Recurrent neural circuits often face inherent complexities in learning and generating their desired outputs, especially when they initially exhibit chaotic spontaneous activity. While the celebrated FORCE learning rule can train chaotic recurrent networks to produce coherent patterns by suppressing chaos, it requires non-local plasticity rules and quick plasticity, raising the question of how synapses adapt on local, biologically plausible timescales to handle potential chaotic dynamics. We propose a novel

Enhanced Preprints Neuroscience A theory and recipe to construct general and biologically plausible integrating continuous attractor neural networks Department of Brain and Cognitive Sciences & McGovern Institute, MIT, Cambridge, United States Integrative Computational Neuroscience Center and Yang-Tan Collective, MIT, Cambridge, United States https://doi.org/10.7554/eLife.107224.1 Reviewed Preprint v1July 28, 2025 Not revised Download Cite Share Share this article Close Cite this article Close Altmetric pro

NEST (NEural Simulation Tool) From Scholarpedia Marc-Oliver Gewaltig and Markus Diesmann (2007), Scholarpedia, 2(4):1430. doi:10.4249/scholarpedia.1430 revision #130182 [ link to/cite this article ] Jump to: navigation , search Post-publication activity Curator: Marc-Oliver Gewaltig Contributors: 0.93 - Markus Diesmann 0.29 - Eugene M. Izhikevich 0.07 - Benjamin Bronner Tobias Denninger Andrew Davison Dr. Marc-Oliver Gewaltig, Blue Brain Project, École Polytechnique Fédérale de Lausanne, CH-1015 Lausanne

Salience Maps Structure and Interpretation of Deep Networks Salience Maps September 10, 2024 • David Bau The field of interpretable machine learning underwent an earthquake in 2012 when Krizhevsky et al published the AlexNet neural network that smashed the competitors in the ImageNet object classification challenge. A Need for Post-Hoc Interpretability While deep networks like AlexNet perform extraordinarily well, they have the following two characteristics: Deep neural layers define long complex

Learning without back-propagation Another really interesting paper that I've come across recently: "NoProp: Training Neural Networks without Back-propagation or Forward-propagation" (Mar 2025) This paper proposes a "back-propagation-free" (kinda) approach to training a denoising (Diffusion / Flow Matching) model. The main difference between typical back-prop and NoProp is that each block (layer) is optimised to denoise independently instead of propagating errors throughout the entire network, think: gradien

--> Login Home Artificial neural networks for intelligent cost estimation – a contribution to strategic cost management in the manufacturing supply chain Bodendorf F, Merkl P, Franke J (2021) Publication Type: Journal article, Original article Publication year: 2021 Journal International Journal of Production Research Taylor & Francis Original Authors: Frank Bodendorf, Philipp Merkl, Jörg Franke Pages Range: 1-22 DOI: 10.1080/00207543.2021.1998697 Abstract In today’s complex supply networks sharing

↓ Skip to main content Altmetric What is this page? Embed badge Share Artificial Neural Networks and Machine Learning – ICANN 2025 Overview of attention for book Table of Contents Altmetric Badge Book Overview Altmetric Badge Chapter 1 ACGCN: A Sequence-Attention-Based Graph Convolutional Model for Enhanced Recommendation Systems Altmetric Badge Chapter 2 Hyperparameter-Free Bi-level Knowledge Graph Optimization for Link Prediction Altmetric Badge Chapter 3 SWIFT: State-Space Wavelet Integrated

Inspired by the extensive success of deep learning, graph neural networks (GNNs) have been proposed to learn expressive node representations and demonstrated promising performance in various graph

Cold-start problem is a fundamental challenge for recommendation tasks. Despite the recent advances on Graph Neural Networks (GNNs) incorporate the high-order collaborative signal to alleviate the problem, the embeddings of the cold-start users and items aren't explicitly optimized, and the cold-start neighbors are not dealt with during the graph convolution in GNNs. This paper proposes to pre-train a GNN model before applying it for recommendation. Unlike the goal of recommendation, the pre-training GNN si

artbrain.org #1 Intro to Neuroaesthetic Theory (1997-99) #2 Cinema and the Brain (2000-02) #3 Buildings, Movies and Brains (2003-04) #4 The Phantom Limb (2005) #5 Conference of Neuroaesthetics (2005) #6 Shifter 16: Pluripotential (2007-11) #7 The Psychopathologies of Cognitive Capitalism: Part One (2013) #8 The Psychopathologies of Cognitive Capitalism: Part Two (2014) #9 The Psychopathologies of Cognitive Capitalism: Part Three (2017) #10 Activist Neuroaesthetics (2021) #11 AI and the Brain (2026) Chaoid G