Dissecting cascade computational components in spiking neural networks

Abstract

Finding out the physical structure of neuronal circuits that governs neuronal responses is an important goal for brain research. With fast advances for large-scale recording techniques, identification of a neuronal circuit with multiple neurons and stages or layers becomes possible and highly demanding. Although methods for mapping the connection structure of circuits have been greatly developed in recent years, they are mostly limited to simple scenarios of a few neurons in a pairwise fashion; and dissecting dynamical circuits, particularly mapping out a complete functional circuit that converges to a single neuron, is still a challenging question. Here, we show that a recent method, termed spike-triggered non-negative matrix factorization (STNMF), can address these issues. By simulating different scenarios of spiking neural networks with various connections between neurons and stages, we demonstrate that STNMF is a persuasive method to dissect functional connections within a circuit. Using spiking activities recorded at neurons of the output layer, STNMF can obtain a complete circuit consisting of all cascade computational components of presynaptic neurons, as well as their spiking activities. For simulated simple and complex cells of the primary visual cortex, STNMF allows us to dissect the pathway of visual computation. Taken together, these results suggest that STNMF could provide a useful approach for investigating neuronal systems leveraging recorded functional neuronal activity.

Representing the dynamics of high-dimensional data with non-redundant wavelets

“… 31 The dimensions revealed by nonlinear methods are more meaningful for computation; 31 , 42 however, they are difficult to be interpreted by biological correspondences, except for a few cases. 31 , 43 , 44 , 45 A practical way of justifying these dimensions is to use a decoder for computation. 13 The WCMI proposed here can extract features well for decoding.…”

Artificial intelligence techniques for retinal prostheses: a comprehensive review and future direction

“…In addition, this model achieved comparable performance with the CNN model and with 15 times power consumption reduction compared with that of CNN model. Another study first proposed an SRNN model with a good performance in regression tasks [177,178]. This model has a high potential to predict the firing rate of multiple ganglion cells.…”

Protocol for dissecting cascade computational components in neural networks of a visual system

“… Revised according to Jia et al (2021). 1

>Weight = [1 1 1 1]; >ModelSubRF_2layers_LNLN_GC(Weight);

Note: The setup of this model is equivalent to a neural network with two layers (synaptic transmission membrane potential). 5 Figure 1 A, right, showcases some examples of stimulus images, which are black and white checkerboard stimulus with a resolution of 8 × 8.…”