dorsal/arxiv
View SchemaFeature detection using spikes: the greedy approach
| Authors | Laurent Perrinet |
|---|---|
| Categories | |
| ArXiv ID | q-bio/0611003 |
| URL | https://arxiv.org/abs/q-bio/0611003 |
| DOI | 10.1016/j.jphysparis.2005.09.012 |
| Journal | Journal of physiology, Paris. 98 (28/11/2005) 530--9 |
Abstract
A goal of low-level neural processes is to build an efficient code extracting the relevant information from the sensory input. It is believed that this is implemented in cortical areas by elementary inferential computations dynamically extracting the most likely parameters corresponding to the sensory signal. We explore here a neuro-mimetic feed-forward model of the primary visual area (VI) solving this problem in the case where the signal may be described by a robust linear generative model. This model uses an over-complete dictionary of primitives which provides a distributed probabilistic representation of input features. Relying on an efficiency criterion, we derive an algorithm as an approximate solution which uses incremental greedy inference processes. This algorithm is similar to 'Matching Pursuit' and mimics the parallel architecture of neural computations. We propose here a simple implementation using a network of spiking integrate-and-fire neurons which communicate using lateral interactions. Numerical simulations show that this Sparse Spike Coding strategy provides an efficient model for representing visual data from a set of natural images. Even though it is simplistic, this transformation of spatial data into a spatio-temporal pattern of binary events provides an accurate description of some complex neural patterns observed in the spiking activity of biological neural networks.
{
"annotation_id": "5c5b6336-d427-47bb-9e98-62e37f811332",
"date_created": "2026-03-02T18:01:35.366000Z",
"date_modified": "2026-03-02T18:01:35.366000Z",
"file_hash": "753a072ee01617a466d02301fe3a6b35ac73c55650ba2c1b257bd51fd769bcfe",
"private": false,
"record": {
"abstract": "A goal of low-level neural processes is to build an efficient code extracting\nthe relevant information from the sensory input. It is believed that this is\nimplemented in cortical areas by elementary inferential computations\ndynamically extracting the most likely parameters corresponding to the sensory\nsignal. We explore here a neuro-mimetic feed-forward model of the primary\nvisual area (VI) solving this problem in the case where the signal may be\ndescribed by a robust linear generative model. This model uses an over-complete\ndictionary of primitives which provides a distributed probabilistic\nrepresentation of input features. Relying on an efficiency criterion, we derive\nan algorithm as an approximate solution which uses incremental greedy inference\nprocesses. This algorithm is similar to \u0027Matching Pursuit\u0027 and mimics the\nparallel architecture of neural computations. We propose here a simple\nimplementation using a network of spiking integrate-and-fire neurons which\ncommunicate using lateral interactions. Numerical simulations show that this\nSparse Spike Coding strategy provides an efficient model for representing\nvisual data from a set of natural images. Even though it is simplistic, this\ntransformation of spatial data into a spatio-temporal pattern of binary events\nprovides an accurate description of some complex neural patterns observed in\nthe spiking activity of biological neural networks.",
"arxiv_id": "q-bio/0611003",
"authors": [
"Laurent Perrinet"
],
"categories": [
"q-bio.NC"
],
"doi": "10.1016/j.jphysparis.2005.09.012",
"journal_ref": "Journal of physiology, Paris. 98 (28/11/2005) 530--9",
"title": "Feature detection using spikes: the greedy approach",
"url": "https://arxiv.org/abs/q-bio/0611003"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "331d1edc-b178-4e12-9c4d-dcacc801bf2f",
"id": "arXiv Dataset IDs",
"type": "Model",
"variant": "snapshot-2026-03-01",
"version": "0.1.0"
},
"user_id": 1000002
}