Apical dendrites as a site for gradient calculations
Blake Richards, Assistant Professor, Associate Fellow of the Canadian Institute for Advanced Research (CIFAR)
Abstract:
Theoretical and empirical results in the neural networks literature demonstrate that effective learning at a real-world scale requires changes to synaptic weights that approximate the gradient of a global loss function. For neuroscientists, this means that the brain must have mechanisms for communicating loss gradients between regions, either explicitly or implicitly. Here, I describe our research into potential means of communicating loss gradients using the unique properties of apical dendrites in pyramidal neurons. I will present modelling work showing that, in principle, ensembles of pyramidal neurons could using the temporal derivative of their activity to estimate cost gradients. I will also show how this can be learned using the discontinuities that spikes induce. Finally, I will discuss specific experimental predictions that arise from these theories.
This event is co-organized by the CBMM Trainee Leadership Council.
Видео Apical dendrites as a site for gradient calculations канала MITCBMM
Abstract:
Theoretical and empirical results in the neural networks literature demonstrate that effective learning at a real-world scale requires changes to synaptic weights that approximate the gradient of a global loss function. For neuroscientists, this means that the brain must have mechanisms for communicating loss gradients between regions, either explicitly or implicitly. Here, I describe our research into potential means of communicating loss gradients using the unique properties of apical dendrites in pyramidal neurons. I will present modelling work showing that, in principle, ensembles of pyramidal neurons could using the temporal derivative of their activity to estimate cost gradients. I will also show how this can be learned using the discontinuities that spikes induce. Finally, I will discuss specific experimental predictions that arise from these theories.
This event is co-organized by the CBMM Trainee Leadership Council.
Видео Apical dendrites as a site for gradient calculations канала MITCBMM
Показать
Комментарии отсутствуют
Информация о видео
Другие видео канала
Adversarial examples and human-ML alignmentThe quest for provably efficient ML algorithmsGLMsingle: a toolbox for improving single-trial fMRI response estimatesRecurrent computations to the rescueA Conversation with Prof. Eric Shea-BrownNeural Data Analysis Tutorial - Part 231 - Resting State & InstaCorr: Part 2 of 2What are the computations underlying primate vs. machine vision?CBMM Undergraduate Summer Research Internships in Neuroscience 2016How does our brain learn new dance moves?Investigacion y Superandola AdversidadComputational models of cognition:Reverse-engineering common sense in the human mind and brain Pt 1Language & VisionReintegrating AI: Skills, Symbols, and the Sensorimotor DilemmaPanel Discussion: Hilbert questions in AIComputation and Learning with Assemblies of NeuronsHippocampal memory, cognition, and the role of sleep - part 2EAC Panel on AI and JobsThe Neural Circuits for Face RecognitionClustering and generalization of abstract structures in reinforcement learningUsing population decoding to understand neural content and coding