Scaling Self-Organizing Maps to Model Large Cortical Networks

Kelkar, Amol; Bednar, James A.; Miikkulainen, Risto

doi:10.1385/ni:2:3:275

Cited by 24 publications

(19 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the one hand, the model neurons defined in our approach represent an ensemble of natural cortical neurons, which are spatially and functionally closely connected. Thus, each model neuron may represent, for example, a cortical column (this concept is mainly used in vision, e.g., by Obermayer et al, 1990 and Bednar et al, 2004). On the other hand, the auditory and the semantic maps are connected by simple associative links rather than connectionist activity propagation such as Hebbian learning.…”

Section: Discussionmentioning

confidence: 99%

Interconnected growing self-organizing maps for auditory and semantic acquisition modeling

Cao

Fang

et al. 2014

Front. Psychol.

View full text Add to dashboard Cite

Based on the incremental nature of knowledge acquisition, in this study we propose a growing self-organizing neural network approach for modeling the acquisition of auditory and semantic categories. We introduce an Interconnected Growing Self-Organizing Maps (I-GSOM) algorithm, which takes associations between auditory information and semantic information into consideration, in this paper. Direct phonetic–semantic association is simulated in order to model the language acquisition in early phases, such as the babbling and imitation stages, in which no phonological representations exist. Based on the I-GSOM algorithm, we conducted experiments using paired acoustic and semantic training data. We use a cyclical reinforcing and reviewing training procedure to model the teaching and learning process between children and their communication partners. A reinforcing-by-link training procedure and a link-forgetting procedure are introduced to model the acquisition of associative relations between auditory and semantic information. Experimental results indicate that (1) I-GSOM has good ability to learn auditory and semantic categories presented within the training data; (2) clear auditory and semantic boundaries can be found in the network representation; (3) cyclical reinforcing and reviewing training leads to a detailed categorization as well as to a detailed clustering, while keeping the clusters that have already been learned and the network structure that has already been developed stable; and (4) reinforcing-by-link training leads to well-perceived auditory–semantic associations. Our I-GSOM model suggests that it is important to associate auditory information with semantic information during language acquisition. Despite its high level of abstraction, our I-GSOM approach can be interpreted as a biologically-inspired neurocomputational model.

show abstract

Section: Discussionmentioning

confidence: 99%

Interconnected growing self-organizing maps for auditory and semantic acquisition modeling

Cao

Fang

et al. 2014

Front. Psychol.

View full text Add to dashboard Cite

show abstract

“…Point‐to‐point accuracy in these maps is achieved through an activity‐dependent mechanism in which synapses that are effective in depolarizing postsynaptic cells are stabilized while those that are ineffective are retracted. These maps are thought to be essential for functions that include depth perception, object recognition, reconstruction of a visual scene and visually guided behaviours (DeAngelis, 2000; Bednar et al ., 2004; Navalpakkam & Itti, 2005; Weber et al ., 2005).…”

Section: Introductionmentioning

confidence: 99%

Nicotine exposure refines visual map topography through an NMDA receptor‐mediated pathway

Yan

Zhao

Butt

et al. 2006

Eur J of Neuroscience

View full text Add to dashboard Cite

The precise mapping of one surface onto another is fundamental to visual system organization and depends upon adequate stimulation of postsynaptic targets to stabilize correctly placed synapses. As exogenous nicotine alters neuronal activity, we investigated whether it would affect the visual map created by retinal ganglion cell terminals in the frog optic tectum. Chronic exposure of the tectum to nicotine decreased the retinal area from which cells project to a given tectal site. This map refinement was also produced by exposure to either the alpha-bungarotoxin sensitive nicotinic receptor agonist, anatoxin-a or the alpha-bungarotoxin-insensitive nicotinic receptor agonist epiboxidine. Immunocytochemical studies using mAb306 and mAb22 demonstrated that alpha-bungarotoxin-sensitive and -insensitive nicotinic receptors, respectively, occupied different tectal sites. Choline acetyltransferase immunoreactivity overlapped with mAb306, but not mAb22, staining. The developing optic tectum was more sensitive to nicotine than the adult tectum and nicotine induced both map refinements and map disruptions in a concentration-dependent manner. Blockade of the N-methyl-D-aspartate (NMDA) receptor with D(-)-2-amino-5-phosphonopentanoic acid (D-APV) prevented nicotine from refining the map in the adult tectum. Exposure to the use-dependent NMDA antagonist MK801 alone had no effect on retinotectal topography but in combination with either NMDA or nicotine it disrupted the map. Exposure to NMDA alone produced refinement. We conclude that the map refinement induced by chronic nicotine treatment has as its basis an increase in the level of NMDA receptor activity. The data are consistent with a model whereby map topography can be bidirectionally affected by either increasing or decreasing NMDA receptor activity.

show abstract

“…This is due to the fact that there are many solutions to the same problem (note, for instance, that solutions are invariant up to a permutation of neurons' addresses). It is possible to decrease these degrees of freedom by including, for instance, topological links between filters (Bednar, Kelkar, & Miikkulainen, 2004). Qualitatively, the main difference between both results is that filters produced by aSSC look more diverse and broad (so that they often overlap), while the filters produced by SparseNet are more localized and thin.…”

Section: Receptive Field Formationmentioning

confidence: 95%

Role of Homeostasis in Learning Sparse Representations

Perrinet

2010

Neural Computation

View full text Add to dashboard Cite

Neurons in the input layer of primary visual cortex in primates develop edge-like receptive fields. One approach to understanding the emergence of this response is to state that neural activity has to efficiently represent sensory data with respect to the statistics of natural scenes. Furthermore, it is believed that such an efficient coding is achieved using a competition across neurons so as to generate a sparse representation, that is, where a relatively small number of neurons are simultaneously active. Indeed, different models of sparse coding, coupled with Hebbian learning and homeostasis, have been proposed that successfully match the observed emergent response. However, the specific role of homeostasis in learning such sparse representations is still largely unknown. By quantitatively assessing the efficiency of the neural representation during learning, we derive a cooperative homeostasis mechanism that optimally tunes the competition between neurons within the sparse coding algorithm. We apply this homeostasis while learning small patches taken from natural images and compare its efficiency with state-of-the-art algorithms. Results show that while different sparse coding algorithms give similar coding results, the homeostasis provides an optimal balance for the representation of natural images within the population of neurons. Competition in sparse coding is optimized when it is fair. By contributing to optimizing statistical competition across neurons, homeostasis is crucial in providing a more efficient solution to the emergence of independent components.

show abstract

Scaling Self-Organizing Maps to Model Large Cortical Networks

Cited by 24 publications

References 26 publications

Interconnected growing self-organizing maps for auditory and semantic acquisition modeling

Interconnected growing self-organizing maps for auditory and semantic acquisition modeling

Nicotine exposure refines visual map topography through an NMDA receptor‐mediated pathway

Role of Homeostasis in Learning Sparse Representations

Contact Info

Product

Resources

About