Cover Image

Reward-based learning under hardware constraints: using a RISC processor embedded in a neuromorphic substrate

In this study, we propose and analyze in simulations a new, highly flexible method of imple- menting synaptic plasticity in a wafer-scale, accelerated neuromorphic hardware system. The study focuses on globally modulated STDP, as a special use-case of this method. Flexibility is achieved by embeddin...

Full description

Saved in:
Bibliographic Details
Main Authors: Friedmann, Simon (Author) , Frémaux, Nicolas (Author) , Schemmel, Johannes (Author) , Gerstner, Wulfram (Author) , Meier, Karlheinz (Author)
Format: Article (Journal)
Language:English
Published: 20 September 2013
In: Frontiers in neuroscience
Year: 2013, Volume: 7, Pages: 1-17
ISSN:1662-453X
DOI:10.3389/fnins.2013.00160
Online Access:Verlag, kostenfrei, Volltext: https://doi.org/10.3389/fnins.2013.00160
Verlag, kostenfrei, Volltext: https://www.frontiersin.org/articles/10.3389/fnins.2013.00160/full
Get full text
Author Notes:Simon Friedmann, Nicolas Frémaux, Johannes Schemmel, Wulfram Gerstner and Karlheinz Meier

MARC

LEADER 00000caa a2200000 c 4500
001 1759240141
003 DE-627
005 20231221092344.0
007 cr uuu---uuuuu
008 210531s2013 xx |||||o 00| ||eng c
024 7 |a 10.3389/fnins.2013.00160  |2 doi 
035 |a (DE-627)1759240141 
035 |a (DE-599)KXP1759240141 
035 |a (OCoLC)1341414905 
040 |a DE-627  |b ger  |c DE-627  |e rda 
041 |a eng 
084 |a 29  |2 sdnb 
100 1 |a Friedmann, Simon  |e VerfasserIn  |0 (DE-588)1038701007  |0 (DE-627)766335380  |0 (DE-576)392728265  |4 aut 
245 1 0 |a Reward-based learning under hardware constraints  |b using a RISC processor embedded in a neuromorphic substrate  |c Simon Friedmann, Nicolas Frémaux, Johannes Schemmel, Wulfram Gerstner and Karlheinz Meier 
264 1 |c 20 September 2013 
300 |a 17 
336 |a Text  |b txt  |2 rdacontent 
337 |a Computermedien  |b c  |2 rdamedia 
338 |a Online-Ressource  |b cr  |2 rdacarrier 
500 |a Gesehen am 31.05.2021 
520 |a In this study, we propose and analyze in simulations a new, highly flexible method of imple- menting synaptic plasticity in a wafer-scale, accelerated neuromorphic hardware system. The study focuses on globally modulated STDP, as a special use-case of this method. Flexibility is achieved by embedding a general-purpose processor dedicated to plasticity into the wafer. To evaluate the suitability of the proposed system, we use a reward modulated STDP rule in a spike train learning task. A single layer of neurons is trained to fire at specific points in time with only the reward as feedback. This model is simulated to measure its performance, i.e. the in- crease in received reward after learning. Using this performance as baseline, we then simulate the model with various constraints imposed by the proposed implementation and compare the performance. The simulated constraints include discretized synaptic weights, a restricted inter- face between analog synapses and embedded processor, and mismatch of analog circuits. We find that probabilistic updates can increase the performance of low-resolution weights, a simple interface between analog synapses and processor is sufficient for learning, and performance is insensitive to mismatch. Further, we consider communication latency between wafer and the conventional control computer system that is simulating the environment. This latency increases the delay, with which the reward is sent to the embedded processor. Because of the time continu- ous operation of the analog synapses, delay can cause a deviation of the updates as compared to the not delayed situation. We find that for highly accelerated systems latency has to be kept to a minimum. This study demonstrates the suitability of the proposed implementation to emulate the selected reward modulated STDP learning rule. It is therefore an ideal candidate for imple- mentation in an upgraded version of the wafer-scale system developed within the BrainScaleS project. 
650 4 |a hardware constraints analysis 
650 4 |a large-scale spiking neural net- works 
650 4 |a neuromorphic hardware 
650 4 |a reinforcement learning 
650 4 |a spike-timing dependent plasticity 
650 4 |a wafer-scale integration 
700 1 |a Frémaux, Nicolas  |d 1979-  |e VerfasserIn  |0 (DE-588)1116513420  |0 (DE-627)87119404X  |0 (DE-576)478442645  |4 aut 
700 1 |a Schemmel, Johannes  |e VerfasserIn  |0 (DE-588)1025834607  |0 (DE-627)72488291X  |0 (DE-576)370821440  |4 aut 
700 1 |a Gerstner, Wulfram  |d 1963-  |e VerfasserIn  |0 (DE-588)1089284381  |0 (DE-627)853080119  |0 (DE-576)459755641  |4 aut 
700 1 |a Meier, Karlheinz  |d 1955-2018  |e VerfasserIn  |0 (DE-588)1025835115  |0 (DE-627)724884114  |0 (DE-576)370822269  |4 aut 
773 0 8 |i Enthalten in  |t Frontiers in neuroscience  |d Lausanne : Frontiers Research Foundation, 2007  |g 7(2013), Artikel-ID 160, Seite 1-17  |h Online-Ressource  |w (DE-627)55908109X  |w (DE-600)2411902-7  |w (DE-576)281378959  |x 1662-453X  |7 nnas  |a Reward-based learning under hardware constraints using a RISC processor embedded in a neuromorphic substrate 
773 1 8 |g volume:7  |g year:2013  |g elocationid:160  |g pages:1-17  |g extent:17  |a Reward-based learning under hardware constraints using a RISC processor embedded in a neuromorphic substrate 
856 4 0 |u https://doi.org/10.3389/fnins.2013.00160  |x Verlag  |x Resolving-System  |z kostenfrei  |3 Volltext 
856 4 0 |u https://www.frontiersin.org/articles/10.3389/fnins.2013.00160/full  |x Verlag  |z kostenfrei  |3 Volltext 
951 |a AR 
992 |a 20210531 
993 |a Article 
994 |a 2013 
998 |g 1025835115  |a Meier, Karlheinz  |m 1025835115:Meier, Karlheinz  |d 130000  |d 130700  |e 130000PM1025835115  |e 130700PM1025835115  |k 0/130000/  |k 1/130000/130700/  |p 5  |y j 
998 |g 1025834607  |a Schemmel, Johannes  |m 1025834607:Schemmel, Johannes  |d 130000  |d 130700  |e 130000PS1025834607  |e 130700PS1025834607  |k 0/130000/  |k 1/130000/130700/  |p 3 
998 |g 1038701007  |a Friedmann, Simon  |m 1038701007:Friedmann, Simon  |d 130000  |d 130700  |e 130000PF1038701007  |e 130700PF1038701007  |k 0/130000/  |k 1/130000/130700/  |p 1  |x j 
999 |a KXP-PPN1759240141  |e 3932665023 
BIB |a Y 
SER |a journal 
JSO |a {"note":["Gesehen am 31.05.2021"],"name":{"displayForm":["Simon Friedmann, Nicolas Frémaux, Johannes Schemmel, Wulfram Gerstner and Karlheinz Meier"]},"recId":"1759240141","origin":[{"dateIssuedKey":"2013","dateIssuedDisp":"20 September 2013"}],"person":[{"family":"Friedmann","given":"Simon","display":"Friedmann, Simon","role":"aut"},{"display":"Frémaux, Nicolas","role":"aut","family":"Frémaux","given":"Nicolas"},{"role":"aut","display":"Schemmel, Johannes","given":"Johannes","family":"Schemmel"},{"role":"aut","display":"Gerstner, Wulfram","given":"Wulfram","family":"Gerstner"},{"given":"Karlheinz","family":"Meier","role":"aut","display":"Meier, Karlheinz"}],"language":["eng"],"relHost":[{"physDesc":[{"extent":"Online-Ressource"}],"title":[{"title":"Frontiers in neuroscience","title_sort":"Frontiers in neuroscience"}],"language":["eng"],"name":{"displayForm":["Frontiers Research Foundation"]},"disp":"Reward-based learning under hardware constraints using a RISC processor embedded in a neuromorphic substrateFrontiers in neuroscience","note":["Gesehen am 04.06.20"],"type":{"media":"Online-Ressource","bibl":"periodical"},"id":{"zdb":["2411902-7"],"issn":["1662-453X"],"eki":["55908109X"]},"part":{"text":"7(2013), Artikel-ID 160, Seite 1-17","pages":"1-17","extent":"17","year":"2013","volume":"7"},"pubHistory":["1.2007 -"],"origin":[{"dateIssuedKey":"2007","publisherPlace":"Lausanne","publisher":"Frontiers Research Foundation","dateIssuedDisp":"2007-"}],"recId":"55908109X"}],"title":[{"subtitle":"using a RISC processor embedded in a neuromorphic substrate","title":"Reward-based learning under hardware constraints","title_sort":"Reward-based learning under hardware constraints"}],"id":{"doi":["10.3389/fnins.2013.00160"],"eki":["1759240141"]},"type":{"bibl":"article-journal","media":"Online-Ressource"},"physDesc":[{"extent":"17 S."}]} 
SRT |a FRIEDMANNSREWARDBASE2020 

Similar Items