text stringlengths 4 4.22k | acronym_word stringclasses 731
values | expansion stringlengths 6 51 | replaced_text stringlengths 20 4.26k |
|---|---|---|---|
In summary , it is evident that their complexities are in increasing order : leakage - based , Max - SR , and generalized EDAS . | SR | secrecy rate | In summary , it is evident that their complexities are in increasing order : leakage - based , Max - secrecy rate , and generalized EDAS . |
The main objective of DDE - MGM is computational efficiency from the aspects of both computing time and memory consumption , while preserving superior classification accuracy as compared to the state - of - the - art methods . | MGM | markov geographic model | The main objective of DDE - markov geographic model is computational efficiency from the aspects of both computing time and memory consumption , while preserving superior classification accuracy as compared to the state - of - the - art methods . |
Especially , there is the smaller descent of SPF - GCN classification accuracy with structure loss increasing from to , e.g. to on Cora , to on Citeseer and to on PubMed . | GCN | graph convolution networks | Especially , there is the smaller descent of SPF - graph convolution networks classification accuracy with structure loss increasing from to , e.g. to on Cora , to on Citeseer and to on PubMed . |
We end the talk inviting the community to join us to further explore this paradigm shift , and to push the concepts of AC to effectively realise multi - purpose , everyday software systems . | AC | autonomic computing | We end the talk inviting the community to join us to further explore this paradigm shift , and to push the concepts of autonomic computing to effectively realise multi - purpose , everyday software systems . |
We start with " AlexNet " as our base CNN and then examine the broad variations proposed over time to suit different applications . | CNN | convolutional neural network | We start with " AlexNet " as our base convolutional neural network and then examine the broad variations proposed over time to suit different applications . |
ConclusionWe find that the FM - Tree is a more durable , faster variant of the B - tree with properties that make it intrinsically better for operating on flash memory . | FM | flash memory | ConclusionWe find that the flash memory - Tree is a more durable , faster variant of the B - tree with properties that make it intrinsically better for operating on flash memory . |
The database is constructed from 915 clinical CT volumes consisting of head and neck images . | CT | computed tomography | The database is constructed from 915 clinical computed tomography volumes consisting of head and neck images . |
The BC distribution follow power law and BC is related to degree as where , is BC exponent . | BC | betweenness centrality | The betweenness centrality distribution follow power law and betweenness centrality is related to degree as where , is betweenness centrality exponent . |
Objectives and ContributionsThe main goals of this thesis are to define a method which enables the characterization of the motion intensity in arbitrary video sequences and also to design , implement , and assess several FEC - based mechanisms with content - awareness to enhance the quality of video delivery from the p... | FEC | forward error correction | Objectives and ContributionsThe main goals of this thesis are to define a method which enables the characterization of the motion intensity in arbitrary video sequences and also to design , implement , and assess several forward error correction - based mechanisms with content - awareness to enhance the quality of vide... |
To our knowledge , the gain design problem for OCC has not been completely convexified in the way we have done in this paper . | OCC | output constrained covariance | To our knowledge , the gain design problem for output constrained covariance has not been completely convexified in the way we have done in this paper . |
The accuracy of the prediction itself however , ranges from for linear regression to for CNN . | CNN | convolutional neural network | The accuracy of the prediction itself however , ranges from for linear regression to for convolutional neural network . |
The Squared Exponential ( SE ) kernel , described as in the table , is known to capture the smoothness property of the objective functions . | SE | squared exponential | The Squared Exponential ( squared exponential ) kernel , described as in the table , is known to capture the smoothness property of the objective functions . |
Having defined the game , we use numerical simulations to derive the existence of equilibrium points , which then identify the optimum behavior for both the FC and the Byzantines in a game - theoretic sense . | FC | fusion center | Having defined the game , we use numerical simulations to derive the existence of equilibrium points , which then identify the optimum behavior for both the fusion center and the Byzantines in a game - theoretic sense . |
Then there exist a player MD strategy and a player MD strategy such that for all states , if or , then the following is true : equation*aligned _ : & _ , s , , ( ) s or _ : & _ , s , , ( ) < s.alignedequation*lemmaproofWe construct a transfinite sequence of subgames , where is an ordinal number , by stepwise removingce... | MD | memoryless deterministic | Then there exist a player memoryless deterministic strategy and a player memoryless deterministic strategy such that for all states , if or , then the following is true : equation*aligned _ : & _ , s , , ( ) s or _ : & _ , s , , ( ) < s.alignedequation*lemmaproofWe construct a transfinite sequence of subgames , where i... |
The AV and RV strategy requires information of at least 75 of nodes . | RV | random vaccination | The AV and random vaccination strategy requires information of at least 75 of nodes . |
Recall from Remarks and that in stratifiable syntax , as used in Quine 's NF , variables do not have predefined levels but we insist on a stratifiability condition that and are only legal if we could assign levels to their variables to stratify them . | NF | new foundations | Recall from Remarks and that in stratifiable syntax , as used in Quine 's new foundations , variables do not have predefined levels but we insist on a stratifiability condition that and are only legal if we could assign levels to their variables to stratify them . |
In order to implement the ARA solution to this problem , the defender must use backward induction . | ARA | adversarial risk analysis | In order to implement the adversarial risk analysis solution to this problem , the defender must use backward induction . |
1 ] point out , the algorithm leads the total quantities and the market price to the values corresponding to the NE for these measures . | NE | nash equilibrium | 1 ] point out , the algorithm leads the total quantities and the market price to the values corresponding to the nash equilibrium for these measures . |
Since the ML implementation is prohibitively complex , we propose a reduced - complexity MSD detection scheme that is able to achieve the optimal ML performance . | ML | maximum likelihood | Since the maximum likelihood implementation is prohibitively complex , we propose a reduced - complexity MSD detection scheme that is able to achieve the optimal maximum likelihood performance . |
Our Lagrange multiplier method ( LM ) is compared against finite differences ( FD ) . | LM | lagrange multiplier method | Our Lagrange multiplier method ( lagrange multiplier method ) is compared against finite differences ( FD ) . |
Given a quad - tuple , by supposing , the square of MMD can be rewritten aswhereDeep Domain Adaptation Based on Transitive Transfer with Multi - SourceIn our method , we will choose a variety of source tasks with diverse similarity to the target task to assist recognizing some new types of SAR targets by transitive tra... | SAR | synthetic aperture radar | Given a quad - tuple , by supposing , the square of MMD can be rewritten aswhereDeep Domain Adaptation Based on Transitive Transfer with Multi - SourceIn our method , we will choose a variety of source tasks with diverse similarity to the target task to assist recognizing some new types of synthetic aperture radar targ... |
Specifically , AT minimizes the loss function : where the inner maximization , , is usually performed with an iterative gradient - based optimization . | AT | adversarial training | Specifically , adversarial training minimizes the loss function : where the inner maximization , , is usually performed with an iterative gradient - based optimization . |
In practice , over long deployment durations the height difference between sensor nodes would result in negligible SPL variations between the sensors ranging 15 - 35 feet in height , as the most significant difference in absolute SPL measurement between different sensors occurs when a noise source is directly below the... | SPL | sound pressure level | In practice , over long deployment durations the height difference between sensor nodes would result in negligible sound pressure level variations between the sensors ranging 15 - 35 feet in height , as the most significant difference in absolute sound pressure level measurement between different sensors occurs when a ... |
The PSIoT - SDN proof of concept components are : A network topology created with the Mininet emulator ; The POX SDN network operating system supporting the IoT - SDN network management ; Open vSwitch switches , controlled by the OpenFlow protocol ; User traffic generators ; The MAM - like BAM module implemented with O... | BAM | bandwidth allocation model | The PSIoT - SDN proof of concept components are : A network topology created with the Mininet emulator ; The POX SDN network operating system supporting the IoT - SDN network management ; Open vSwitch switches , controlled by the OpenFlow protocol ; User traffic generators ; The MAM - like bandwidth allocation model mo... |
We show that the best feature representation produced by ResNet outperforms several baselines for image privacy prediction that consider CNN - based models and SVM models trained on traditional visual features such as SIFT and global GIST descriptor . | SVM | support vector machine | We show that the best feature representation produced by ResNet outperforms several baselines for image privacy prediction that consider CNN - based models and support vector machine models trained on traditional visual features such as SIFT and global GIST descriptor . |
We showed that it is feasible to run an I / O bound analysis task on HPC resources with a Lustre parallel file system and achieve good scaling behavior up to 384 CPU cores with an almost 300-fold speed - up compared to serial execution . | HPC | high performance computing | We showed that it is feasible to run an I / O bound analysis task on high performance computing resources with a Lustre parallel file system and achieve good scaling behavior up to 384 CPU cores with an almost 300-fold speed - up compared to serial execution . |
Frame structure of standard RTSIn the uplink , all frame modifications are limited to the AP side to reduce STAs ' computing consumption . | AP | access point | Frame structure of standard RTSIn the uplink , all frame modifications are limited to the access point side to reduce STAs ' computing consumption . |
corollary[[Proposition 8.1]bcffm ] If GI is then the polynomial hierarchy collapses to the second level , that is , . | GI | graph isomorphism | corollary[[Proposition 8.1]bcffm ] If graph isomorphism is then the polynomial hierarchy collapses to the second level , that is , . |
SID performance with Text - dependent databaseSpeaker - identification under quite conditionsSpeaker identification performance of CI users were predicted for text dependent database based on CI auditory stimuli derived from electrodograms . | CI | cochlear implant | SID performance with Text - dependent databaseSpeaker - identification under quite conditionsSpeaker identification performance of cochlear implant users were predicted for text dependent database based on cochlear implant auditory stimuli derived from electrodograms . |
However , the DL and UL time fractions were set as fixed in , even though the DL and UL time fractions can be optimized to further improve the system performance . | DL | downlink | However , the downlink and UL time fractions were set as fixed in , even though the downlink and UL time fractions can be optimized to further improve the system performance . |
table*[t]Accuracy on Target SP Stringsets Early Stoppingtab : resultsSPES4.5pttabularcccccccccc2c2Training & 2Test & 3cLSTM & 3cs - RNN & 2RPNI & & & 10 & 30 & 100 & 10 & 30 & 100 & 6SP2 & 21k & 1 & 0.871 ( 0.04 ) & 0.954 ( 0.05 ) & 0.992 ( 0.00 ) & 0.910 ( 0.05 ) & 0.994 ( 0.01 ) & 0.992 ( 0.01 ) & 1.000 & & 2 & 0.960... | SP | strictly piecewise | table*[t]Accuracy on Target strictly piecewise Stringsets Early Stoppingtab : resultsstrictly piecewiseES4.5pttabularcccccccccc2c2Training & 2Test & 3cLSTM & 3cs - RNN & 2RPNI & & & 10 & 30 & 100 & 10 & 30 & 100 & 6strictly piecewise2 & 21k & 1 & 0.871 ( 0.04 ) & 0.954 ( 0.05 ) & 0.992 ( 0.00 ) & 0.910 ( 0.05 ) & 0.994... |
These properties make an AIDA sentence highly reusable , and we can treat it as an anchor to formally link , for example , papers that claim or refute it . | AIDA | atomic , independent , declarative , and absolute | These properties make an atomic , independent , declarative , and absolute sentence highly reusable , and we can treat it as an anchor to formally link , for example , papers that claim or refute it . |
The words occurring between the SDP only takes part in the training instead of the whole words present in the sentences to generate SDP embedding . | SDP | shortest dependency path | The words occurring between the shortest dependency path only takes part in the training instead of the whole words present in the sentences to generate shortest dependency path embedding . |
At the second experiment , we compare the proposed highly efficient MSC - trackers with the state - of - the - art real - time trackers , which shows the superiority of our MSC - trackers . | MSC | multi - layer same - resolution compressed | At the second experiment , we compare the proposed highly efficient multi - layer same - resolution compressed - trackers with the state - of - the - art real - time trackers , which shows the superiority of our multi - layer same - resolution compressed - trackers . |
In addition , we observe that , in general , when the circuitry power increases , the average number of PNs per CN reduces for both the MC - based and the SV - based games . | MC | marginal contribution | In addition , we observe that , in general , when the circuitry power increases , the average number of PNs per CN reduces for both the marginal contribution - based and the SV - based games . |
Droplet was subjected to continuous DC of 1.5 V. ( d ) Square setup without DC , ( e ) Square setup with DC , ( f ) Round setup without DC , ( g ) Round setup with DC . | DC | disconnected components | Droplet was subjected to continuous disconnected components of 1.5 V. ( d ) Square setup without disconnected components , ( e ) Square setup with disconnected components , ( f ) Round setup without disconnected components , ( g ) Round setup with disconnected components . |
Differential privacy ( DP ) is a framework that guarantees strict bounds for the amount of leaked private information , even in the presence of arbitrary side information . | DP | differential privacy | Differential privacy ( differential privacy ) is a framework that guarantees strict bounds for the amount of leaked private information , even in the presence of arbitrary side information . |
2015 In January , Microsoft announces the Hololens , a headset to fuse AR and VR to be made available later in 2015 . | VR | virtual reality | 2015 In January , Microsoft announces the Hololens , a headset to fuse AR and virtual reality to be made available later in 2015 . |
By exploiting DA and advanced classification methods , it is capable of identifying a variety of network intrusions , especially emerging ones . | DA | data augmentation | By exploiting data augmentation and advanced classification methods , it is capable of identifying a variety of network intrusions , especially emerging ones . |
Approximately and improvements are introduced for SAD and RMSE metrics respectively . | SAD | spectral angle distance | Approximately and improvements are introduced for spectral angle distance and RMSE metrics respectively . |
The context consists of three parts : 1 ) the output of the SP ( i.e. the cluster representation ) , 2 ) the next cluster probabilities from the TP , and 3 ) the expected value of any rewards that the architecture will receive if in the next step , the input falls into a particular cluster ( interpreted as goals)(In f ... | TP | temporal pooler | The context consists of three parts : 1 ) the output of the SP ( i.e. the cluster representation ) , 2 ) the next cluster probabilities from the temporal pooler , and 3 ) the expected value of any rewards that the architecture will receive if in the next step , the input falls into a particular cluster ( interpreted as... |
recently proposed relevance - based word embedding models for learning word representations based on the objectives that matter for IR applications . | IR | information retrieval | recently proposed relevance - based word embedding models for learning word representations based on the objectives that matter for information retrieval applications . |
Methods to handle manifold uncertainties The hybrid ET and EAA approachET is a useful and powerful approach to combine various types of uncertain information from different sources . | ET | evidence theory | Methods to handle manifold uncertainties The hybrid evidence theory and EAA approachevidence theory is a useful and powerful approach to combine various types of uncertain information from different sources . |
Further , the RDF snippet in Listing shows a semantic definition of a particular study arm as an instance of the sco : InterventionArm . | RDF | resource description framework | Further , the resource description framework snippet in Listing shows a semantic definition of a particular study arm as an instance of the sco : InterventionArm . |
Model Selection by : The authors proposed a deep multi - modal architecture for product classification in e - commerce , wherein they learn a decision - level fusion policy to choose between image and text CNN for an input product . | CNN | convolutional neural network | Model Selection by : The authors proposed a deep multi - modal architecture for product classification in e - commerce , wherein they learn a decision - level fusion policy to choose between image and text convolutional neural network for an input product . |
Consider a finite directed acyclic graph , consisting of AFC nodes belonging to set , a set of sources ( MTC devices ) , and a set of destinations , such that . | AFC | atomic function computation | Consider a finite directed acyclic graph , consisting of atomic function computation nodes belonging to set , a set of sources ( MTC devices ) , and a set of destinations , such that . |
An incoming individual is converted to a POS skeleton , and the algorithm checks if the skeleton is contained in the skeleton set . | POS | part of speech | An incoming individual is converted to a part of speech skeleton , and the algorithm checks if the skeleton is contained in the skeleton set . |
P. Mogensen , W. Na , I. Kovacs , F. Frederiksen , A. Pokhariyal , K. Pedersen , T. Kolding , K. Hugl , and M. Kuusela , " LTE capacity compared to the shannon bound , " in Proc . | LTE | long term evolution | P. Mogensen , W. Na , I. Kovacs , F. Frederiksen , A. Pokhariyal , K. Pedersen , T. Kolding , K. Hugl , and M. Kuusela , " long term evolution capacity compared to the shannon bound , " in Proc . |
Optimization problemWe consider utility functions that are a weighted sum of three aspects : the charging time , which includes the travel time to the CS , the possible time spent queuing and the effective time for charging ; the charging price , where we use the amount of energy generated from renewable sources as a p... | CS | charging station | Optimization problemWe consider utility functions that are a weighted sum of three aspects : the charging time , which includes the travel time to the charging station , the possible time spent queuing and the effective time for charging ; the charging price , where we use the amount of energy generated from renewable ... |
For example , if the predictions correspond to walking at a slower pace , the joint angles will be misaligned ( frequency - shift ) and MSE computed will diverge over time . | MSE | mean squared error | For example , if the predictions correspond to walking at a slower pace , the joint angles will be misaligned ( frequency - shift ) and mean squared error computed will diverge over time . |
Prob - PIT defines a log - likelihood function based on the prior distributions and the separation errors of all possible permutations . | PIT | permutation invariant training | Prob - permutation invariant training defines a log - likelihood function based on the prior distributions and the separation errors of all possible permutations . |
In summary , the objective results confirm the effectiveness of the proposed MGE criterion in improving DNN accuracy , and that MGE is complementary to the stacking of bottleneck features . | MGE | minimum generation error | In summary , the objective results confirm the effectiveness of the proposed minimum generation error criterion in improving DNN accuracy , and that minimum generation error is complementary to the stacking of bottleneck features . |
These operators are then reused to approximately perform more ALS sweeps until the changes to the factor matrices are deemed large , at which point , the pairwise operators are recomputed . | ALS | alternating least squares | These operators are then reused to approximately perform more alternating least squares sweeps until the changes to the factor matrices are deemed large , at which point , the pairwise operators are recomputed . |
Even though both of these articles and many others demonstrate the incredible potential of using RL , these models also have their fair share of weaknesses . | RL | reinforcement learning | Even though both of these articles and many others demonstrate the incredible potential of using reinforcement learning , these models also have their fair share of weaknesses . |
Recent methods , however , have transitioned towards using advanced ML techniques such as Sparse Coding and Deep Learning as a proxy to sophisticated cognitive models . | ML | machine learning | Recent methods , however , have transitioned towards using advanced machine learning techniques such as Sparse Coding and Deep Learning as a proxy to sophisticated cognitive models . |
Using this information , the AP can identify the worst channel 's condition and then adjust the transmission rate and FEC . | AP | access point | Using this information , the access point can identify the worst channel 's condition and then adjust the transmission rate and FEC . |
We might use deeper networks to gain a few points in AUC and CAP curves . | CAP | cumulative accuracy profit | We might use deeper networks to gain a few points in AUC and cumulative accuracy profit curves . |
To model the spatial dependency of joints , we cast the graph structure into a sequence of joints and exactly develop a relevant RNN architecture . | RNN | recurrent neural network | To model the spatial dependency of joints , we cast the graph structure into a sequence of joints and exactly develop a relevant recurrent neural network architecture . |
The RWA model achieves a classification accuracy near before the LSTM model . | RWA | recurrent weighted average | The recurrent weighted average model achieves a classification accuracy near before the LSTM model . |
When 60 s worth of SPL data has been generated , it is written to a dedicated data partition on the SD card ready for data upload . | SD | secure digital | When 60 s worth of SPL data has been generated , it is written to a dedicated data partition on the secure digital card ready for data upload . |
Suppose the AP first wins the channel contention and sends a MU - RTS . | RTS | request to send | Suppose the AP first wins the channel contention and sends a MU - request to send . |
The intuition behind this design is that if where is indistinguishable from where , the shared encoder will learn to extract features and structures across different language domains , and thus make the following layers of QA models easier to learn language - independent general knowledge . | QA | question answering | The intuition behind this design is that if where is indistinguishable from where , the shared encoder will learn to extract features and structures across different language domains , and thus make the following layers of question answering models easier to learn language - independent general knowledge . |
The table highlights that our LIM embeddings contain information on the speaker identity , leading to a CER ( ) ranging from 2.84 to 1.21 in all the considered settings . | CER | classification error rate | The table highlights that our LIM embeddings contain information on the speaker identity , leading to a classification error rate ( ) ranging from 2.84 to 1.21 in all the considered settings . |
The images and vectors are used as training data for a CNN that , afterward , should be able to predict a vector for an unseen image . | CNN | convolutional neural network | The images and vectors are used as training data for a convolutional neural network that , afterward , should be able to predict a vector for an unseen image . |
where TP , FP , FN , TN represent true positive , false positive , false negative and true negative , respectively . | TN | true negative | where TP , FP , FN , true negative represent true positive , false positive , false negative and true negative , respectively . |
The only exception is the LIVE1 dataset at QF = 10 and 20 for which LERaG is slightly higher than our proposed method . | QF | quality factor | The only exception is the LIVE1 dataset at quality factor = 10 and 20 for which LERaG is slightly higher than our proposed method . |
Since the CNN feature is trained on labeled data , semantic cues can be preserved , and our method ( the third row ) is able to return challenging candidates which are semantically related to the query . | CNN | convolutional neural network | Since the convolutional neural network feature is trained on labeled data , semantic cues can be preserved , and our method ( the third row ) is able to return challenging candidates which are semantically related to the query . |
For MPI - based parallel HDF5 , both the compute and I / O time on Bridges were consistently larger than their corresponding values on SDSC Comet and LSU SuperMIC . | MPI | multiple parallel instances | For multiple parallel instances - based parallel HDF5 , both the compute and I / O time on Bridges were consistently larger than their corresponding values on SDSC Comet and LSU SuperMIC . |
For easy comparison , we present the related PMF models with a unified framework . | PMF | probabilistic matrix factorization | For easy comparison , we present the related probabilistic matrix factorization models with a unified framework . |
The same process was applied using the different AE models we introduced earlier . | AE | autoencoder | The same process was applied using the different autoencoder models we introduced earlier . |
In previous work , we also presented two studies on the manual and automatic creation , respectively , of AIDA sentences in the biomedical field . | AIDA | atomic , independent , declarative , and absolute | In previous work , we also presented two studies on the manual and automatic creation , respectively , of atomic , independent , declarative , and absolute sentences in the biomedical field . |
The energy predictor is also built atop GP regression , with energy consumption values directly measured from the hardware platform . | GP | gaussian process | The energy predictor is also built atop gaussian process regression , with energy consumption values directly measured from the hardware platform . |
To mitigate pilot contamination caused by PR , we develop a pilot scheduling algorithm under the criterion of minimizing the sum MSE of channel estimation of D2D links . | PR | pilot reuse | To mitigate pilot contamination caused by pilot reuse , we develop a pilot scheduling algorithm under the criterion of minimizing the sum MSE of channel estimation of D2D links . |
For the successful iterates we do not further increase the length of our moves by limiting as follows : Similarly , for the unsuccessful iterations we halve the distance by a factor of by setting as it is shown next : A short description of our algorithm as a GPS method for solving the problem stated in Eq . | GPS | general pattern search | For the successful iterates we do not further increase the length of our moves by limiting as follows : Similarly , for the unsuccessful iterations we halve the distance by a factor of by setting as it is shown next : A short description of our algorithm as a general pattern search method for solving the problem stated... |
This is because , in the NNNF strategy , the distance of the nearest user to the BS , i.e. , , approaches to zero and hence the term in is small which makes the difference of the bounds and the exact values insignificant . | BS | base station | This is because , in the NNNF strategy , the distance of the nearest user to the base station , i.e. , , approaches to zero and hence the term in is small which makes the difference of the bounds and the exact values insignificant . |
Only those users , whose second best link is better than at least existing links for that node , where is a configurable number , are selected for DC ( step 14 ) . | DC | dual connectivity | Only those users , whose second best link is better than at least existing links for that node , where is a configurable number , are selected for dual connectivity ( step 14 ) . |
We thus design a CNN encoder - decoder model based on AST structures . | CNN | convolutional neural network | We thus design a convolutional neural network encoder - decoder model based on AST structures . |
So the EHS - CNOMA with MRC provides linearly increased ESC for . | EHS | enhanced hybrid simultaneous | So the enhanced hybrid simultaneous - CNOMA with MRC provides linearly increased ESC for . |
t ] Predictive attack probabilities computed by the defender using ARA We now illustrate how could be obtained by modeling the attacker 's strategic analysis process using Eqs . ( ) | ARA | adversarial risk analysis | t ] Predictive attack probabilities computed by the defender using adversarial risk analysis We now illustrate how could be obtained by modeling the attacker 's strategic analysis process using Eqs . ( ) |
tabularp0.8cmp5.5cmp6 cm 3cAmyloid Load ( PiB Positivity ) Set 1 & PiB Angular L / R & PiB Cingulum Ant L / R & PiB Cingulum Post L / R & PiB Frontal Med Orb L / R & PiB Precuneus L / R & PiB Temporal Sup L / R & PiB Temporal Mid L / R & PiB SupraMarginal L Set 2 & FA Cerebral peduncle R & FA Cerebral peduncle L & MD C... | MD | mean diffusivity | tabularp0.8cmp5.5cmp6 cm 3cAmyloid Load ( PiB Positivity ) Set 1 & PiB Angular L / R & PiB Cingulum Ant L / R & PiB Cingulum Post L / R & PiB Frontal Med Orb L / R & PiB Precuneus L / R & PiB Temporal Sup L / R & PiB Temporal Mid L / R & PiB SupraMarginal L Set 2 & FA Cerebral peduncle R & FA Cerebral peduncle L & mean... |
The best performing algorithms are aggregated using a probability vote in order to create a model which has the largest area under ROC curve of all the developed models . | ROC | receiver operating characteristic | The best performing algorithms are aggregated using a probability vote in order to create a model which has the largest area under receiver operating characteristic curve of all the developed models . |
theoremthm : expectedCostTA - SKYGiven a subspace skyline query , the expected number of sorted accesses performed by TA - SKY on an tuple boolean relation with probability of having value on attribute being is , alignm^_i=1^n iP_stop(i)alignwhere is computed using Equations eq : stopi-1 , eq : stopi-2 , and eq : stopi... | TA | threshold algorithm | theoremthm : expectedCostthreshold algorithm - SKYGiven a subspace skyline query , the expected number of sorted accesses performed by threshold algorithm - SKY on an tuple boolean relation with probability of having value on attribute being is , alignm^_i=1^n iP_stop(i)alignwhere is computed using Equations eq : stopi... |
Given a test set of creatives , suppose represent the images with the lowest CTR values and represent the images with the highest CTR . | CTR | click through rates | Given a test set of creatives , suppose represent the images with the lowest click through rates values and represent the images with the highest click through rates . |
As stated by , the ( ) meets the two conditions of Exact - MBR encoding matrix . | MBR | minimum bandwidth regenerating | As stated by , the ( ) meets the two conditions of Exact - minimum bandwidth regenerating encoding matrix . |
Second , Specify VM Parameters as image size , VM memory ( MB ) , number of CPUs and VMM name . | VM | virtual machine | Second , Specify virtual machine Parameters as image size , virtual machine memory ( MB ) , number of CPUs and virtual machineM name . |
In this study , we want to propose a method for TCP in order to find the bugs made by the development team as soon as possible . | TCP | test case prioritization | In this study , we want to propose a method for test case prioritization in order to find the bugs made by the development team as soon as possible . |
The proposed ACNN - Seg method is compared against : the current state - of - the - art cine MR 2D slice by slice segmentation method ( 2D - FCN ) , 3D - UNet model , cascaded 3D - UNet and convolutional AE model ( AE - Seg ) , sub - pixel 3D - CNN segmentation model ( 3D - Seg ) proposed in Sec . , | MR | magnetic resonance | The proposed ACNN - Seg method is compared against : the current state - of - the - art cine magnetic resonance 2D slice by slice segmentation method ( 2D - FCN ) , 3D - UNet model , cascaded 3D - UNet and convolutional AE model ( AE - Seg ) , sub - pixel 3D - CNN segmentation model ( 3D - Seg ) proposed in Sec . , |
2 ) The performances of BOA and WCA are better than GSP ( with relative improvements of 8.9 and 2.2 respectively ) , which indicates that both our proposed approach and the game - theoretic approach for revenue optimization are effective . | GSP | generalized second price | 2 ) The performances of BOA and WCA are better than generalized second price ( with relative improvements of 8.9 and 2.2 respectively ) , which indicates that both our proposed approach and the game - theoretic approach for revenue optimization are effective . |
The database is constructed from 915 clinical CT volumes consisting of head and neck images . | CT | computed tomography | The database is constructed from 915 clinical computed tomography volumes consisting of head and neck images . |
Using the session key and IV from OpenSSL server logs , a search of malware client application memory yielded interesting facts : the key occurs frequently in different extract files , memory extract files sizes containing the key are within specific ranges , and two unusual ASCII strings are present near encryption ke... | IV | initialization vector | Using the session key and initialization vector from OpenSSL server logs , a search of malware client application memory yielded interesting facts : the key occurs frequently in different extract files , memory extract files sizes containing the key are within specific ranges , and two unusual ASCII strings are present... |
This will give NPR in MDFT FB . | NPR | near perfect reconstruction | This will give near perfect reconstruction in MDFT FB . |
Lastly , adopting different gold standards can affect ML - DSS significantly . | ML | machine learning | Lastly , adopting different gold standards can affect machine learning - DSS significantly . |
By reducing the depth of the path , the ADAG and SE require comparisons of binary classifiers . | SE | strong elimination | By reducing the depth of the path , the ADAG and strong elimination require comparisons of binary classifiers . |
Lastly , the model is formulated as a full pipeline based on a NN method that can be learned by backpropagation and a stochastic gradient solver . | NN | neural network | Lastly , the model is formulated as a full pipeline based on a neural network method that can be learned by backpropagation and a stochastic gradient solver . |
Frontier - Based Belief Propagation on the GPU We present all algorithms examined as realizations of a frontier - based BP framework . | BP | belief propagation | Frontier - Based Belief Propagation on the GPU We present all algorithms examined as realizations of a frontier - based belief propagation framework . |
Different from the temporal RNN , spatial RNN could recognize actions by a glimpse of one frame ( when the size of temporal window equals 1 ) . | RNN | recurrent neural network | Different from the temporal recurrent neural network , spatial recurrent neural network could recognize actions by a glimpse of one frame ( when the size of temporal window equals 1 ) . |
The CSG complexity measureAs mentioned before , a dataset with a low eigenvalue spectrum indicates a low inter - class overlap and thus easily separable classes . | CSG | cumulative spectral gradient | The cumulative spectral gradient complexity measureAs mentioned before , a dataset with a low eigenvalue spectrum indicates a low inter - class overlap and thus easily separable classes . |
sec : zkstatement , and presents an RPC interface to receive and answer requests from Proof Consumers ( ) . | RPC | remote procedure calls | sec : zkstatement , and presents an remote procedure calls interface to receive and answer requests from Proof Consumers ( ) . |
Proof of Proposition Claim : Under Assumption in the section , the MSE of SGLD with a decreasing step size sequence as in Theorem is bounded , for a constant independent of and a constant depending on and , as * where First , we adopt the MSE formula for the decreasing - step - size SG - MCMC with Euler integrator ( 1-... | MSE | mean squared error | Proof of Proposition Claim : Under Assumption in the section , the mean squared error of SGLD with a decreasing step size sequence as in Theorem is bounded , for a constant independent of and a constant depending on and , as * where First , we adopt the mean squared error formula for the decreasing - step - size SG - M... |
Source CodeThe source code for the CA tree implementations and the benchmarks can be found online ( https://www.it.uu.se/research/group/languages/software/im_tr_ca ) . | CA | contention adaptions | Source CodeThe source code for the contention adaptions tree implementations and the benchmarks can be found online ( https://www.it.uu.se/research/group/languages/software/im_tr_ca ) . |
End of preview. Expand in Data Studio
README.md exists but content is empty.
- Downloads last month
- 5