FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2974889342> ?p ?o ?g. }

• W2974889342 abstract "Abstract Natural brains perform miraculously well in learning new tasks from a small number of samples, whereas sample efficient learning is still a major open problem in the field of machine learning. Here, we raise the question, how the neural coding scheme affects sample efficiency, and make first progress on this question by proposing and analyzing a learning algorithm that uses a simple reinforce-type plasticity mechanism and does not require any gradients to learn low dimensional mappings. It harnesses three bio-plausible mechanisms, namely, population codes with bell shaped tuning curves, continous attractor mechanisms and probabilistic synapses, to achieve sample efficient learning. We show both theoretically and by simulations that population codes with broadly tuned neurons lead to high sample efficiency, whereas codes with sharply tuned neurons account for high final precision. Moreover, a dynamic adaptation of the tuning width during learning gives rise to both, high sample efficiency and high final precision. We prove a sample efficiency guarantee for our algorithm that lies within a logarithmic factor from the information theoretical optimum. Our simulations show that for low dimensional mappings, our learning algorithm achieves comparable sample efficiency to multi-layer perceptrons trained by gradient descent, although it does not use any gradients. Furthermore, it achieves competitive sample efficiency in low dimensional reinforcement learning tasks. From a machine learning perspective, these findings may inspire novel approaches to improve sample efficiency. From a neuroscience perspective, these findings suggest sample efficiency as a yet unstudied functional role of adaptive tuning curve width." @default.
• W2974889342 created "2019-09-26" @default.
• W2974889342 creator A5012297652 @default.
• W2974889342 creator A5046495847 @default.
• W2974889342 creator A5062687538 @default.
• W2974889342 date "2019-09-19" @default.
• W2974889342 modified "2023-10-17" @default.
• W2974889342 title "Adaptive Tuning Curve Widths Improve Sample Efficient Learning" @default.
• W2974889342 cites W1503270516 @default.
• W2974889342 cites W1542791059 @default.
• W2974889342 cites W1596533183 @default.
• W2974889342 cites W1603077448 @default.
• W2974889342 cites W1621791442 @default.
• W2974889342 cites W1806891645 @default.
• W2974889342 cites W1967004047 @default.
• W2974889342 cites W1978509807 @default.
• W2974889342 cites W1991769007 @default.
• W2974889342 cites W1992476998 @default.
• W2974889342 cites W1995257611 @default.
• W2974889342 cites W1996579288 @default.
• W2974889342 cites W1997782587 @default.
• W2974889342 cites W1998218174 @default.
• W2974889342 cites W1999358078 @default.
• W2974889342 cites W2001566989 @default.
• W2974889342 cites W2001971840 @default.
• W2974889342 cites W2002985446 @default.
• W2974889342 cites W2005686453 @default.
• W2974889342 cites W2005929157 @default.
• W2974889342 cites W2007208966 @default.
• W2974889342 cites W2008240154 @default.
• W2974889342 cites W2011711856 @default.
• W2974889342 cites W2029374903 @default.
• W2974889342 cites W2034699576 @default.
• W2974889342 cites W2043019763 @default.
• W2974889342 cites W2046424683 @default.
• W2974889342 cites W2052036764 @default.
• W2974889342 cites W2055339690 @default.
• W2974889342 cites W2056631950 @default.
• W2974889342 cites W2061897041 @default.
• W2974889342 cites W2065286706 @default.
• W2974889342 cites W2075637500 @default.
• W2974889342 cites W2083252150 @default.
• W2974889342 cites W2086994991 @default.
• W2974889342 cites W2087298647 @default.
• W2974889342 cites W2088400612 @default.
• W2974889342 cites W2090581299 @default.
• W2974889342 cites W2098086734 @default.
• W2974889342 cites W2098639993 @default.
• W2974889342 cites W2103212315 @default.
• W2974889342 cites W2103234826 @default.
• W2974889342 cites W2103236816 @default.
• W2974889342 cites W2103775384 @default.
• W2974889342 cites W2107244895 @default.
• W2974889342 cites W2111935653 @default.
• W2974889342 cites W2115990514 @default.
• W2974889342 cites W2116360511 @default.
• W2974889342 cites W2119717200 @default.
• W2974889342 cites W2127473914 @default.
• W2974889342 cites W2132272758 @default.
• W2974889342 cites W2137825550 @default.
• W2974889342 cites W2138667780 @default.
• W2974889342 cites W2150012741 @default.
• W2974889342 cites W2154168931 @default.
• W2974889342 cites W2161900876 @default.
• W2974889342 cites W2164012213 @default.
• W2974889342 cites W2164548674 @default.
• W2974889342 cites W2166092475 @default.
• W2974889342 cites W2166226199 @default.
• W2974889342 cites W2166405697 @default.
• W2974889342 cites W2235552820 @default.
• W2974889342 cites W2346736747 @default.
• W2974889342 cites W2600778099 @default.
• W2974889342 cites W2610861664 @default.
• W2974889342 cites W2766736793 @default.
• W2974889342 cites W2795738670 @default.
• W2974889342 cites W2896860503 @default.
• W2974889342 cites W2938321354 @default.
• W2974889342 cites W4229992520 @default.
• W2974889342 cites W4247766879 @default.
• W2974889342 cites W4253020087 @default.
• W2974889342 cites W4256520039 @default.
• W2974889342 doi "https://doi.org/10.1101/775163" @default.
• W2974889342 hasPublicationYear "2019" @default.
• W2974889342 type Work @default.
• W2974889342 sameAs 2974889342 @default.
• W2974889342 citedByCount "0" @default.
• W2974889342 crossrefType "posted-content" @default.
• W2974889342 hasAuthorship W2974889342A5012297652 @default.
• W2974889342 hasAuthorship W2974889342A5046495847 @default.
• W2974889342 hasAuthorship W2974889342A5062687538 @default.
• W2974889342 hasBestOaLocation W29748893421 @default.
• W2974889342 hasConcept C11413529 @default.
• W2974889342 hasConcept C119857082 @default.
• W2974889342 hasConcept C121332964 @default.
• W2974889342 hasConcept C12713177 @default.
• W2974889342 hasConcept C134306372 @default.
• W2974889342 hasConcept C144024400 @default.
• W2974889342 hasConcept C149923435 @default.
• W2974889342 hasConcept C154945302 @default.
• W2974889342 hasConcept C198531522 @default.

• next