Qun Song

DENFIS: dynamic evolving neural-fuzzy inference system and its application for time-series prediction

This paper introduces a new type of fuzzy inference systems, denoted as dynamic evolving neural-fuzzy inference system (DENFIS), for adaptive online and offline learning, and their application for dynamic time series prediction. DENFIS evolve through incremental, hybrid (supervised/unsupervised), learning, and accommodate new input data, including new features, new classes, etc., through local element tuning. New fuzzy rules are created and updated during the operation of the system. At each time moment, the output of DENFIS is calculated through a fuzzy inference system based on m-most activated fuzzy rules which are dynamically chosen from a fuzzy rule set. Two approaches are proposed: (1) dynamic creation of a first-order Takagi-Sugeno-type fuzzy rule set for a DENFIS online model; and (2) creation of a first-order Takagi-Sugeno-type fuzzy rule set, or an expanded high-order one, for a DENFIS offline model. A set of fuzzy rules can be inserted into DENFIS before or during its learning process. Fuzzy rules can also be extracted during or after the learning process. An evolving clustering method (ECM), which is employed in both online and offline DENFIS models, is also introduced. It is demonstrated that DENFIS can effectively learn complex temporal sequences in an adaptive way and outperform some well-known, existing models.

NFI: a neuro-fuzzy inference method for transductive reasoning

This paper introduces a novel neural fuzzy inference method-NFI for transductive reasoning systems. NFI develops further some ideas from DENFIS-dynamic neuro-fuzzy inference systems for both online and offline time series prediction tasks. While inductive reasoning is concerned with the development of a model (a function) to approximate data in the whole problem space (induction), and consecutively-using this model to predict output values for a new input vector (deduction), in transductive reasoning systems a local model is developed for every new input vector, based on some closest to this vector data from an existing database (also generated from an existing model). NFI is compared with both inductive connectionist systems (e.g., MLP, DENFIS) and transductive reasoning systems (e.g., K-NN) on three case study prediction/identification problems. The first one is a prediction task on Mackey Glass time series; the second one is a classification on Iris data; and the last one is a real medical decision support problem of estimating the level of renal function of a patient, based on measured clinical parameters for the purpose of their personalised treatment. The case studies have demonstrated better accuracy obtained with the use of the NFI transductive reasoning in comparison with the inductive reasoning systems.

GA-parameter optimisation of evolving connectionist systems for classification and a case study from bioinformatics

The paper describes an algorithm for parameter optimisation of evolving connectionist systems (ECOS) in an offline processing mode. The algorithm is illustrated on a case study of a classification system that uses gene expression data to predict an outcome of a treatment of cancer disease.

Evolutionary computation for dynamic parameter optimisation of evolving connectionist systems for on-line prediction of time series with changing dynamics

The paper describes a method of using evolutionary computation technique for parameter optimisation of evolving connectionist systems (ECOS) that operate in an online, life-long learning mode. ECOS evolve their structure and functionality from an incoming stream of data in either a supervised-, of/and in an unsupervised mode. The algorithm is illustrated on a case study of predicting a chaotic time-series that changes its dynamics over time. With the on-line parameter optimisation of ECOS, a faster adaptation and a better prediction is achieved. The method is practically applicable for real time applications.

Personalized modeling based gene selection for microarray data analysis

This paper presents a novel gene selection method based on personalized modeling. Identifying a compact set of genes from gene expression data is a critical step in bioinformatics research. Personalized modeling is a recently introduced technique for constructing clinical decision support systems. In this paper we have provided a comparative study using the proposed Personalized Modeling based Gene Selection method (PMGS) on two benchmark microarray datasets (Colon cancer and Central Nervous System cancer data). The experimental results show that our method is able to identify a small number of informative genes which can lead to reproducible and acceptable predictive performance without expensive computational cost. These genes are of importance for specific groups of people for cancer diagnosis and prognosis.

Integrating regression formulas and kernel functions into locally adaptive knowledge-based neural networks: A case study on renal function evaluation

OBJECTIVE In many medical areas, there exist different regression formulas to predict/evaluate a medical outcome on the same problem, each of them being efficient only in a particular sub-space of the problem space. The paper aims at the development of a generic, incremental learning model that includes all available regression formulas for a particular prediction problem to define local areas of the problem space with their best performing formula along with useful explanation rules. Another objective of the paper is to develop a specific model for renal function evaluation using nine existing formulas. METHODS AND MATERIALS We have used a connectionist neuro-fuzzy approach and have developed a knowledge-based neural network model (KBNN) which incorporates and adapts incrementally several existing regression formulas and kernel functions. The model incorporates different non-linear regression functions as neurons in its hidden layer and adapts these functions through incremental learning from data in particular local areas of the space. More specifically, each hidden neural node has a pair of functions associated with it--one regression formula, that represents existing knowledge and one Gaussian kernel function, that defines the sub-space of the whole problem space, in which the formula is locally adapted to new data. All these functions are aggregated and changed through incremental learning. The proposed KBNN model is illustrated using a medical dataset of observed patient glomerular filtration rate (GFR) measurements for renal function evaluation. In this case study, the regression function for each cluster is selected by the model from nine formulas commonly used by medical practitioners to predict GFR. 441 GFR data vectors from 141 patients taken from 12 sites in Australia and New Zealand have been used as a case study experimental data set. RESULTS The proposed GFR prediction model, based on the proposed generic KBNN model, outperforms at least by 10% accuracy any of the individual regression formulas or a standard neural network model. Furthermore, we have derived locally adapted regression formulas to perform best on local clusters of data along with useful explanatory rules. CONCLUSION The proposed KBNN model manifests better accuracy then existing regression formulas or neural network models for renal function evaluation and extracts modified formulas that perform well in local areas of the problem space.

Transductive knowledge based fuzzy inference system for personalized modeling

This paper introduces a novel transductive knowledge based fuzzy inference system (TKBFIS) and its application for creating personalized models. In transductive systems a local model is developed for every new input vector, based on some closest data to this vector from the training data set. A higher-order TSK type fuzzy inference engine is applied in TKBFIS. Some existing formulas or equations, which are used to represent the knowledge and usually have a non-linear form, are taken as consequent parts of the fuzzy rules. The TKBFIS uses a gradient descent algorithm for its training. In this paper, the TKBFIS is illustrated with a case study of personalized modeling for renal function estimation of patients and the result is compared with other transductive or inductive methods.

Dynamic neural fuzzy inference system

This paper proposes an extension to the original offline version of DENFIS. The new algorithm, DyNFIS, replaces original triangular membership function with Gaussian membership function and use back-propagation to further optimizes the model. Fuzzy rules are created for each clustering centre based on the clustering outcome of evolving clustering method. For each test data, the output of DyNFIS is calculated through fuzzy inference system based on m-most activated fuzzy rules and these rules are updated based on back-propagation to minimize the error. DyNFIS shows improvement on multiple benchmark data and satisfactory result in NN3 forecast competition.

Evolving Connectionist Systems Based Role Allocation of Robots for Soccer Playing

For a group of robots (multi-agents) to complete a task, it is important for each of them to play a certain role changing with the environment of the task. One typical example is robotic soccer in which a team of mobile robots perform soccer playing behaviors. Traditionally, a robots role is determined by a closed-form function of a robots postures relative to the target which usually cannot accurately describe real situations. In this paper, the robot role allocation problem is converted to the one of pattern classification. Evolving classification function (ECF), a special evolving connectionist systems (ECOS), is used to identify the suitable role of a robot from the data collected from the robot system in real time. The software and hardware platforms are established for data collection, learning and verification for this approach. The effectiveness of the approach are verified by the experimental studies

Integrating Local and Personalised Modelling with Global Ontology Knowledge Bases for Biomedical and Bioinformatics Decision Support

A novel ontology based decision support framework and a development platform are described, which allow for the creation of global knowledge representation for local and personalised modelling and decision support. The main modules are: an ontology module; and a machine learning module. Both modules evolve through continuous learning from new data. Results from the machine learning procedures can be entered back to the ontology thus enriching its knowledge base and facilitating new discoveries. This framework supports global, local and personalised modelling. The latter is a process of model creation for a single person, based on their personal data and the information available in the ontology. Several methods for local and personalised modelling, both traditional and new, are described. A case study is presented on brain-gene-disease ontology, where a set of 12 genes related to central nervous system cancer are revealed from existing data and local profiles of patients are derived. Through ontology analysis, these genes are found to be related to different functions, areas, and other diseases of the brain. Two other case studies discussed in the paper are chronic disease ontology and risk evaluation, and cancer gene ontology and prognosis.