Matches in SemOpenAlex for { <https://semopenalex.org/work/W2989650185> ?p ?o ?g. }
- W2989650185 abstract "To acquire a new skill, humans learn better and faster if a tutor, based on their current knowledge level, informs them of how much attention they should pay to particular content or practice problems. Similarly, a machine learning model could potentially be trained better with a scorer that adapts to its current learning state and estimates the importance of each training data instance. Training such an adaptive scorer efficiently is a challenging problem; in order to precisely quantify the effect of a data instance at a given time during the training, it is typically necessary to first complete the entire training process. To efficiently optimize data usage, we propose a reinforcement learning approach called Differentiable Data Selection (DDS). In DDS, we formulate a scorer network as a learnable function of the training data, which can be efficiently updated along with the main model being trained. Specifically, DDS updates the scorer with an intuitive reward signal: it should up-weigh the data that has a similar gradient with a dev set upon which we would finally like to perform well. Without significant computing overhead, DDS delivers strong and consistent improvements over several strong baselines on two very different tasks of machine translation and image classification." @default.
- W2989650185 created "2019-12-05" @default.
- W2989650185 creator A5001170954 @default.
- W2989650185 creator A5004315399 @default.
- W2989650185 creator A5013793053 @default.
- W2989650185 creator A5040523461 @default.
- W2989650185 creator A5062362044 @default.
- W2989650185 creator A5068811427 @default.
- W2989650185 date "2019-11-22" @default.
- W2989650185 modified "2023-09-27" @default.
- W2989650185 title "Optimizing Data Usage via Differentiable Rewards." @default.
- W2989650185 cites W1527783480 @default.
- W2989650185 cites W1836465849 @default.
- W2989650185 cites W1848260265 @default.
- W2989650185 cites W1905522558 @default.
- W2989650185 cites W2012897754 @default.
- W2989650185 cites W2034368206 @default.
- W2989650185 cites W2090411045 @default.
- W2989650185 cites W2095705004 @default.
- W2989650185 cites W2101105183 @default.
- W2989650185 cites W2111362445 @default.
- W2989650185 cites W2117278770 @default.
- W2989650185 cites W2117539524 @default.
- W2989650185 cites W2119717200 @default.
- W2989650185 cites W2124659975 @default.
- W2989650185 cites W2125653933 @default.
- W2989650185 cites W2132984949 @default.
- W2989650185 cites W2144600658 @default.
- W2989650185 cites W2156689847 @default.
- W2989650185 cites W2194775991 @default.
- W2989650185 cites W2251590347 @default.
- W2989650185 cites W2256388387 @default.
- W2989650185 cites W2296073425 @default.
- W2989650185 cites W2549139847 @default.
- W2989650185 cites W2584268338 @default.
- W2989650185 cites W2591789009 @default.
- W2989650185 cites W2785542264 @default.
- W2989650185 cites W2786471719 @default.
- W2989650185 cites W2804935296 @default.
- W2989650185 cites W2810075754 @default.
- W2989650185 cites W2887920589 @default.
- W2989650185 cites W2890095519 @default.
- W2989650185 cites W2898846200 @default.
- W2989650185 cites W2901026139 @default.
- W2989650185 cites W2902347140 @default.
- W2989650185 cites W2906424389 @default.
- W2989650185 cites W2919188216 @default.
- W2989650185 cites W2949973181 @default.
- W2989650185 cites W2951065878 @default.
- W2989650185 cites W2951775809 @default.
- W2989650185 cites W2952474700 @default.
- W2989650185 cites W2953199126 @default.
- W2989650185 cites W2963081269 @default.
- W2989650185 cites W2963088995 @default.
- W2989650185 cites W2963263347 @default.
- W2989650185 cites W2963366389 @default.
- W2989650185 cites W2963371670 @default.
- W2989650185 cites W2963403868 @default.
- W2989650185 cites W2963577254 @default.
- W2989650185 cites W2963696295 @default.
- W2989650185 cites W2963831310 @default.
- W2989650185 cites W2963962154 @default.
- W2989650185 cites W2964005754 @default.
- W2989650185 cites W2964022663 @default.
- W2989650185 cites W2964085268 @default.
- W2989650185 cites W2964121744 @default.
- W2989650185 cites W2964137095 @default.
- W2989650185 cites W2964308564 @default.
- W2989650185 cites W2964327384 @default.
- W2989650185 cites W3118608800 @default.
- W2989650185 cites W3137695714 @default.
- W2989650185 hasPublicationYear "2019" @default.
- W2989650185 type Work @default.
- W2989650185 sameAs 2989650185 @default.
- W2989650185 citedByCount "4" @default.
- W2989650185 countsByYear W29896501852019 @default.
- W2989650185 countsByYear W29896501852020 @default.
- W2989650185 crossrefType "posted-content" @default.
- W2989650185 hasAuthorship W2989650185A5001170954 @default.
- W2989650185 hasAuthorship W2989650185A5004315399 @default.
- W2989650185 hasAuthorship W2989650185A5013793053 @default.
- W2989650185 hasAuthorship W2989650185A5040523461 @default.
- W2989650185 hasAuthorship W2989650185A5062362044 @default.
- W2989650185 hasAuthorship W2989650185A5068811427 @default.
- W2989650185 hasConcept C111919701 @default.
- W2989650185 hasConcept C119857082 @default.
- W2989650185 hasConcept C134306372 @default.
- W2989650185 hasConcept C14036430 @default.
- W2989650185 hasConcept C154945302 @default.
- W2989650185 hasConcept C177264268 @default.
- W2989650185 hasConcept C199360897 @default.
- W2989650185 hasConcept C202615002 @default.
- W2989650185 hasConcept C2779960059 @default.
- W2989650185 hasConcept C33923547 @default.
- W2989650185 hasConcept C41008148 @default.
- W2989650185 hasConcept C51632099 @default.
- W2989650185 hasConcept C78458016 @default.
- W2989650185 hasConcept C86803240 @default.
- W2989650185 hasConcept C97541855 @default.
- W2989650185 hasConcept C98045186 @default.