Massively distributed digital implementation of an integrate-and-fire LEGION network for visual scene segmentation

Girau, Bernard; Torres-Huitzil, César

doi:10.1016/j.neucom.2006.11.009

Cited by 30 publications

(26 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although there is no requirement that a practical system should be based on a physiological account, ASA approaches based on neural models are attractive because they are comprised of simple, parallel and distributed components and are therefore well suited to hardware implementation (see also Chapter 10). For example, Cosp and Madrenas (2003) describe a very large-scale integration (VLSI) implementation of Wang and Terman's LEGION network, comprising a grid of 16 × 16 neural oscillators (see also Girau and Torres-Huitzil 2007). Similar devices built on a larger scale could provide the basis for hardware implementation of physiological ASA models, and their compact size and low power consumption could make them suitable for application in hearing aids.…”

Section: Discussionmentioning

confidence: 99%

Physiological Models of Auditory Scene Analysis

Brown

2010

Computational Models of the Auditory System

View full text Add to dashboard Cite

Human listeners are remarkably adept at perceiving speech and other sounds in unfavorable acoustic environments. Typically, the sound source of interest is contaminated by other acoustic sources, and listeners are therefore faced with the problem of unscrambling the mixture of sounds that arrives at their ears. Nonetheless, human listeners can segregate one voice from a mixture of many voices at a cocktail party, or follow a single melodic line in a performance of orchestral music. Much as the visual system must combine information about edges, colors and textures in order to identify perceptual wholes (e.g., a face or a table), so the auditory system must solve an analogous auditory scene analysis (ASA) problem in order to recover a perceptual description of a single sound source (Bregman 1990). Understanding how the ASA problem is solved at the physiological level is one of the greatest challenges of hearing science, and is one that lies at the core of the "systems" approach of this book.The emerging field of computational auditory scene analysis (CASA) aims to develop machine systems that mimic the ability of human listeners to perceptually segregate acoustic mixtures (Wang and Brown 2006). However, most CASA systems are motivated by engineering applications (e.g., robust automatic speech recognition in noise), and take inspiration from psychophysical and physiological accounts of human ASA without slavishly adhering to them. The review in this chapter is therefore necessarily selective, and concerns only those computational models of ASA that are based on physiologically plausible mechanisms.The next section briefly reviews the psychology of ASA, and then Sect. 8.3 considers the likely physiological basis of ASA in general terms, and its relationship with the wider issue of feature binding in the brain. Computer models of specific ASA

show abstract

Section: Discussionmentioning

confidence: 99%

Physiological Models of Auditory Scene Analysis

Brown

2010

Computational Models of the Auditory System

View full text Add to dashboard Cite

show abstract

“…Their model segments images and binds their pixels to form shapes of objects using local excitatory lateral connections. Girau et al [5] had implemented integrate-andfire neurons to the standard LEGION (Local Excitatory Global Inhibitory Oscillator Network) architecture to segment grey-level images. In order to segment images, the LEGION model groups oscillators that receive their input from similar features in an image.…”

Section: Introductionmentioning

confidence: 99%

Image Processing with Spiking Neuron Networks

Meftah

Lézoray

Chaturvedi

et al. 2013

Studies in Computational Intelligence

View full text Add to dashboard Cite

Abstract. Artificial neural networks have been well developed so far. First two generations of neural networks have had a lot of successful applications. Spiking Neuron Networks (SNNs) are often referred to as the third generation of neural networks which have potential to solve problems related to biological stimuli. They derive their strength and interest from an accurate modeling of synaptic interactions between neurons, taking into account the time of spike emission.SNNs overcome the computational power of neural networks made of threshold or sigmoidal units. Based on dynamic event-driven processing, they open up new horizons for developing models with an exponential capacity of memorizing and a strong ability to fast adaptation. Moreover, SNNs add a new dimension, the temporal axis, to the representation capacity and the processing abilities of neural networks. In this chapter, we present how SNN can be applied with efficacy in image clustering, segmentation and edge detection. Results obtained confirm the validity of the approach.

show abstract

“…Several architectures have already been presented, ranging from large systems [13], [14], [15] able to process very large networks (> 10, 000 of neurons) at many times real-time speed, to compact designs where small networks are directly mapped onto hardware [6], [16], [17], [18] using small neuron processing elements (PEs). These implementations span a small part of a larger design space which allows to make a trade-off between area (chip area and memory footprint) and calculation time.…”

Section: Introductionmentioning

confidence: 99%

Compact hardware for real-time speech recognition using a Liquid State Machine

Schrauwen

D’Haene

Verstraeten

et al. 2007

2007 International Joint Conference on Neural Networks

View full text Add to dashboard Cite

Hardware implementations of Spiking Neural Networks are numerous because they are well suited for implementation in digital and analog hardware, and outperform classic neural networks. This work presents an application driven digital hardware exploration where we implement realtime, isolated digit speech recognition using a Liquid State Machine (a recurrent neural network of spiking neurons where only the output layer is trained). First we test two existing hardware architectures, but they appear to be too fast and thus area consuming for this application. Then we present a scalable, serialised architecture that allows a very compact implementation of spiking neural networks that is still fast enough for real-time processing. This work shows that there is actually a large hardware design space of Spiking Neural Network hardware that can be explored. Existing architectures only spanned part of it.

show abstract

Massively distributed digital implementation of an integrate-and-fire LEGION network for visual scene segmentation

Cited by 30 publications

References 13 publications

Physiological Models of Auditory Scene Analysis

Physiological Models of Auditory Scene Analysis

Image Processing with Spiking Neuron Networks

Compact hardware for real-time speech recognition using a Liquid State Machine

Contact Info

Product

Resources

About