Matches in SemOpenAlex for { <https://semopenalex.org/work/W2897250287> ?p ?o ?g. }
Showing items 1 to 83 of
83
with 100 items per page.
- W2897250287 abstract "Building evaluation functions for chess variants is a challenging goal. At this time, only AlphaZero succeeded with millions of self-play records produced by using thousands of tensor processing units (TPUs), which are not available for most researchers. This paper presents the challenge of training evaluation functions on the basis of deep convolutional neural networks using decent data and computing resources, where regularization is crucial as complex models trained with limited data are more prone to overfitting. We present a novel training scheme by introducing a uniformity regularization (UR) network. In the proposed approach, a value network and a discriminator network share common convolutional layers and both networks are trained simultaneously. Loss functions for them are based on the difference between the score of a random move and that of an experts’ move as a comparison training method. The value network is expected to give precise scores for all positions, while the discriminator makes qualitative evaluations for move pairs, and acts as a regularizer that penalizes differences in evaluation results to ensure all samples are uniformly discriminated. Due to the existence of shared layers, such regularization is beneficial for improving the overall accuracy of the value network. Experimental results for chess and shogi demonstrate the proposed method surpassed the standard L 2 regularization method, and successfully helped obtain decently accurate value networks." @default.
- W2897250287 created "2018-10-26" @default.
- W2897250287 creator A5015877402 @default.
- W2897250287 creator A5073984860 @default.
- W2897250287 date "2018-08-01" @default.
- W2897250287 modified "2023-10-02" @default.
- W2897250287 title "Building Evaluation Functions for Chess and Shogi with Uniformity Regularization Networks" @default.
- W2897250287 cites W1497290640 @default.
- W2897250287 cites W1583460156 @default.
- W2897250287 cites W1750311484 @default.
- W2897250287 cites W1896424170 @default.
- W2897250287 cites W2095705004 @default.
- W2897250287 cites W2099001564 @default.
- W2897250287 cites W2144513243 @default.
- W2897250287 cites W2151368309 @default.
- W2897250287 cites W2156387975 @default.
- W2897250287 cites W2165999534 @default.
- W2897250287 cites W2194775991 @default.
- W2897250287 cites W2257979135 @default.
- W2897250287 cites W2590374252 @default.
- W2897250287 cites W2593414223 @default.
- W2897250287 cites W2626792426 @default.
- W2897250287 cites W2766447205 @default.
- W2897250287 cites W2772709170 @default.
- W2897250287 cites W2800677456 @default.
- W2897250287 cites W2911296969 @default.
- W2897250287 cites W3098637454 @default.
- W2897250287 doi "https://doi.org/10.1109/cig.2018.8490455" @default.
- W2897250287 hasPublicationYear "2018" @default.
- W2897250287 type Work @default.
- W2897250287 sameAs 2897250287 @default.
- W2897250287 citedByCount "3" @default.
- W2897250287 countsByYear W28972502872018 @default.
- W2897250287 crossrefType "proceedings-article" @default.
- W2897250287 hasAuthorship W2897250287A5015877402 @default.
- W2897250287 hasAuthorship W2897250287A5073984860 @default.
- W2897250287 hasConcept C119857082 @default.
- W2897250287 hasConcept C124101348 @default.
- W2897250287 hasConcept C144133560 @default.
- W2897250287 hasConcept C154945302 @default.
- W2897250287 hasConcept C162853370 @default.
- W2897250287 hasConcept C22019652 @default.
- W2897250287 hasConcept C2776135515 @default.
- W2897250287 hasConcept C2779803651 @default.
- W2897250287 hasConcept C41008148 @default.
- W2897250287 hasConcept C4216890 @default.
- W2897250287 hasConcept C50644808 @default.
- W2897250287 hasConcept C76155785 @default.
- W2897250287 hasConcept C81363708 @default.
- W2897250287 hasConcept C89249532 @default.
- W2897250287 hasConcept C94915269 @default.
- W2897250287 hasConceptScore W2897250287C119857082 @default.
- W2897250287 hasConceptScore W2897250287C124101348 @default.
- W2897250287 hasConceptScore W2897250287C144133560 @default.
- W2897250287 hasConceptScore W2897250287C154945302 @default.
- W2897250287 hasConceptScore W2897250287C162853370 @default.
- W2897250287 hasConceptScore W2897250287C22019652 @default.
- W2897250287 hasConceptScore W2897250287C2776135515 @default.
- W2897250287 hasConceptScore W2897250287C2779803651 @default.
- W2897250287 hasConceptScore W2897250287C41008148 @default.
- W2897250287 hasConceptScore W2897250287C4216890 @default.
- W2897250287 hasConceptScore W2897250287C50644808 @default.
- W2897250287 hasConceptScore W2897250287C76155785 @default.
- W2897250287 hasConceptScore W2897250287C81363708 @default.
- W2897250287 hasConceptScore W2897250287C89249532 @default.
- W2897250287 hasConceptScore W2897250287C94915269 @default.
- W2897250287 hasLocation W28972502871 @default.
- W2897250287 hasOpenAccess W2897250287 @default.
- W2897250287 hasPrimaryLocation W28972502871 @default.
- W2897250287 hasRelatedWork W2744043610 @default.
- W2897250287 hasRelatedWork W2897250287 @default.
- W2897250287 hasRelatedWork W2998075155 @default.
- W2897250287 hasRelatedWork W3116187740 @default.
- W2897250287 hasRelatedWork W3138048262 @default.
- W2897250287 hasRelatedWork W3162092403 @default.
- W2897250287 hasRelatedWork W3186919929 @default.
- W2897250287 hasRelatedWork W4214844979 @default.
- W2897250287 hasRelatedWork W4220755519 @default.
- W2897250287 hasRelatedWork W4246357244 @default.
- W2897250287 isParatext "false" @default.
- W2897250287 isRetracted "false" @default.
- W2897250287 magId "2897250287" @default.
- W2897250287 workType "article" @default.