KIII Model Research Articles

An Improved Biomimetic Olfactory Model and Its Application in Traffic Sign Recognition

In human and other organisms’ perception, olfaction plays a vital role, and biomimetic olfaction models offer a pathway for studying olfaction. The most optimal existing biomimetic olfaction model is the KIII model proposed by Professor Freeman; however, it still exhibits certain limitations. This study aims to address these limitations: In the feature extraction stage, it introduces adaptive histogram equalization, Gaussian filtering, and discrete cosine transform methods, effectively enhancing and extracting high-quality image features, thereby bolstering the model’s recognition capabilities. To tackle the computational cost issue associated with solving the numerical solutions of neuronal dynamics equations in the KIII model, it replaces the original method with the faster Euler method, reducing time expenses while maintaining good recognition results. In the decision-making stage, several different dissimilarity metrics are compared, and the results indicate that the Spearman correlation coefficient performs best in this context. The improved KIII model is applied to a new domain of traffic sign recognition, demonstrating that it outperforms the baseline KIII model and exhibits certain advantages compared to other models.

A Novel Evaluation Strategy to Artificial Neural Network Model Based on Bionics.

With the continuous deepening of Artificial Neural Network (ANN) research, ANN model structure and function are improving towards diversification and intelligence. However, the model is more evaluated from the pros and cons of the problem-solving results and the lack of evaluation from the biomimetic aspect of imitating neural networks is not inclusive enough. Hence, a new ANN models evaluation strategy is proposed from the perspective of bionics in response to this problem in the paper. Firstly, four classical neural network models are illustrated: Back Propagation (BP) network, Deep Belief Network (DBN), LeNet5 network, and olfactory bionic model (KIII model), and the neuron transmission mode and equation, network structure, and weight updating principle of the models are analyzed qualitatively. The analysis results show that the KIII model comes closer to the actual biological nervous system compared with other models, and the LeNet5 network simulates the nervous system in depth. Secondly, evaluation indexes of ANN are constructed from the perspective of bionics in this paper: small-world, synchronous, and chaotic characteristics. Finally, the network model is quantitatively analyzed by evaluation indexes from the perspective of bionics. The experimental results show that the DBN network, LeNet5 network, and BP network have synchronous characteristics. And the DBN network and LeNet5 network have certain chaotic characteristics, but there is still a certain distance between the three classical neural networks and actual biological neural networks. The KIII model has certain small-world characteristics in structure, and its network also exhibits synchronization characteristics and chaotic characteristics. Compared with the DBN network, LeNet5 network, and the BP network, the KIII model is closer to the real biological neural network.

Research on an olfactory neural system model and its applications based on deep learning

The idea of constructing the biological neural system model as realistic as possible can not only provide a new artificial neural network (ANN), but also offer an effective object to study biological neural systems. As a very meaningful attempt about the idea, a bionic model of olfactory neural system, KIII model, is introduced in this paper. There are the unique characteristics of KIII model different from those of general ANNs. The KIII model realistically simulates the structure of the real olfactory neural system and the process of odor molecules gradually transformed by the core components of the olfactory system including olfactory receptor, olfactory bulb and olfactory cortex. The neuron model of the KIII model is constructed and optimized based on neurophysiological experimental data and accurately reflects the response of olfactory neurons to odor stimulation. In particular, the noise introduced to KIII model can further improve the performance of the model. In addition, the KIII model is analyzed based on the idea of deep learning. The qualitative analysis shows that there are obvious similarities between the KIII model and the deep learning model. Furthermore, with the epileptic electroencephalograph (EEG) recognition task, two groups of experiments are designed to comprehensively analyze the performance of the KIII model. In the first group of experiments, a typical pattern recognition experiment with feature extraction stage is shown. The features of epileptic EEG were extracted based on Empirical Mode Decomposition (EMD), and the KIII model was used as a classifier. The experimental results show that the KIII model only needs a small number of iterations to memorize different modes and has a high recognition rate, over 91%. In the second group of experiments, a direct recognition experiment without feature extraction stage is shown. The original epileptic EEG signals as KIII model inputs directly were recognized. The experimental results show that there is still an excellent performance in the KIII model, over 96%, and the recognition result is similar to the characteristics of the deep learning model. The theoretical analysis and experimental results prove that KIII model with the idea of deep learning is an excellent bionic model of olfactory neural system and gets a good balance between high bionics and good performance, which is a good reference for related research.

A New Classification Method for Stored Grain Insect Infestation Using KIII and SVM Based Electronic Nose

Insect infestation is a common problem for stored grain. In this paper, a novel pattern recognition approach combining an olfactory neural network entitled KIII with support vector machine (SVM) is proposed and used in conjunction with an electronic nose to generate recognition models. Using this approach, feature vectors are firstly processed by KIII model which stimulates information processing function of olfactory bulb, and then classified by SVM. Through optimization of SVM model parameters, the data are mapped into high dimension space and the stored wheat samples with different degrees of insect damage are distinguished successfully. The experimental results demonstrate that the proposed method can achieve up to 100% classification rate and significantly outperforms the conventional KIII-Minimum Euclidean Distance Classifier.

Handwriting character classification using Freeman’s olfactory KIII model

KIII model is an olfactory model proposed by W. J. Freeman referring to a physiological structure of mammal's olfactory system. The KIII model has been applied to kinds of pattern recognition systems, for example, electronic nose, tea classification, etc. However, the dynamics of neurons in the KIII model is given by Hodgkin-Huxley's second-order differential equation and it consumes a very high computation cost. In this paper, we propose a simplified dynamics of chaotic neuron instead of the Hodgkin-Huxley dynamics at first, and secondly, we propose to use Fourier transformation with high resolution capability to extract features of time series behaviors of internal states of M1 nodes in KIII model instead of the conventional standard deviation method. Furthermore, paying attention to the point that human brain does visual processing as same as olfactory processing in the sense of information processing, a handwriting image recognition problem is treated as a new application field of KIII model. Through the computer simulation of the handwriting character classification, it is shown that the proposed method is useful by the comparison of experiment results with both computation time and recognition accuracy.

Increasing the separability of chemosensor array patterns with Hebbian/anti-Hebbian learning

The olfactory bulb is able to enhance the contrast between odor representations through a combination of excitatory and inhibitory circuits. Inspired by this mechanism, we propose a new Hebbian/anti-Hebbian learning rule to increase the separability of sensor-array patterns in a neurodynamics model of the olfactory system: the KIII. In the proposed learning rule, a Hebbian term is used to build associations within odors and an anti-Hebbian term is used to reduce correlated activity across odors. The KIII model with the new learning rule is characterized on synthetic data and validated on experimental data from an array of temperature-modulated metal-oxide sensors. Our results show that the performance of the model is comparable to that obtained with Linear Discriminant Analysis (LDA). Furthermore, the model is able to increase pattern separability for different concentrations of three odorants: allyl-alcohol, tert-butanol, and benzene, even though it is only trained with the gas sensor response to the highest concentration.

A chaotic neural network mimicking an olfactory system and its application on image recognition

Based on the research of a biological olfactory system, a novel chaotic neural network model — K set model has been established. This chaotic neural network not only simulates the real brain activity of an olfactory system, but also presents a novel chaotic concept for signal processing and pattern recognition. The characteristics of the K set models are investigated and show that a KIII model can be used for image pattern classification.

Habituation in the kiii olfactory model with chemical sensor arrays

This paper presents a novel combination of chemical sensors and the KIII model for simulating mixture perception with a habituation process triggered by local activity. Stimuli are generated by partitioning feature space with labeled lines. Pattern completion is demonstrated through coherent oscillations across granule populations using experimental odor mixtures.

On the constructive role of noise in stabilizing itinerant trajectories in chaotic dynamical systems.

This work aims at studying dynamical models of neural networks, which exhibit phase transitions between states of various complexities. We use the biologically motivated KIII model, which has demonstrated excellent performance as a robust dynamical memory device. KIII is a high-dimensional dynamical system with extremely fragmented boundaries between limit cycles, tori, fixed points, and chaotic attractors. We study the role of additive noise in the development of itinerant trajectories. Noise not only stabilizes aperiodic trajectories, but there is an optimum noise level with highly itinerant behavior. We speculate that the previously found optimum classification performance of KIII as a function of the noise level, also identified as chaotic resonance, is related to chaotic itinerant oscillations among various ordered states.

Basic principles of the KIV model and its application to the navigation problem.

EEG measurements indicate the presence of common-mode, coherent oscillations shared by multiple cortical areas. In previous studies the KIII model has been introduced, which interprets the experimental observations as nonlinear, spatially distributed dynamical oscillations of locally coupled neural populations. KIII can account for the fast and robust classification and pattern recognition in sensory cortices. In order to describe selection of action, planning, and spatial orientation functions, in this paper we expand KIII into the KIV model. KIV approximates the operation of the corticostriatal-hippocampal system. KIV consists of three KI, eight KII and three KIII components, including sensory and cortical systems, as well as the hippocampus, amygdala, and the septum. KIV implements various types of dynamic neural activities. The neural activity patterns determine the emergence of global spatial encoding to implement the orientation function of a simulated animal. Our results indicate the mechanisms, which we believe support the generation of cognitive maps in the hippocampus based on the sensory input-based destabilization of cortical spatio-temporal patterns. In this paper, we describe the conceptual design of the KIV model. We outline the biological background and motivation of the basic principles that are applied to design the KIV computational model. We use the KIV model to explain how the hippocampal neural circuitry functions are constructed and controlled by the corticostriatal-hippocampal loops, supplemented with specific subcortical units. In the second part, we implement these principles using the example of the hippocampal formation as a KIII unit. We demonstrate the learning and navigation principles using the Evolving Multi-module Mobile Agent (EMMA) in 2D software environment.

Pattern completion through phase coding in population neurodynamics

This article presents an alternative phase coding mechanism for Freeman's KIII model of population neurodynamics. Motivated by experimental evidence that supports the existence of a neural code based on synchronous oscillations, we propose an analogy between synchronization in neural populations and phase locking in KIII channels. An efficient method is proposed to extract phase differences across granule channels from their state-space trajectories. First, the scale invariance of the KIII model with respect to phase information is established. The phase code is then compared against the conventional amplitude code in terms of their bit-wise and across-fiber pattern recovery capabilities using decision-theoretic principles and a Hamming-distance classifier. Graph isomorphism in the Hebbian connections is exploited to perform an exhaustive evaluation of patterns on an 8-channel KIII model. Simulation results show that phase information outperforms amplitude information in the recovery of incomplete or corrupted stimuli.

The KIV model—nonlinear spatio-temporal dynamics of the primordial vertebrate forebrain

EEG measurements indicate the presence of common-mode, coherent oscillations in various cortical areas. In previous studies the KIII model has been introduced, which interprets the experimental observation as nonlinear, spatially distributed dynamical oscillations of coupled neural populations. In this paper we combine multiple KIII sets into the KIV model, which approximates the operation of the basic vertebrate forebrain together with the basal ganglia and motor systems. This paper outlines a summary description of the essential components of the KIV model, as the basis for future modeling of their cooperative dynamics guided by analysis of multichannel EEG in animals and humans.

Using singular-value decompositions to classify spatial patterns generated by a nonlinear dynamic model of the olfactory system

Freeman's KIII model, a nonlinear dynamic software version of the olfactory system, has been applied for classifying drugs via their preprocessed IR spectra. In this paper, singular-value decomposition (SVD), rather than Euclidean distance classifiers, are successfully introduced to analyze the generated spatial amplitude modulation patterns. Results are compared and discussed.

Fragmented attractor boundaries in the KIII model of sensory information processing: A potential evidence of Cantor encoding in cognitive processes

Spatio-temporal neuro-dynamics is a quickly developing field of brain research and Tsuda's work is a significant contribution toward establishing theoretical foundations in this area. It is conceivable that the fragmented attractor landscapes and dynamical memory patterns identified earlier in various K-sets are biologically plausible manifestations of attractor ruins, chaotic itinerancy, and Cantor encoding as applied to sensory information processing.

Interpretation of analytical patterns from the output of chaotic dynamical memories.

A fundamental study has been carried out to use a nonlinear dynamic system, Freeman's KIII model, for pattern recognition in analytical chemistry. Filtering of the output of the neuron activity traces in the gamma range has been studied by applying a fast Fourier transform filter. Different possibilities for interpretation of the system output have been investigated. Preliminary results are demonstrated for classification of drugs based on their spectra in the mid infrared range.

Biologically Modeled Noise Stabilizing Neurodynamics for Pattern Recognition

We present a distributed KIII model for the olfactory neural system. Low-level Gaussian noise is introduced to the receptors and anterior olfactory nucleus, which biologically models the peripheral and central sources of noise. The additive noise numerically makes the model stable and robust in respect to repeated input-induced state transitions, while improving the simulations of EEG potentials and multiunit activity from the olfactory system. This hybrid dynamics generates a 1/f aperiodic state, which provides an unpatterned basal state for every module to stay in while there is no significant stimulus. Any external input may guide the system to a certain patterned state. The mechanism is fast, fully parallel, under modulatory control, and flexible in absorbing new patterns from unpredictable environments.

Taming chaos: stabilization of aperiodic attractors by noise [olfactory system model

A model named of the olfactory system contains an array of 64 coupled oscillators simulating the olfactory bulb (OB), with negative and positive feedback through low-pass filter lines from single oscillators simulating the anterior olfactory nucleus (AON) and prepyriform cortex (PC). It is implemented in C to run on Macintosh, IBM, or UNIX platforms. The output can be set by parameter optimization to point, limit cycle, quasi-periodic, or aperiodic (presumably chaotic) attractors. The first three classes of solutions are stable under variations of parameters and perturbations by input, but they are biologically unrealistic. Chaotic solutions simulate the properties of time-dependent densities of olfactory action potentials and EEGs, but they transit into the basins of point, limit cycle, or quasi-periodic attractors after only a few seconds of simulated run time. Despite use of double precision arithmetic giving 64-bit words, the KIII model is exquisitely sensitive to changes in the terminal bit of parameters and inputs. The global stability decreases as the number of coupled oscillators in the OB is increased, indicating that attractor crowding reduces the size of basins in the model to the size of the digitizing step (/spl sim/10/sup -16/). Chaotic solutions having biological verisimilitude are robustly stabilized by introducing low-level, additive noise from a random number generator at two biologically determined points: rectified, spatially incoherent noise on each receptor input line, and spatially coherent noise to the AON, a global control point receiving centrifugal inputs from various parts of the forebrain. Methods are presented for evaluating global stability in the high dimensional system from measurements of multiple chaotic outputs. Ranges of stability are shown for variations of connection weights (gains) in the KIII model. The system is devised for pattern classification.

COMPARISON OF EEG TIME SERIES FROM RAT OLFACTORY SYSTEM WITH MODEL COMPOSED OF NONLINEAR COUPLED OSCILLATORS

The so-called “spontaneous,” background, or “basal” time series of electroencephalographic (EEG) data from rat olfactory cortex are compared with those generated by a model of the ol-factory system, denoted KIII, composed of nonlinear coupled second order ODEs with feedback delay. Each ODE (KO) represents a neural mass having either excitatory or inhibitory output to the ODEs to which it connects. Connected mutually inhibitory or mutually excitatory pairs represent populations of similar neurons interacting to form one structure (KI set). A connected set (KII set) of excitatory and inhibitory KI sets (KIe and KIi) represents the activity of a gross structure in the olfactory system (the olfactory bulb—OB, anterior olfactory nucleus—AON, and prepyriform cortex—PPC or PC). The set of KII sets connected with dispersion-delay feedback pathways (KIII set) form a model of the first central stage of olfactory perceptual processing upon olfactory stimulation, with transition on receptor input from a self-sustained basal state of activity to an induced burst. Both basal and stimulated states are studied using amplitude histograms, power spectra, attractor reconstruction, visual inspection of traces, and correlation. A comparison shows that the two systems are similar, having broad spectra, correlation functions that rapidly go to zero, and gaussian amplitude histograms. Activity is nearly periodic and higher amplitude in the excited state and aperiodic, lower amplitude in the basal state with close to 1/f falloff. Cross spectra between OB and PPC show wide frequency bands of coherent high frequency activity in both the model and rat. The excited state nearly periodic frequencies differ between the KIII model and rat (30–60 Hz in KIII, 50–100 Hz in rat EEG), because the KIII model was developed for the cat and the rabbit, which have lower frequencies than the rat. These benchmarks may be used to evaluate the physiological validity of the output produced by a proposed model, as all components of the system and its connectivity were derived from or evaluated by physiological data.