Necessary Conditions for Reliable Propagation of Slowly Time-Varying Firing Rate

Hasanzadeh, Navid; Rezaei, Mohammadreza; Faraz, Sayan; Popović, Miloš R.; Lankarany, Milad

doi:10.3389/fncom.2020.00064

Cited by 4 publications

(2 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While this complicates back-propagation, the proposed method would remain the same as the only nonlocal information required in each synapse is the binary reward signal. It is of interest to determine the applicability of the static policy proposed in this work to CNNs as there is substantially more structure imposed on the application of kernels and these kernel "synapses" generally coexist with a set of fully connected terminal layers (MLP) [38][39][40]. Since 2D convolutions can be expressed as matrix multiplications with circulant matrices [41], it may be that the proposed algorithm would generalize to the CNN architecture.…”

Section: Discussionmentioning

confidence: 99%

Gradient-Free Neural Network Training via Synaptic-Level Reinforcement Learning

2022

Self Cite

View full text Add to dashboard Cite

An ongoing challenge in neural information processing is the following question: how do neurons adjust their connectivity to improve network-level task performance over time (i.e., actualize learning)? It is widely believed that there is a consistent, synaptic-level learning mechanism in specific brain regions, such as the basal ganglia, that actualizes learning. However, the exact nature of this mechanism remains unclear. Here, we investigate the use of universal synaptic-level algorithms in training connectionist models. Specifically, we propose an algorithm based on reinforcement learning (RL) to generate and apply a simple biologically-inspired synaptic-level learning policy for neural networks. In this algorithm, the action space for each synapse in the network consists of a small increase, decrease, or null action on the connection strength. To test our algorithm, we applied it to a multilayer perceptron (MLP) neural network model. This algorithm yields a static synaptic learning policy that enables the simultaneous training of over 20,000 parameters (i.e., synapses) and consistent learning convergence when applied to simulated decision boundary matching and optical character recognition tasks. The trained networks yield character-recognition performance comparable to identically shaped networks trained with gradient descent. The approach has two significant advantages in comparison to traditional gradient-descent-based optimization methods. First, the robustness of our novel method and its lack of reliance on gradient computations opens the door to new techniques for training difficult-to-differentiate artificial neural networks, such as spiking neural networks (SNNs) and recurrent neural networks (RNNs). Second, the method’s simplicity provides a unique opportunity for further development of local information-driven multiagent connectionist models for machine intelligence analogous to cellular automata.

show abstract

Section: Discussionmentioning

confidence: 99%

Gradient-Free Neural Network Training via Synaptic-Level Reinforcement Learning

2022

Self Cite

View full text Add to dashboard Cite

show abstract

“…There are several ways to analyze the propagation of activity in the cortex; some of them can be based only on structural connectivity [44], in amplification caused by changes in projections weights [24] and some based on alterations in firing rates [67]. However, to the best of our knowledge, no studies explore how to acquire signal propagation paths based on directed functional connectivity.…”

Section: The Propagation Of Activity Between Simulated Cortical Areasmentioning

confidence: 99%

Directed Functional and Structural Connectivity in a Large-Scale Model for the Mouse Cortex

Nunes

Reyes

Mejías

et al. 2021

Preprint

View full text Add to dashboard Cite

Inferring the structural connectivity from electrophysiological measurements is a fundamental challenge in systems neuroscience. Directed functional connectivity measures, such as the Generalized Partial Directed Correlation (GPDC), provide estimates of the causal influence between areas. However, such methods have a limitation because their estimates depend on the number of brain regions simultaneously recorded. We analyzed this problem by evaluating the effectiveness of GPDC to estimate the connectivity of a ground-truth, data-constrained computational model of a large-scale mouse cortical network. The model contains 19 cortical areas modeled using spiking neural populations, and directed weights for long-range projections were obtained from a tract-tracing cortical connectome. We show that the GPDC estimates correlate positively with structural connectivity. Moreover, the correlation between structural and directed functional connectivity is comparable even when using only a few cortical areas for GPDC estimation, a typical scenario for electro-physiological recordings. Finally, GPDC measures also provided a measure of the flow of information among cortical areas.

show abstract

The thalamus and tinnitus: Bridging the gap between animal data and findings in humans

Koops

Eggermont

2021

Hearing Research

View full text Add to dashboard Cite

Necessary Conditions for Reliable Propagation of Slowly Time-Varying Firing Rate

Cited by 4 publications

References 44 publications

Gradient-Free Neural Network Training via Synaptic-Level Reinforcement Learning

Gradient-Free Neural Network Training via Synaptic-Level Reinforcement Learning

Directed Functional and Structural Connectivity in a Large-Scale Model for the Mouse Cortex

The thalamus and tinnitus: Bridging the gap between animal data and findings in humans

Contact Info

Product

Resources

About