Analog Memories in a Balanced Rate-Based Network of E-I Neurons
Authors: Dylan Festa, Guillaume Hennequin, Mate Lengyel
NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We directly optimize networks of excitatory and inhibitory neurons to force sets of arbitrary analog patterns to become stable fixed points of the dynamics. The resulting networks operate in the balanced regime, are robust to corruptions of the memory cue as well as to ongoing noise, and incidentally explain the reduction of trial-to-trial variability following stimulus onset that is ubiquitously observed in sensory and motor cortices. For recall, we initialize neuronal activities at a noisy version of one of the target patterns, and study the subsequent evolution of the network state. The network performs well if its dynamics clean up the noise and home in on the target pattern (autoassociative behavior) and if it achieves this robustly even in the face of large amounts of noise. |
| Researcher Affiliation | Academia | Computational & Biological Learning Lab, Department of Engineering University of Cambridge, UK |
| Pseudocode | No | The paper describes the mathematical model and optimization procedure in detail but does not provide structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any concrete access information (e.g., repository link, explicit statement of code release, or mention of supplementary materials containing code) for the described methodology. |
| Open Datasets | No | For the remaining target patterns, {vµ exc}µ=2,...,m were generated by inverting (using g 1) firing rates that were sampled from a log-normal distribution with a mean matching the baseline firing rate, rbaseline (Fig. 1a) and a variance of 5 Hz. |
| Dataset Splits | No | The paper describes generating target patterns and optimizing network parameters to embed them, but it does not specify explicit training/validation/test dataset splits in the conventional sense for reproducibility. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., exact GPU/CPU models, processor types, or memory amounts) used for running its experiments. |
| Software Dependencies | No | The paper mentions using "a variant of the low-storage BFGS algorithm included in the open source library NLopt [17]" but does not provide specific version numbers for NLopt or any other software dependencies. |
| Experiment Setup | Yes | Table 1: Parameter settings (n E 100, τE 20 ms, ηs 0.02, n I 50, τI 10 ms, ηF 0.001, m 30, rbaseline 5 Hz). We set γ to 0.04, such that g(vi) spans a few tens of Hz when vi spans a few tens of m V, as experimentally observed in cortical areas (e.g. cat s V1 [16]). |