David Chik

2009 Special Issue: A neural model of selective attention and object segmentation in the visual scene: An approach based on partial synchronization and star-like architecture of connections

A brain-inspired computational system is presented that allows sequential selection and processing of objects from a visual scene. The system is comprised of three modules. The selective attention module is designed as a network of spiking neurons of the Hodgkin-Huxley type with star-like connections between the central unit and peripheral elements. The attention focus is represented by those peripheral neurons that generate spikes synchronously with the central neuron while the activity of other peripheral neurons is suppressed. Such dynamics corresponds to the partial synchronization mode. It is shown that peripheral neurons with higher firing rates are preferentially drawn into partial synchronization. We show that local excitatory connections facilitate synchronization, while local inhibitory connections help distinguishing between two groups of peripheral neurons with similar intrinsic frequencies. The module automatically scans a visual scene and sequentially selects regions of interest for detailed processing and object segmentation. The contour extraction module implements standard image processing algorithms for contour extraction. The module computes raw contours of objects accompanied by noise and some spurious inclusions. At the next stage, the object segmentation module designed as a network of phase oscillators is used for precise determination of object boundaries and noise suppression. This module has a star-like architecture of connections. The segmented object is represented by a group of peripheral oscillators working in the regime of partial synchronization with the central oscillator. The functioning of each module is illustrated by an example of processing of the visual scene taken from a visual stream of a robot camera.

Visual perception of ambiguous figures: synchronization based neural models

We develop and study two neural network models of perceptual alternations. Both models have a star-like architecture of connections with a central element connected to a set of peripheral elements. A particular perception is simulated in terms of partial synchronization between the central element and some sub-group of peripheral elements. The first model is constructed from phase oscillators and the mechanism of perceptual alternations is based on chaotic intermittency under fixed parameter values. Similar to experimental evidence, the distribution of times between perceptual alternations is represented by the gamma distribution. The second model is built of spiking neurons of the Hodgkin–Huxley type. The mechanism of perceptual alternations is based on plasticity of inhibitory synapses which increases the inhibition from the central unit to the neural assembly representing the current percept. As a result another perception is formed. Simulations show that the second model is in good agreement with behavioural data on switching times between percepts of ambiguous figures and with experimental results on binocular rivalry of two and four percepts.

Selective attention model with spiking elements

A new biologically plausible model of visual selective attention is developed based on synaptically coupled Hodgkin-Huxley neurons. The model is designed according to a two-layer architecture of excitatory and inhibitory connections which comprises two central neurons and a population of peripheral neurons. Two types of inhibition from the central neurons are present: fixed inhibition which is responsible for the formation of the attention focus, and short-term plastic inhibition which is responsible for the shift of attention. The regimes of synchronous dynamics associated with the development of the attentional focus are studied. In particular, the regime of partial synchronization between spiking activity of the central and peripheral neurons is interpreted as object selection to the focus of attention. It is shown that peripheral neurons with higher firing rates are selected preferentially by the attention system. The model correctly reproduces some observations concerning the mechanisms of attentional control, such as the coherence of spikes in the population of neurons included in the focus of attention, and the inhibition of neurons outside the focus of attention. Sequential selection of stimuli simultaneously present in the visual scene is demonstrated by the model in the frequency domain in both a formal example and a real image.

Theta-alpha cross-frequency synchronization facilitates working memory control – a modeling study

Despite decades of research, the neural mechanism of central executive and working memory is still unclear. In this paper, we propose a new neural network model for the real-time control of working memory. The key idea is to consider separately the role of neural activation from that of oscillatory phase. Neural populations encoding different information would not confuse each other when the populations have different oscillatory phases. Depending on the current situation, relevant memories bind together through phase-locking between theta-frequency oscillation of a Central Unit and alpha-frequency oscillations of the relevant group of Memory Units. The Central Unit dynamically controls which Memory Units should be synchronized (and the encoded memory would be processed), and which units should be out of phase (the encoded memory is standby and would not be processed yet). Simulations of two working memory tasks are provided as examples. The model is in agreement with many recent experimental results of human scalp EEG analysis, which reported observations of neural synchronization and cross-frequency coupling during working memory tasks. This model offers a possible explanation of the underlying mechanism for these experiments.

Spiking neural network model for memorizing sequences with forward and backward recall

We present an oscillatory network of conductance based spiking neurons of Hodgkin-Huxley type as a model of memory storage and retrieval of sequences of events (or objects). The model is inspired by psychological and neurobiological evidence on sequential memories. The building block of the model is an oscillatory module which contains excitatory and inhibitory neurons with all-to-all connections. The connection architecture comprises two layers. A lower layer represents consecutive events during their storage and recall. This layer is composed of oscillatory modules. Plastic excitatory connections between the modules are implemented using an STDP type learning rule for sequential storage. Excitatory neurons in the upper layer project star-like modifiable connections toward the excitatory lower layer neurons. These neurons in the upper layer are used to tag sequences of events represented in the lower layer. Computer simulations demonstrate good performance of the model including difficult cases when different sequences contain overlapping events. We show that the model with STDP type or anti-STDP type learning rules can be applied for the simulation of forward and backward replay of neural spikes respectively.

Partial synchronization of neural activity and information processing

We study dynamics of neural activity in brain-inspired neural networks which comprise both low and high layers of information processing. Information propagates from the low layer which includes “peripheral neurons” (PNs), and the dynamics is controlled by the feedback from the higher layer of “central neurons” (CNs). We use the Hodgkin-Huxley type model to describe spike generation properties of neural elements. Synaptic connections are of excitatory and inhibitory type and some of them have fixed connection strengths and some are adjustable according to Hebbian type learning rule. The regime of partial synchronization between spiking activity of the CNs and PNs has been found. It is shown that PNs with higher firing rates are selected preferentially by the central neurons. In the case of local connections between PNs, we have found that local excitatory connections facilitate synchronization; while local inhibitory connections help distinguishing two groups of PNs with similar intrinsic frequencies. We hypothesize that the regime of partial synchronization can be used to simulate neural mechanisms of perception and attention. In particular, sequential selection of stimuli simultaneously present in the visual scene is demonstrated by the model which deals with a real image in the frequency domain.

Spiking neural network models for memorizing sequences with forward and backward recall

or> Meeting abstracts - A single PDF containing all abstracts in this Supplement is available here . http://www. biomedcentral.co m/content/pdf/14 71-2202 -10-S1-info.pdf

Selective Attention Model of Moving Objects

Tracking moving objects is a vital visual task for the survival of an animal. We describe oscillatory neural network models of visual attention with a central element that can track a moving target among a set of distracters on the screen. At the initial stage, the model forms focus of attention on an arbitrary object that is considered as a target. Other objects are treated as distracters. We present here two models: 1) synchronisation based AMCO model of phase oscillators and 2) spiking neural model which is based on the idea of resource-limited parallel visual pointers. Selective attention and the tracking process are represented by partial synchronization between the central unit and subgroup of peripheral elements. The simulation results are in overall agreement with the findings from psychological experiments: overlapping between target and distractor is the main source of error. Future investigations include the dependence between tracking performance and neuron frequency.

A Method to Deal with Prospective Risks at Home in Robotic Observations by Using a Brain-Inspired Model

Home robotics is a continuously growing field in academic research as well as commercial market. People are becoming more interested in advanced intelligent robots that can do housework and take care of children and elderly. A brain-inspired intelligent system is a possible solution to make the robot capable of learning and predicting risks at home. In order to solve difficult problems such as ambiguous situations and unclear causality, we propose a robotic system inspired from human working memory functions, which consists of an Event Map for storing observed information, and a Causality Map for representing causal relationships through supervised learning. The two maps couple together to enable the robot to evaluate various situations based on the appropriate context. More importantly, the Causality Map takes into account the dynamical aspects of physical attributes (e.g. the decreasing temperature of a hot pot). Our case studies showed that this is a satisfactory solution for predicting many risky situations at home.

How Difficult Is It for Robots to Maintain Home Safety? – A Brain-Inspired Robotics Point of View

The cognition-based human intelligence, that is driven by emotion and feeling will definitely change the robot learning, memory, attention and decision making mechanism. The aim of paper is to give in depth investigation on how the robot learning based on emotion and feeling will give a new dimension to its performance. It means that self-learning is not just dependent upon a logical brain and proper embodiment; rather a feedback in terms of emotion and feeling based on experience is required for self-learning. It is the feedback in the form of feeling and emotion that plays a vital role in a complete self-learning process and this makes the robot more human.