2022 roadmap on neuromorphic computing and engineering

Neuromorphic Computing and Engineering, Journal Year: 2022, Volume and Issue: 2(2), P. 022501 - 022501

Published: Jan. 12, 2022

Abstract Modern computation based on von Neumann architecture is now a mature cutting-edge science. In the architecture, processing and memory units are implemented as separate blocks interchanging data intensively continuously. This transfer responsible for large part of power consumption. The next generation computer technology expected to solve problems at exascale with 10 18 calculations each second. Even though these future computers will be incredibly powerful, if they type architectures, consume between 20 30 megawatts not have intrinsic physically built-in capabilities learn or deal complex our brain does. These needs can addressed by neuromorphic computing systems which inspired biological concepts human brain. new has potential used storage amounts digital information much lower consumption than conventional processors. Among their applications, an important niche moving control from centers edge devices. aim this roadmap present snapshot state provide opinion challenges opportunities that holds in major areas technology, namely materials, devices, circuits, algorithms, ethics. collection perspectives where leading researchers community own view about current research area. We hope useful resource providing concise yet comprehensive introduction readers outside field, those who just entering well established community.

Language: Английский

---

Theory Before the Test: How to Build High-Verisimilitude Explanatory Theories in Psychological Science

Perspectives on Psychological Science, Journal Year: 2021, Volume and Issue: 16(4), P. 682 - 697

Published: Jan. 6, 2021

Drawing on the philosophy of psychological explanation, we suggest that science, by focusing effects, may lose sight its primary explananda: capacities. We revisit Marr's levels-of-analysis framework, which has been remarkably productive and useful for cognitive explanation. discuss ways in framework be extended to other areas psychology, such as social, developmental, evolutionary bringing new benefits these fields. then show how theoretical analyses can endow a theory with minimal plausibility even before contact empirical data: call this cycle. Finally, explain our proposal contribute addressing critical issues including leverage effects understand capacities better.

Language: Английский

---

Artificial Neural Networks for Neuroscientists: A Primer

Neuron, Journal Year: 2020, Volume and Issue: 107(6), P. 1048 - 1070

Published: Sept. 1, 2020

Language: Английский

---

Training Spiking Neural Networks Using Lessons From Deep Learning

Proceedings of the IEEE, Journal Year: 2023, Volume and Issue: 111(9), P. 1016 - 1054

Published: Sept. 1, 2023

The brain is the perfect place to look for inspiration develop more efficient neural networks. inner workings of our synapses and neurons provide a glimpse at what future deep learning might like. This article serves as tutorial perspective showing how apply lessons learned from several decades research in learning, gradient descent, backpropagation, neuroscience biologically plausible spiking networks (SNNs). We also explore delicate interplay between encoding data spikes process; challenges solutions applying gradient-based SNNs; subtle link temporal backpropagation spike timing-dependent plasticity; move toward online learning. Some ideas are well accepted commonly used among neuromorphic engineering community, while others presented or justified first time here. A series companion interactive tutorials complementary this using Python package, snnTorch , made available: https://snntorch.readthedocs.io/en/latest/tutorials/index.html.

Language: Английский

---

Deep Reinforcement Learning and Its Neuroscientific Implications

Neuron, Journal Year: 2020, Volume and Issue: 107(4), P. 603 - 616

Published: July 13, 2020

Language: Английский

---

Burst-dependent synaptic plasticity can coordinate learning in hierarchical circuits

Nature Neuroscience, Journal Year: 2021, Volume and Issue: 24(7), P. 1010 - 1019

Published: May 13, 2021

Language: Английский

---

Inductive biases for deep learning of higher-level cognition

Proceedings of the Royal Society A Mathematical Physical and Engineering Sciences, Journal Year: 2022, Volume and Issue: 478(2266)

Published: Oct. 1, 2022

A fascinating hypothesis is that human and animal intelligence could be explained by a few principles (rather than an encyclopaedic list of heuristics). If was correct, we more easily both understand our own build intelligent machines. Just like in physics, the themselves would not sufficient to predict behaviour complex systems brains, substantial computation might needed simulate human-like intelligence. This suggest studying kind inductive biases humans animals exploit help clarify these provide inspiration for AI research neuroscience theories. Deep learning already exploits several key biases, this work considers larger list, focusing on those which concern mostly higher-level sequential conscious processing. The objective clarifying particular they potentially us benefiting from humans’ abilities terms flexible out-of-distribution systematic generalization, currently area where large gap exists between state-of-the-art machine

Language: Английский

---

Deep learning as a tool for ecology and evolution

Methods in Ecology and Evolution, Journal Year: 2022, Volume and Issue: 13(8), P. 1640 - 1660

Published: May 30, 2022

Abstract Deep learning is driving recent advances behind many everyday technologies, including speech and image recognition, natural language processing autonomous driving. It also gaining popularity in biology, where it has been used for automated species identification, environmental monitoring, ecological modelling, behavioural studies, DNA sequencing population genetics phylogenetics, among other applications. relies on artificial neural networks predictive modelling excels at recognizing complex patterns. In this review we synthesize 818 studies using deep the context of ecology evolution to give a discipline‐wide perspective necessary promote rethinking inference approaches field. We provide an introduction machine contrast with mechanistic inference, followed by gentle primer learning. applications discuss its limitations efforts overcome them. practical biologists interested their toolkit identify possible future find that being rapidly adopted evolution, 589 (64%) published since beginning 2019. Most use convolutional (496 studies) supervised identification but tasks molecular data, sounds, data or video as input. More sophisticated uses biology are appear. Operating within paradigm, can be viewed alternative modelling. desirable properties good performance scaling increasing complexity, while posing unique challenges such sensitivity bias input data. expect rapid adoption will continue, especially automation biodiversity monitoring discovery from genetic Increased unsupervised visualization clusters gaps, simplification multi‐step analysis pipelines, integration into graduate postgraduate training all likely near future.

Language: Английский

---

Spine dynamics in the brain, mental disorders and artificial neural networks

Nature reviews. Neuroscience, Journal Year: 2021, Volume and Issue: 22(7), P. 407 - 422

Published: May 28, 2021

Language: Английский

---

The neuroconnectionist research programme

Nature reviews. Neuroscience, Journal Year: 2023, Volume and Issue: 24(7), P. 431 - 450

Published: May 30, 2023

Language: Английский

---