x, mu, sigma_sq)
% function [ess, auto_cor] = ESS(x, mu, sigma_sq)
%
% Returns an estimate of effective sample sizes of a Markov chain 'x(:,i)'
% for each i. The estimates are based on the monotone positive sequence estimator
% of "Practical Markov Chain Monte Carlo" by Geyer (1992). The ... |
github | bsxfan/meta-embeddings-master | NUTS.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/NUTS-matlab-master/NUTS.m | 8,078 | utf_8 | 7703b67677b18bbfd8fbcce670366463 | function [theta, alpha_ave, nfevals, logp, grad] = NUTS(f, epsilon, theta0, logp0, grad0, max_tree_depth)
% function [theta, grad, logp, nfevals, alpha_ave] = NUTS(f, epsilon, theta0, max_tree_depth, logp0, grad0)
%
% Carries out one iteration of No-U-Turn-Sampler.
%
% Args:
% f - function handle: returns the log proba... |
github | bsxfan/meta-embeddings-master | ReNUTS.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/NUTS-matlab-master/Recycle/ReNUTS.m | 10,790 | utf_8 | 916dd2121fbfb133e73fe81c36d2961b | function [theta, re_theta, alpha_ave, nfevals, logp, grad] = ...
ReNUTS(f, epsilon, theta0, nsample, logp0, grad0, max_tree_depth)
% function [theta, grad, logp, nfevals, alpha_ave] = ...
% ReNUTS(f, epsilon, theta0, n_recycle, logp0, grad0, max_tree_depth)
%
% Carries out one iteration of Recycled No-U-Turn-Sa... |
github | bsxfan/meta-embeddings-master | d2p.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/t-sne/d2p.m | 3,155 | utf_8 | dc48b1dd0688d11671d81af50a0970de | function [P, beta] = d2p(D, u, tol)
%D2P Identifies appropriate sigma's to get kk NNs up to some tolerance
%
% [P, beta] = d2p(D, kk, tol)
%
% Identifies the required precision (= 1 / variance^2) to obtain a Gaussian
% kernel with a certain uncertainty for every datapoint. The desired
% uncertainty can be specified... |
github | bsxfan/meta-embeddings-master | x2p.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/t-sne/x2p.m | 3,297 | utf_8 | e0d6a8b9bcdd6ebd97037e46252fe200 | function [P, beta] = x2p(X, u, tol)
%X2P Identifies appropriate sigma's to get kk NNs up to some tolerance
%
% [P, beta] = x2p(xx, kk, tol)
%
% Identifies the required precision (= 1 / variance^2) to obtain a Gaussian
% kernel with a certain uncertainty for every datapoint. The desired
% uncertainty can be specifie... |
github | bsxfan/meta-embeddings-master | create_PYCRP.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/CRP/create_PYCRP.m | 13,438 | utf_8 | 97417297bf1d08ceaa6be2bbfe82c2bb | function PYCRP = create_PYCRP(alpha,beta,e,n)
% alpha: alpha>=0, concentration
% beta: 0<= beta <=1, discount
if nargin==0
%test_this2();
test_Gibbsmatrix()
return;
end
if nargin==4
PYCRP = create_PYCRP(1,0);
PYCRP.set_expected_number_tables(e,n,alpha,be... |
github | bsxfan/meta-embeddings-master | trandn.m | .m | meta-embeddings-master/code/Niko/matlab/fous-y-tout/dpmm/trunc_gaussian/trandn.m | 3,445 | utf_8 | 319949967a84b816a6cc4d0b882c4c98 | function x=trandn(l,u)
%% truncated normal generator
% * efficient generator of a vector of length(l)=length(u)
% from the standard multivariate normal distribution,
% truncated over the region [l,u];
% infinite values for 'u' and 'l' are accepted;
% * Remark:
% If you wish to simulate a random variable
% 'Z' from the ... |
github | bsxfan/meta-embeddings-master | VB4HTPLDA_iteration.m | .m | meta-embeddings-master/code/Niko/matlab/clean/VB4HTPLDA/VB4HTPLDA_iteration.m | 4,532 | utf_8 | 3639a9571c692326b5102546e66240c3 | function [F,W,obj] = VB4HTPLDA_iteration(nu,F,W,R,labels,weights)
% Iteration of VB algorithm for HT-PLDA training. See HTPLDA_SGME_train_VB()
% for details. The model parameters F and W are updated.
%
% Inputs:
% nu: scalar, df > 0 (nu=inf is allowed: it signals G-PLDA)
% F: D-by-d factor loading matrix, D > d
% ... |
github | bsxfan/meta-embeddings-master | VB4HTPLDA_iteration_2.m | .m | meta-embeddings-master/code/Niko/matlab/clean/VB4HTPLDA/VB4HTPLDA_iteration_2.m | 4,534 | utf_8 | 50e11eb8be6844dbc9c70f60f49c3c23 | function [F,W,obj] = VB4HTPLDA_iteration_2(nu,F,W,R,labels,weights)
% Iteration of VB algorithm for HT-PLDA training. See HTPLDA_SGME_train_VB()
% for details. The model parameters F and W are updated.
%
% Inputs:
% nu: scalar, df > 0 (nu=inf is allowed: it signals G-PLDA)
% F: D-by-d factor loading matrix, D > d
... |
github | bsxfan/meta-embeddings-master | VB4HTPLDA_demo.m | .m | meta-embeddings-master/code/Niko/matlab/clean/VB4HTPLDA/VB4HTPLDA_demo.m | 7,965 | utf_8 | 52ba6282c157b56dd41335179ade4206 | function VB4HTPLDA_demo
% Demo and test code for VB training and SGME scoring of HT-PLDA model.
%
% Training and evaluation data are (independently) sampled from a model with
% randomly generated data. A VB algorithm is used to estimate the parameters
% of this model from the training data. The accuracy of the traine... |
github | bsxfan/meta-embeddings-master | MLNDA_MAP_obj.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/MLNDA_MAP_obj.m | 1,526 | utf_8 | c25e0cb0548424de1497aa4331334429 | function [y,back] = MLNDA_MAP_obj(newData,newLabels,oldData,oldLabels,oldWeight,F,W,fi,params,nu)
if nargin==0
test_this();
return;
end
ht = exist('nu','var') && ~isempty(nu) && ~isinf(nu);
[newR,logdetJnew,back1] = fi(params,newData);
[oldR,logdetJold,back3] = fi(params,oldD... |
github | bsxfan/meta-embeddings-master | logdetNice.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/logdetNice.m | 864 | utf_8 | 3dbf0e9b2c0ef0c688e5d05ea7956379 | function [logdet,back] = logdetNice(sigma,R,d)
if nargin==0
test_this();
return;
end
RR = R.'*R;
S = RR/sigma + diag(1./d);
[L,U] = lu(S);
dim = size(R,1);
logdet = ( sum(log(diag(U).^2)) + sum(log(d.^2)) + dim*log(sigma^2) ) /2;
back = @back_this;
functio... |
github | bsxfan/meta-embeddings-master | logdetNice4.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/logdetNice4.m | 1,046 | utf_8 | 62f8d2e7cac7cc028322c8d71f81b2e9 | function [logdet,back] = logdetNice4(D,L,R)
if nargin==0
test_this();
return;
end
[~,rank] = size(L);
DL = bsxfun(@ldivide,D,L);
RDL = R.'*DL;
S = RDL + eye(rank);
[Ls,Us] = lu(S);
logdet = ( sum(log(diag(Us).^2)) + sum(log(D.^2)) ) /2;
back = @back_this;
... |
github | bsxfan/meta-embeddings-master | htplda_llh.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/htplda_llh.m | 2,103 | utf_8 | d1bda1755ab4932ab4a22c90e64c4d1b | function [llh,back] = htplda_llh(R,labels,F,W,nu)
% Inputs:
% R: D-by-N, i-vectors
% labels: sparse, logical K-by-N, one hot columns, K speakers, N recordings
if nargin==0
test_this();
return;
end
[D,d] = size(F);
FW = F.'*W;
FWR = FW*R;
S = FWR*labels.'; %f... |
github | bsxfan/meta-embeddings-master | create_nice_Trans4.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/create_nice_Trans4.m | 2,926 | utf_8 | acc1b16001cab1bbc286b4f1c9d385de | function [f,fi,paramsz,fe] = create_nice_Trans4(dim,rank)
% Creates affine transform, having a matrix: M = D + L*R.', where D is
% diagonal and L and R are of low rank. The forward transform is:
% f(X) = M \ X + offset
if nargin==0
test_this();
return;
end
paramsz = dim + 2*dim*rank +... |
github | bsxfan/meta-embeddings-master | create_nice_Trans.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/create_nice_Trans.m | 3,126 | utf_8 | 564dc31754d76cf3ff3e1b29812bda01 | function [f,fi,paramsz,fe] = create_nice_Trans(dim,rank)
% Creates affine transform, having a matrix: M = sigma I + L*D*L.', where
% L is of low rank and D is diagonal. The forward transform is:
% f(X) = M \ X + offset
if nargin==0
test_this();
return;
end
paramsz = 1 + dim*rank + ran... |
github | bsxfan/meta-embeddings-master | logdetNice3.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/logdetNice3.m | 980 | utf_8 | 64ca857b7b319139b15a8c416ff1ac41 | function [logdet,back] = logdetNice3(sigma,L,R)
if nargin==0
test_this();
return;
end
[dim,rank] = size(L);
RL = R.'*L;
S = RL/sigma + eye(rank);
[Ls,Us] = lu(S);
logdet = ( sum(log(diag(Us).^2)) + dim*log(sigma^2) ) /2;
back = @back_this;
function [d... |
github | bsxfan/meta-embeddings-master | splda_llh.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/splda_llh.m | 1,253 | utf_8 | 9fa88758bc715c7d0d76209bd1936f6a | function [llh,back] = splda_llh(R,labels,F,W)
% Inputs:
% R: D-by-N, i-vectors
% labels: sparse, logical K-by-N, one hot columns, K speakers, N recordings
if nargin==0
test_this();
return;
end
FW = F.'*W;
FWR = FW*R;
S = FWR*labels.'; %first order natural parameter for sp... |
github | bsxfan/meta-embeddings-master | create_nice_Trans3.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/create_nice_Trans3.m | 2,943 | utf_8 | e197b146f2ee6588c45fc36d860a3db9 | function [f,fi,paramsz,fe] = create_nice_Trans3(dim,rank)
% Creates affine transform, having a matrix: M = sigma I + L*R.', where
% L and R are of low rank. The forward transform is:
% f(X) = M \ X + offset
if nargin==0
test_this();
return;
end
paramsz = 1 + 2*dim*rank + dim;
f =... |
github | bsxfan/meta-embeddings-master | create_nice_Trans2.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/create_nice_Trans2.m | 2,933 | utf_8 | aba4972b59d6acfb19b6a85c190bb897 | function [f,fi,paramsz,fe] = create_nice_Trans2(dim,rank)
if nargin==0
test_this();
return;
end
paramsz = 1 + dim*rank + rank*rank + dim;
f = @f_this;
fi = @fi_this;
fe = @expand;
function T = f_this(P,R)
[sigma,L,D,offset] = expand(P);
M = sigma*eye... |
github | bsxfan/meta-embeddings-master | logdetNice2.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/logdetNice2.m | 929 | utf_8 | 469e8c40a8fb711c201f8fbeaf1c8b32 | function [logdet,back] = logdetNice2(sigma,R,D)
if nargin==0
test_this();
return;
end
RR = R.'*R;
[Ld,Ud] = lu(D);
invD = inv(Ud)/Ld;
S = RR/sigma + invD;
[L,U] = lu(S);
dim = size(R,1);
logdet = ( sum(log(diag(U).^2)) + sum(log(diag(Ud).^2)) + dim*log(sigma^2) ) /2... |
github | bsxfan/meta-embeddings-master | LinvSR.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/param_adaptation/LinvSR.m | 562 | utf_8 | cc684a75c7911f7d518550620e6b8fc7 | function [Y,back] = LinvSR(L,S,R)
if nargin==0
test_this();
return;
end
Z = S\R;
Y = L*Z;
back = @back_this;
function [dL,dS,dR] = back_this(dY)
% Y = L*Z
dL = dY*Z.';
dZ = L.'*dY;
% Z = S\R;
dR = S.'\dZ;
... |
github | bsxfan/meta-embeddings-master | splda_adaptation_obj.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/param_adaptation/splda_adaptation_obj.m | 1,783 | utf_8 | 6c6dfe6b4fc0b2eedc0ac2e4802de451 | function [y,back] = splda_adaptation_obj(newData,labels,oldF,oldW,params,num_new_Fcols,W_adj_rank,slow)
if nargin==0
test_this();
return;
end
if ~exist('slow','var')
slow = false;
end
[dim,Frank] = size(oldF);
[Fcols,Fscal,Cfac] = unpack_SPLDA_adaptation_params(params... |
github | bsxfan/meta-embeddings-master | logdetLU.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/param_adaptation/logdetLU.m | 393 | utf_8 | f89f69d0255592e7c1933492e7894c79 | function [y,back] = logdetLU(M)
if nargin==0
test_this();
return;
end
[L,U] = lu(M);
y = sum(log(diag(U).^2))/2;
back = @back_this;
function dM = back_this(dy)
%dM = dy*(inv(U)/L).';
dM = dy*(L.'\inv(U.'));
end
end
function test_this()
... |
github | bsxfan/meta-embeddings-master | posteriorNorm2_slow.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/param_adaptation/posteriorNorm2_slow.m | 1,855 | utf_8 | c490aa24f95c0ecb840e47f99af01af6 | function [y,back] = posteriorNorm2_slow(A,B,b,priorFac)
% Computes, for every i: log N( 0 | Pi\A(:,i), inv(Pi) ), where
%
% precisions are Pi = I + b(i)*B
%
% This is the slow version, used only to verify correctness of the function
% value and derivatives of the fast version, posteriorNorm_fast().
%
% Inputs:
% ... |
github | bsxfan/meta-embeddings-master | adaptSPLDA.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/param_adaptation/adaptSPLDA.m | 1,791 | utf_8 | fce312e9e18adccb87760f2788ced832 | function [Ft,Wt,back] = adaptSPLDA(Fcols,Fscal,Cfac,F,W)
if nargin==0
test_this();
return;
end
Frank = length(Fscal);
Ft = [bsxfun(@times,F,Fscal), Fcols];
% Wt = inv(Ct), Ct = inv(W) + Cfac*Cfac'
% Wt = W - W*Cfac*inv(I + Cfac'*W*Cfac)*Cfac'*W
WCfac = W*C... |
github | bsxfan/meta-embeddings-master | matinv.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/param_adaptation/matinv.m | 345 | utf_8 | 140429773a1c0fb820d3ebe7e4d16619 | function [Y,back] = matinv(S)
if nargin==0
test_this();
return;
end
Y = inv(S);
back = @back_this;
function dS = back_this(dY)
dS = -Y.'*dY*Y.';
end
end
function test_this()
n = 4;
S = randn(n,n);
testBackprop(@matinv... |
github | bsxfan/meta-embeddings-master | posteriorNorm_mindiv_slow.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/param_adaptation/posteriorNorm_mindiv_slow.m | 1,999 | utf_8 | 3a06a111867c52058d73244d0ba36117 | function [y,Ezz,back] = posteriorNorm_mindiv_slow(A,B,b)
% Computes, for every i: log N( 0 | Pi\A(:,i), inv(Pi) ), where
%
% precisions are Pi = I + b(i)*B
%
% This is the slow version, used only to verify correctness of the function
% value and derivatives of the fast version, posteriorNorm_fast().
%
% Inputs:
% ... |
github | bsxfan/meta-embeddings-master | splda_llh_full.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/param_adaptation/splda_llh_full.m | 3,374 | utf_8 | 9f2a6f749c0dfeb7e8f3ae83f26c362e | function [llh,back] = splda_llh_full(labels,F,W,R,slow)
% Like splda_llh(), but backpropagates into all of F,W and R
%
% Inputs:
% R: D-by-N, i-vectors
% labels: sparse, logical K-by-N, one hot columns, K speakers, N recordings
% F,W: SPLA parameters
% slow: [optional, default = false] logical, use slow = true ... |
github | bsxfan/meta-embeddings-master | posteriorNorm_slow.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/param_adaptation/posteriorNorm_slow.m | 1,815 | utf_8 | a9b32fbd06cb0f91327d57dac2f41bbd | function [y,back] = posteriorNorm_slow(A,B,b)
% Computes, for every i: log N( 0 | Pi\A(:,i), inv(Pi) ), where
%
% precisions are Pi = I + b(i)*B
%
% This is the slow version, used only to verify correctness of the function
% value and derivatives of the fast version, posteriorNorm_fast().
%
% Inputs:
% A: dim-by-... |
github | bsxfan/meta-embeddings-master | posteriorNorm_fast.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/param_adaptation/posteriorNorm_fast.m | 2,131 | utf_8 | f950d84510798e37996a0f242fd5f9be | function [y,back] = posteriorNorm_fast(A,B,b)
% Computes, for every i: log N( 0 | Pi\A(:,i), inv(Pi) ), where
%
% precisions are Pi = I + b(i)*B
%
% This is the fast version, which simultaneously diagonalizes all the Pi,
% using eigenanalysis of B.
%
% Inputs:
% A: dim-by-n, natural parameters (precision *mean) f... |
github | bsxfan/meta-embeddings-master | rkhs_inner_prod.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/mmd/rkhs_inner_prod.m | 516 | utf_8 | 8df105a9a35bd9faf72c137fd3d1d6e8 | function y = rkhs_inner_prod(sigma,B1,a1,B2,a2)
dim = length(a1);
I = eye(dim);
Bconv = (I+sigma*B1)\B1; % inv(sigmaI + inv(B1))
aconv = (I+sigma*B1)\a1; % Bconv*(B1\a)
y = gauss_prod_int(aconv,Bconv,a2,B2);
end
function y = log_gauss_norm(B,a)
[logd,mu,back] = logdet_solveLU(B,a);
... |
github | bsxfan/meta-embeddings-master | log_gauss_norm.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/mmd/log_gauss_norm.m | 527 | utf_8 | 9dba44b2abd5c0513bb1b75566fab019 | function [y,back] = log_gauss_norm(B,a)
if nargin==0
test_this();
return;
end
[logd,mu,back1] = logdet_solveLU(B,a);
y = (logd -mu.'*a)/2;
back = @back_this;
function [dB,da] = back_this(dy)
dlogd = dy/2;
da = (-dy/2)*mu;
dmu = (-dy/2)*a;
... |
github | bsxfan/meta-embeddings-master | logdet_solveLU.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/mmd/logdet_solveLU.m | 508 | utf_8 | c3ec88a4a91de8ea84a175ef1857bfa8 | function [y,mu,back] = logdet_solveLU(B,a)
if nargin==0
test_this();
return;
end
[L,U] = lu(B);
y = sum(log(diag(U).^2))/2; %logdet
mu = U\(L\a); %solve
back = @back_this;
function [dB,da] = back_this(dy,dmu)
dB = dy*(L.'\inv(U.'));
da = L.'\... |
github | bsxfan/meta-embeddings-master | rkhs_proj_onto_I0_slow.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/mmd/rkhs_proj_onto_I0_slow.m | 1,766 | utf_8 | d012ab760719343ddc73bd3f4aa42563 | function [y,back] = rkhs_proj_onto_I0_slow(sigma,A,b,B)
% RKHS inner products of multivariate Gaussians onto standard normal. The
% RKHS kernel is K(x,y) = ND(x|y,sigma I). The first order natural
% parameters of the Gaussians are in the columns of A. The precisions are
% proportional to the fixed B, with scaling conta... |
github | bsxfan/meta-embeddings-master | L_BFGS.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/optimization/L_BFGS.m | 5,008 | utf_8 | 8a3bd29f4c568d13b7c0f7e466850904 | function [w,y,mem,logs] = L_BFGS(obj,w,maxiters,timeout,mem,stpsz0,callback)
% L-BFGS Quasi-Newton unconstrained optimizer.
% -- This has a small interface change from LBFGS.m --
%
% Inputs:
% obj: optimization objective, with interface: [y,grad] = obj(w),
% where w is the parameter vector, y is the scalar o... |
github | bsxfan/meta-embeddings-master | testBackprop_rs.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/test/testBackprop_rs.m | 2,198 | utf_8 | 9d88acaf06002d97bf8eb2fdc07bf7b8 | function total = testBackprop_rs(block,X,delta,mask)
%same as testFBblock, but with real step
if ~iscell(X)
X = {X};
end
dX = cellrndn(X);
if exist('mask','var')
assert(length(mask)==length(X));
dX = cellmask(dX,mask);
end
cX1 = cellstep(X,dX,delta);
cX2 = cell... |
github | bsxfan/meta-embeddings-master | testBackprop_multi.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/test/testBackprop_multi.m | 2,361 | utf_8 | ced4a5b7e925d5c00a2d991fb34d012c | function total = testBackprop_multi(block,nout,X,mask)
% same as testBackprop, but handles multiple outputs
if ~iscell(X)
X = {X};
end
dX = cellrndn(X);
if exist('mask','var')
assert(length(mask)==length(X));
dX = cellmask(dX,mask);
end
cX = cellcomplex(X,dX);
D... |
github | bsxfan/meta-embeddings-master | testBackprop.m | .m | meta-embeddings-master/code/Niko/matlab/SRE18/MLNDA_SRE18/test/testBackprop.m | 2,015 | utf_8 | fc73fa404f5441c097eb63f249106078 | function total = testBackprop(block,X,mask)
if ~iscell(X)
X = {X};
end
dX = cellrndn(X);
if exist('mask','var')
assert(length(mask)==length(X));
dX = cellmask(dX,mask);
end
cX = cellcomplex(X,dX);
DX = cell(size(X));
[Y,back] = block(X{:});
... |
github | gaoxifeng/Robust-Hexahedral-Re-Meshing-master | m2p.m | .m | Robust-Hexahedral-Re-Meshing-master/extern/libigl/python/matlab/m2p.m | 786 | utf_8 | 9c36f898544f3e8561f950ac4cc06626 | % Converts a Matlab matrix to a python-wrapped Eigen Matrix
function [ P ] = m2p( M )
if (isa(M, 'double'))
% Convert the matrix to a python 1D array
a = py.array.array('d',reshape(M,1,numel(M)));
% Then convert it to a eigen type
t = py.igl.eigen.MatrixXd(a.tolist());
% Fina... |
github | gaoxifeng/Robust-Hexahedral-Re-Meshing-master | p2m.m | .m | Robust-Hexahedral-Re-Meshing-master/extern/libigl/python/matlab/p2m.m | 585 | utf_8 | 15f32817630ba8b7a51a95ba7e4d5f2b | % Converts a python-wrapped Eigen Matrix to a Matlab matrix
function [ M ] = p2m( P )
if py.repr(py.type(P)) == '<class ''igl.eigen.MatrixXd''>'
% Convert it to a python array first
t = py.array.array('d',P);
% Reshape it
M = reshape(double(t),P.rows(),P.cols());
elseif py.repr(p... |
github | janhavelka/Stochastic-Finite-Element-Method-master | trisurfc.m | .m | Stochastic-Finite-Element-Method-master/functions/trisurfc.m | 14,500 | utf_8 | 850e8c97d2b3250b7af61ed01af5dd00 | function [cout,hout] = trisurfc(el,xin,yin,zin,N,shift,magnit)
% Contouring and surface function for functions defined on triangular meshes
%
% xin, yin, zin, are arrays of x,y,z values of the points for your surface.
% So [x(1) y(1) z(1)] defines the first point, etc.
%
% The last input N defines the contourin... |
github | janhavelka/Stochastic-Finite-Element-Method-master | position_figure.m | .m | Stochastic-Finite-Element-Method-master/functions/position_figure.m | 1,958 | utf_8 | f5af9ed0a926dc7aa2830616a6cd90aa | % Similar to subplot(), position_figure() divides the screen into rectangular
% panes to position figures in.
%
% Syntax:
% position_figure(no_rows, no_columns, fig_no)
%
% where
% 'no_rows' is the total number of figure-rows
% 'no_columns' is the total number of figure-... |
github | janhavelka/Stochastic-Finite-Element-Method-master | getStiffness.m | .m | Stochastic-Finite-Element-Method-master/functions/mesh_basics/getStiffness.m | 2,024 | utf_8 | 95c79401069289772f0168e518bb5393 |
function K=getStiffness(MESH,lambda)
% Jan Havelka (jnhavelka@gmail.com)
% Copyright 2016, Czech Technical University in Prague
%
% This program is free software: you can redistribute it and/or modify
% it under the terms of the GNU General Public License as published by
% the Free Software Foundation, ... |
github | janhavelka/Stochastic-Finite-Element-Method-master | nwspgr.m | .m | Stochastic-Finite-Element-Method-master/Stochastic-Collocation/functions/nwspgr.m | 55,365 | utf_8 | d201003f6a079cff4647bd2dda9defda | %*****************************************************************************
%nwSpGr: Nodes and weights for numerical integration on sparse grids (Smolyak)
%(c) 2007 Florian Heiss, Viktor Winschel
%Nov 11, 2007
%*****************************************************************************
%**********************... |
github | jlowenz/rrwm-master | makePointMatchingProblem.m | .m | rrwm-master/utils/makePointMatchingProblem.m | 1,904 | utf_8 | d8b1523c4e12502cda5a4a603d756604 | %% Make test problem
function [ problem ] = makePointMatchingProblem( Set )
%% Get values from structure
strNames = fieldnames(Set);
for i = 1:length(strNames), eval([strNames{i} '= Set.' strNames{i} ';']); end
%% Set number of nodes
if bOutBoth, nP1 = nInlier + nOutlier; else nP1 = nInlier; end
nP2 = nInlier + nOutlie... |
github | jlowenz/rrwm-master | wrapper_GM.m | .m | rrwm-master/utils/wrapper_GM.m | 1,050 | utf_8 | 1576ee070fa2b9ddbae721e3c9b98b1a | %% Makes current problem into Graph Matching form
function [accuracy score time X Xraw] = wrapper_GM(method, cdata)
% Make function evaluation script
str = ['feval(@' func2str(method.fhandle)];
for j = 1:length(method.variable), str = [str ',cdata.' method.variable{j} ]; end
if ~isempty(method.param), for i = 1:length(... |
github | jlowenz/rrwm-master | RRWM.m | .m | rrwm-master/Methods/RRWM/RRWM.m | 4,758 | utf_8 | ea71ece6a20675bf027025e76e265427 | %% Implementaion of PageRank matching algortihm
function [ X ] = RRWM( M, group1, group2, varargin)
% MATLAB demo code of Reweighted Random Walks Graph Matching of ECCV 2010
%
% Minsu Cho, Jungmin Lee, and Kyoung Mu Lee,
% Reweighted Random Walks for Graph Matching,
% Proc. European Conference on Computer Vision (ECC... |
github | qingsongma/blend-master | make.m | .m | blend-master/tools/libsvm-3.22/matlab/make.m | 888 | utf_8 | 4a2ad69e765736f8cca8e3b721fb7ebd | % This make.m is for MATLAB and OCTAVE under Windows, Mac, and Unix
function make()
try
% This part is for OCTAVE
if (exist ('OCTAVE_VERSION', 'builtin'))
mex libsvmread.c
mex libsvmwrite.c
mex -I.. svmtrain.c ../svm.cpp svm_model_matlab.c
mex -I.. svmpredict.c ../svm.cpp svm_model_matlab.c
% This part is fo... |
github | kertansul/caffe-segnet-cudnn5-master | classification_demo.m | .m | caffe-segnet-cudnn5-master/matlab/demo/classification_demo.m | 5,412 | utf_8 | 8f46deabe6cde287c4759f3bc8b7f819 | function [scores, maxlabel] = classification_demo(im, use_gpu)
% [scores, maxlabel] = classification_demo(im, use_gpu)
%
% Image classification demo using BVLC CaffeNet.
%
% IMPORTANT: before you run this demo, you should download BVLC CaffeNet
% from Model Zoo (http://caffe.berkeleyvision.org/model_zoo.html)
%
% *****... |
github | ncohn/Windsurf-master | Windsurf_Run_Batch.m | .m | Windsurf-master/Windsurf_Run_Batch.m | 261 | utf_8 | 55529be4c74bf42a0cc4951e09301793 | %Windsurf_Run_Batch.m - Code which allows to run multiple simulations at once via parfor loop
%Created By: N. Cohn, Oregon State University
function XBCDM_Run_Batch(Directory)
load([Directory, filesep, 'windsurf_setup.mat']);
Windsurf_Run;
end
|
github | ncohn/Windsurf-master | Windsurf_BeachNourishment.m | .m | Windsurf-master/Windsurf_BeachNourishment.m | 3,484 | utf_8 | 13189e5a9562fdfceaee5f6b8627766e | %Windsurf_BeachNourishment
%Code for modifying the beach profile based on volume (type 1), contour (type 2), or fixed time (type 3) changes at some
%fixed time intervals based on specified user inputs
%
%Created By: N. Cohn, Oregon State University
if project.flag.nourishment ~= 0 && run_number > 1
... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.