Blooming and pruning: learning from mistakes with memristive synapses
Abstract Blooming and pruning is one of the most important developmental mechanisms of the biological brain in the first years of life, enabling it to adapt its network structure to the demands of the environment. The mechanism is thought to be fundamental for the development of cognitive skills. In...
Main Authors: | , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Nature Portfolio
2024-04-01
|
Series: | Scientific Reports |
Subjects: | |
Online Access: | https://doi.org/10.1038/s41598-024-57660-4 |
_version_ | 1797219827506479104 |
---|---|
author | Kristina Nikiruy Eduardo Perez Andrea Baroni Keerthi Dorai Swamy Reddy Stefan Pechmann Christian Wenger Martin Ziegler |
author_facet | Kristina Nikiruy Eduardo Perez Andrea Baroni Keerthi Dorai Swamy Reddy Stefan Pechmann Christian Wenger Martin Ziegler |
author_sort | Kristina Nikiruy |
collection | DOAJ |
description | Abstract Blooming and pruning is one of the most important developmental mechanisms of the biological brain in the first years of life, enabling it to adapt its network structure to the demands of the environment. The mechanism is thought to be fundamental for the development of cognitive skills. Inspired by this, Chialvo and Bak proposed in 1999 a learning scheme that learns from mistakes by eliminating from the initial surplus of synaptic connections those that lead to an undesirable outcome. Here, this idea is implemented in a neuromorphic circuit scheme using CMOS integrated HfO2-based memristive devices. The implemented two-layer neural network learns in a self-organized manner without positive reinforcement and exploits the inherent variability of the memristive devices. This approach provides hardware, local, and energy-efficient learning. A combined experimental and simulation-based parameter study is presented to find the relevant system and device parameters leading to a compact and robust memristive neuromorphic circuit that can handle association tasks. |
first_indexed | 2024-04-24T12:39:50Z |
format | Article |
id | doaj.art-d240a07a7079409997238de754e162bc |
institution | Directory Open Access Journal |
issn | 2045-2322 |
language | English |
last_indexed | 2024-04-24T12:39:50Z |
publishDate | 2024-04-01 |
publisher | Nature Portfolio |
record_format | Article |
series | Scientific Reports |
spelling | doaj.art-d240a07a7079409997238de754e162bc2024-04-07T11:19:14ZengNature PortfolioScientific Reports2045-23222024-04-0114111110.1038/s41598-024-57660-4Blooming and pruning: learning from mistakes with memristive synapsesKristina Nikiruy0Eduardo Perez1Andrea Baroni2Keerthi Dorai Swamy Reddy3Stefan Pechmann4Christian Wenger5Martin Ziegler6Micro- and Nanoelectronic Systems, Department of Electrical Engineering and Information Technology, TU IlmenauIHP - Leibniz-Institut fuer innovative MikroelektronikIHP - Leibniz-Institut fuer innovative MikroelektronikIHP - Leibniz-Institut fuer innovative MikroelektronikChair of Micro- and Nanosystems Technology, Technical University of MunichIHP - Leibniz-Institut fuer innovative MikroelektronikMicro- and Nanoelectronic Systems, Department of Electrical Engineering and Information Technology, TU IlmenauAbstract Blooming and pruning is one of the most important developmental mechanisms of the biological brain in the first years of life, enabling it to adapt its network structure to the demands of the environment. The mechanism is thought to be fundamental for the development of cognitive skills. Inspired by this, Chialvo and Bak proposed in 1999 a learning scheme that learns from mistakes by eliminating from the initial surplus of synaptic connections those that lead to an undesirable outcome. Here, this idea is implemented in a neuromorphic circuit scheme using CMOS integrated HfO2-based memristive devices. The implemented two-layer neural network learns in a self-organized manner without positive reinforcement and exploits the inherent variability of the memristive devices. This approach provides hardware, local, and energy-efficient learning. A combined experimental and simulation-based parameter study is presented to find the relevant system and device parameters leading to a compact and robust memristive neuromorphic circuit that can handle association tasks.https://doi.org/10.1038/s41598-024-57660-4Memristive devicesNeuromorphic computingLearning from mistakes |
spellingShingle | Kristina Nikiruy Eduardo Perez Andrea Baroni Keerthi Dorai Swamy Reddy Stefan Pechmann Christian Wenger Martin Ziegler Blooming and pruning: learning from mistakes with memristive synapses Scientific Reports Memristive devices Neuromorphic computing Learning from mistakes |
title | Blooming and pruning: learning from mistakes with memristive synapses |
title_full | Blooming and pruning: learning from mistakes with memristive synapses |
title_fullStr | Blooming and pruning: learning from mistakes with memristive synapses |
title_full_unstemmed | Blooming and pruning: learning from mistakes with memristive synapses |
title_short | Blooming and pruning: learning from mistakes with memristive synapses |
title_sort | blooming and pruning learning from mistakes with memristive synapses |
topic | Memristive devices Neuromorphic computing Learning from mistakes |
url | https://doi.org/10.1038/s41598-024-57660-4 |
work_keys_str_mv | AT kristinanikiruy bloomingandpruninglearningfrommistakeswithmemristivesynapses AT eduardoperez bloomingandpruninglearningfrommistakeswithmemristivesynapses AT andreabaroni bloomingandpruninglearningfrommistakeswithmemristivesynapses AT keerthidoraiswamyreddy bloomingandpruninglearningfrommistakeswithmemristivesynapses AT stefanpechmann bloomingandpruninglearningfrommistakeswithmemristivesynapses AT christianwenger bloomingandpruninglearningfrommistakeswithmemristivesynapses AT martinziegler bloomingandpruninglearningfrommistakeswithmemristivesynapses |