Geok See Ng

• BMath(Hons. Comp. Sci), MEng (Elect. Eng.), PhD(Comp. Eng.)
• Professor at SUTD, CIC

Interpretabilty is one of the desired characteristics in various classification task. Rule-based system and fuzzy logic can be used for interpretation in classification. The main drawback of rule-based system is that it may contain large complex rules for classification and sometimes it becomes very difficult in interpretation. Rule reduction is al...

Due to their aptitude in both accurate data processing and human comprehensible reasoning, neural fuzzy inference systems have been widely adopted in various application domains as decision support systems. Especially in real-world scenarios such as decision making in financial transactions, the human experts may be more interested in knowing the c...

In this paper, entropy term is used in the learning phase of a neural network. As learning progresses, more hidden nodes get into saturation. The early creation of such hidden nodes may impair generalisation. Hence entropy approach is proposed to dampen the early creation of such nodes. The entropy learning also helps to increase the importance of...

Several methods have been proposed by researchers to detect cyber attacks in Cyber-Physical Systems (CPSs). This paper proposes a comprehensive approach for conducting experiments to assess the effectiveness of such methods in the context of a robot (Amigobot) that includes both cyber and physical components. The proposed approach includes a method...

Effectiveness of seven methods for detecting stealthy attacks on Cyber Physical Systems (CPS) was investigated using an experimental study. The Amigobot robot was used as the CPS. The experiments were conducted in simulation as well as on the physical robot. Three types of stealthy attacks were implemented: surge, bias, and geometric. Two variation...

Bank failure prediction is an important study for regulators in the banking industry because the failure of a bank leads to devastating consequences. If bank failures are correctly predicted, early warnings can be sent to the responsible authorities for precaution purposes. Therefore, a reliable bank failure prediction or early warning system is in...

An experiment was conducted to investigate the effectiveness of the Cumulative Sum (CUSUM) approach for detecting cyber attacks on Cyber Physical Systems (CPS). The Amigobot robot was used as the CPS in this study. Three types of stealthy attacks were considered, namely, surge, bias, and geometric. While a similar study has been reported earlier us...

Hedging against potential market risk and maximizing profit from stock trading are some of the driving forces for the development of financial decision support system. Although presently many types of financial decision support available, there are still a lot of room for improvement. Hence, there is a significant importance to identify and develop...

Artificial ventilation is a crucial supporting treatment for Intensive Care Unit. However, as the ventilator control becomes increasingly more complex, it is non-trivial for less experienced clinicians to control the settings. In this paper, the novel Hebbian based Rule Reduction (HeRR) neuro-fuzzy system is applied to model this control problem fo...

Medical decision making is often linked to survival and wellbeing of patients. As such, it is paramount for clinical decision support system to not only provides human-understandable explanation, but also human-relatable reasoning. A human decision making model Complementary Decision Making System (CDMS) is proposed. CDMS is based on complementary...

This paper describes the programming of a reconfigurable environment to handle inverse dynamics computation for robotics control. Instruction parallelism/pipelining and avoidance of carry propagation while evaluating a lengthy sequence of sum of products is proposed. The difficulties of programming a reconfigurable platform are overcome by defining...

As an extension of the traditional normalized radial basis function (NRBF) model, the extended normalized RBF (ENRBF) model was proposed by Xu [RBF nets, mixture experts, and Bayesian Ying-Yang learning, Neurocomputing 19 (1998) 223–257]. In this paper, we perform a supplementary study on ENRBF with several properly designed experiments and some fu...

Early detection is paramount to reduce the high death rate of ovarian cancer. Unfortunately, current detection tool is not sensitive. New techniques such as deoxyribonucleic acid (DNA) micro-array and proteomics data are difficult to analyze due to high dimensionality, whereas conventional methods such as blood test are neither sensitive nor specif...

The approach described in this paper uses an array of Field Programmable Gate Array (FPGA) devices to implement a fault tolerant hardware system that can be compared to the running of fault tolerant software on a traditional processor. Fault tolerance is achieved is achieved by using FPGA with on the fly partial programmability feature. Major consi...

This paper proposes a novel hybrid intelligent system denoted as genetic algorithm and rough set incorporated neural fuzzy inference system (GARSINFIS). Its network structure dynamically changes along with the evolving genetic algorithm based rough set clustering (GARSC) technique. When input data set is applied, only the most essential information...

Clinical Decision Support System (CDSS) is a promising tool that can alleviate the high medical error rate. However, most of the CDSS are not adopted in clinical settings due to the lack of trust amongst the physicians. Thus, the development of CDSS should cater to the psychological need of physicians. One major issue preventing the wide acceptance...

In the banking industry, it is highly desirable to identify potential bank failure or high-risk banks. Successful early warning systems (EWS) would provide capabilities to avoid adverse financial repercussions and a massive bail out costs for the failing banks. Very often, these failures are due to financial distress. Various traditional statistica...

A computational intelligent system that models the human cognitive abilities may promise significant performance in problem learning because human is effective in learning and problem solving. Functionally modelling the human cognitive abilities not only avoids the details of the underlying neural mechanisms performing the tasks, but also reduces t...

Early detection of breast cancer is the key to improve survival rate. Thermogram is a promising front-line screening tool as it is able to warn women of breast cancer up to 10 years in advance. However, analysis and interpretation of thermogram are heavily dependent on the analysts, which may be inconsistent and error-prone. In order to boost the a...

Humans are effective in learning and problem solving. Thus, a computational intelligence method that models human being not only increases in credibility, but also increases in performance. One of the reasons behind effective learning is that the brain employs an effective memory consolidation mechanism. Hence, a method that functionally models the...

Motion planning is an essential task for humanoid robots. However, it is still very challenging to obtain good motion performance in humanoid motion planning, because of its high DOFs (degree of freedoms), variable mechanical structure and nonlinearity. In humanoid motion planning, the motion performance can be given only after one whole cycle moti...

Decision making is important in problem solving. Human makes decisions everyday rather effectively. This suggests that functionally model the decision making process could potentially offer a competent decision support system. Hence, a Complementary Decision Making System (CDMS), which functionally models the neural decision making in human visual...

Imbalanced dataset is a phenomenon seen in many real life applications, especially in medical field. The conventional computational intelligence algorithms cannot effectively handle the imbalanced data because they are designed for balanced data distribution. Complementary learning fuzzy neural network is proposed as one of the approach for learnin...

There are two important issues in neuro-fuzzy modeling: (1) interpretability—the ability to describe the behavior of the system in an interpretable way—and (2) accuracy—the ability to approximate the outcome of the system accurately. As these two objectives usually exert contradictory requirements on the neuro-fuzzy model, certain compromise has to...

Genetic complementary learning (GCL) is a biological brain-inspired learning system based on human pattern recognition, and genes selection process. It is a confluence of the hippocampal complementary learning and the evolutionary genetic algorithm. With genetic algorithm providing the possibility of optimal solution, and complementary learning pro...

Among various biometric verification systems, fingerprint verification is one of the most reliable and widely accepted. One essential part of fingerprint verification is the minutiae extraction system. Most existing minutiae extraction methods require image preprocessing or post processing resulting in additional complex computation and time. Hence...

Time series prediction is traditionally handled by linear models such as autoregressive and moving-average. However they are unable to adequately deal with the non-linearity in the data. Neural networks are non-linear models that are suitable to handle the non-linearity in time series. When designing a neural network for prediction, two critical fa...

Autonomous exploration by a team of robots has many important applications in rescue operations, clearing of mine fields and other military applications, and even space exploration. With limited range of sensors robots have to divide exploration tasks among themselves working under multiple constraints. An optimal covering of two-dimensional area b...

As powerful probabilistic models for optimal policy search, partially observable Markov decision processes (POMDPs) still suffer from the problems such as hidden state and uncertainty in action effects. In this paper, a novel approximate algorithm -Genetic algorithm based Q-Learning (GAQ-Learning), is proposed to solve the POMDP problems. In the pr...

Without Abstract

Ovarian cancer is a major cause of deaths worldwide. As a result, women are not diagnosed until the cancer has advanced to later stages. Accurate prognosis is required to determine the suitable therapeutic decision. Since abnormalities of hemostasis and increased risk of thrombosis are observed in cancer patient, assay involving hemostatic paramete...

There have been many applications in the area of handwritten character recognition. Over the last decade much research has gone into developing algorithms to accurately convert captured images of handwriting to text. At the same time, research into neuro fuzzy classification models has proven to solve many complex problems. In this paper, Adaptive...

Wavelet decomposition reconstructs a signal by a series of scaled and translated wavelets. Incorporating discrete wavelet decomposition theory with neural network techniques, wavelet networks have recently emerged as a powerful tool for many applications in the field of signal processing, such as data compression and function approximation. In this...

Traffic prediction is a critical element in traffic control today. With the increase of transportation, an effective traffic prediction allows to prevent traffic problems. This research aims to propose a novel approach to traffic prediction using Ying-Yang fuzzy cerebellar model articulation controller (YY-FCMAC). The model is motivated from the fa...

Uncertainty exists in various complex problems. Yet, human is able to effectively handle these uncertainties and makes appropriate decision. Thus, modeling of human uncertainty process should improve the performance of learning system in uncertain environment. A mechanism for human uncertainty monitoring is the broad and narrow generalization in ca...

Snake neurotoxins are important experimental tool in pharmacological research. Over the years, the number of snake neurotoxin sequences identified is increasing at a very fast pace. However, only a small portion of them are experimentally characterized from more than 200,000 variants estimated to exist in nature. In this paper, we report a systemat...

Proteomics is a promising microbiology technology that will play a critical role in future oncology and drug discovery. The protein expression pattern is useful in clinical diagnosis. However, the protein expression pattern is high-dimensional and hence, renders manual analysis difficult. Thus, computational intelligence method is often applied wit...

Artificial ventilation is a crucial treatment to the patients in Intensive Care Unit. However, as the ventilator increasingly becomes more complex, it is not easy for less experienced clinicians to control the settings. The objective of the paper is to model the FiO2 settings by clinician, using a neuro-fuzzy hybrid system. Two important issues, th...

As computational power of modern computer increases exponentially, more efficient computerized solutions are possible for complex real world applications. However, the solutions are usually not interpretable to human beings such as the opaqueness of traditional neural networks. In this paper, we propose a fuzzy neural network that is empowered by g...

Accurate prediction of the translation initiation sites (TIS) in eukaryotes is paramount for better understanding of the translation process, gene structure, as well as protein coding, and for more reliable amino acid prediction, etc. However, detecting TIS is not a simple task. Hence, computational biology is adopted to assist in the detection. Un...

Hemostatic parameters or parameters related to blood clotting are useful for diagnosis and prognosis. This is because abnormalities in hemostatic parameters are observed in various diseases. However, these parameters vary among localities and individuals. Thus, computational intelligent tool is required to aid the diagnosis and prognosis using hemo...

Proteomics is a promising microbiology technology that will play a critical role in future oncology and drug discovery. The protein expression pattern is useful in clinical diagnosis. However, the protein expression pattern is high-dimensional and hence, renders manual analysis difficult. Thus, computational intelligence method is often applie...

Artificial ventilation is a crucial treatment to the patients in Intensive Care Unit. However, as the ventilator increasingly becomes more complex, it is not easy for less experienced clinicians to control the settings. The objective of the paper is to model the FiO2 settings by clinician, using a neuro-fuzzy hybrid system. Two important issue...

A key property of Chinese characters is that they are composed of fundamental parts called radicals. In this paper, a method to recognise (offline) the radicals of handwritten Chinese characters is proposed that is an extension of the authors' previous work based on active shape modelling. Three stages are involved: a set of example radicals is fir...

Fuzzy logic allows mapping of an input space to an output space. The mechanism for doing this is through a set of IF-THEN statements, commonly known as fuzzy rules. In order for a fuzzy rule to perform well, the fuzzy sets must be carefully designed. A major problem plaguing the effective use of this approach is the difficulty of automatically and...

Speaker authentication has been developed rapidly in the last few decades. This research work attempts to extract the hidden features of human voice that is able to simulate human auditory system characteristics in speaker authentication. The hidden features are then presented as inputs to a Multi-Layer Perceptron Neural Network and Generic Self-or...

Early detection and accurate staging of ovarian cancer are the keys to improving survival rate. However, at present there is no single diagnosis modality that is sufficiently sensitive. DNA microarray analysis is an emerging technique that has potential for ameliorating the hardship in early detection and staging of ovarian disease. However, microa...

Traditional technical analysis for stock market prediction is error-prone, especially for multiyear trend prediction. Hence, computational intelligence provides an attractive alternative. Among the plethora of methods, statistics and artificial neural network are the most popular. However, they are black boxes that are not interpretable. Genetic co...

We present a novel geometric approach to the popular post nonlinear (PNL) BSS problem. A PNL mixing system includes two stages: a linear mixing followed by a nonlinear transformation. In our method, the process to linearize the nonlinear observed signals, the most critical task in PNL model, is carried out by a geometric transformation. The basic i...

Two important issues when constructing a neural network (NN) for time series prediction: proper selection of (1) the input dimension and (2) the time delay between the inputs. These two parameters determine the structure, computing complexity and accuracy of the NN. This paper is to formulate an autonomous data-driven approach to identify a parsimo...

DNA microarray is an emerging technique in ovarian cancer diagnosis. However, very often, microarray data is ultra-huge and difficult to analyze. Thus, it is desirable to utilize fuzzy neural network (FNN) approach for assisting the diagnosis and analysis process. Amongst FNN, complementary learning FNN is able to rapidly derive fuzzy sets and form...

We propose a novel architecture of neuro-fuzzy system called modified self-organizing Takagi-Sugeno-Kang fuzzy neural network (MS-TSKfnn) that uses ART-like clustering called discrete incremental clustering (DIC). The network is able to handle online data input with significant high performance. Its ability of entirely self-organizing to form the n...

The existing Self-Organizing Takagi Sugeno Kang Fuzzy Neural Networks (S-TSKfnn) structure uses virus infection clustering (VIC) method to generate fuzzy rules. In this paper, we propose a novel architecture called Modified S-TSKfnn (MS-TSKfnn) that uses ART-like clustering called discrete incremental clustering (DIC). By doing so, MS-TSKfnn is abl...

Almost all attempts made to improve the quality of speech that has been corrupted by additive noise would require the use of noise reference. The inaccuracy of estimating the noise reference in the latter system could cause additional musical artifacts being introduced at the output of the processed speech. These musical artifacts could be unbearab...

Taking advantage of both the scaling property of wavelets and the high learning ability of neural networks, wavelet networks have recently emerged as a powerful tool for many applications in the field of signal processing, such as data compression and function approximation. It is an implementation of wavelet transform, decomposing a signal into a...

The evolution of the Internet has linked the entire world together and revolutionized the way people communicate and exchange information. Wireless application protocol (WAP) further bridges the gaps by allowing people to link to the Internet through a mobile device. This paper introduces the concepts, and presents the methodology for the implement...

Handwritten digits classification has many useful applications. This has prompted decades of research into algorithms to produce an effective system of classifying handwritten images into text. Image processing and feature extraction play a large role in this process. An intelligent system is one, which is taught, and one, which uses this learning...

Handwritten digits classification has many useful applications. This has prompted decades of research into algorithms to produce an effective system of classifying handwritten images into text. Image processing and feature extraction play a large role in this process. An intelligent system is one, which is taught, and one, which uses this learning...

Minutia matching is the most popular approach to fingerprint recognition. We analyzed a novel fingerprint feature named adjacent orientation vector, or AOV, for fingerprint matching. In the first stage, AOV is used to find possible minutiae pairs. Then one minutiae set is rotated and translated. This is followed by a preliminary matching to ensure...

Handwritten digits classification has many useful applications. This has prompted decades of research into algorithms to produce an effective system of classifying handwritten images into text. Image processing and feature extraction play a large role in this process. An intelligent system is one, which is taught and uses its learning for classific...

In fuzzy neural network systems, fuzzy membership functions play a key role in making the fuzzy sets organize the input data knowledge in an appropriate and representative manner. Earlier clustering techniques exploit some uniform, convex algebraic functions, such as Gaussian, triangular or trapezoidal to represent the fuzzy sets. However, due to t...

Sensitivity is initially investigated for the construction of a network prior to its design. Sensitivity analysis applied to network pruning seems particularly useful and valuable when network training involves a large amount of redundant data. This paper proposes a novel learning algorithm for the construction of radial basis function (RBF) classi...

In this paper, entropy is a term used in the learning phase of a neural network. As learning progresses, more hidden nodes get into saturation. The early creation of such hidden nodes may impair generalisation. Hence an entropy approach is proposed to dampen the early creation of such nodes by using a new computation called entropy cycle. Entropy l...

Fax machines have become a ubiquitous equipment in today's office environment due to its ability to send information around with speed and convenience. Increasingly many personal computers nowadays are able to play the role of a fax machine by adding in a fax card and the relevant software. This paper proposes a Fax Management System employing char...

Currently, not many attempts are made to use neural-fuzzy inference system for recognizing primitive features of an input image. The objective of this paper is to propose a method of feature extraction so as the features obtained can be trained in a novel neural-fuzzy inference system called POP-CHAR. Common features of digit characters are extract...

This paper proposes active handwriting models, in which kernel principal component analysis is applied to capture nonlinear handwriting variations. In the recognition phase, the chamfer distance transform and a dynamic tunneling algorithm (DTA) are employed to search for the optimal shape parameters. The proposed methodology is successfully applied...

A type of network called the Contender Network (CN) was earlier proposed by Ng, Erdogan and Ng (1995). A classification algorithm is used to assign weighted vote in a monotonically decreasing function of the rank in CN. Modification to the CN classification algorithm known as the conscience algorithm is presented. However, a new problem is encounte...

. In this paper, an additional entropy penalty term is used to steer the direction of the hidden node's activation in the process of learning. A state with minimum entropy means that most nodes are operating in the non-linear zones (i.e. saturation zones) near the extreme ends of the Sigmoid curve. As the training proceeds, redundant hidden nodes'...

An entropy penalty term is used to steer the direction of the hidden node's activation in the process of learning. A state with minimum entropy means that nodes are operating near the extreme values of the Sigmoid curve. As the training proceeds, redundant hidden nodes' activations are pushed towards their extreme value, while relevant nodes remain...

The problem of occlusion in a two-dimensional scene introduces errors into many existing vision algorithms that cannot be resolved. Occlusion occurs where two or more objects in a given image touch or overlap one another. Since occlusion will be present in all but the most constrained environment, the recognition of partially occluded objects is im...

The objective of this paper is to show that a combination of votes from various pattern classifiers is better than a single vote from each individual classifier. A proposed support function is used in the combination of votes. The combination of outputs is motivated by the fact that decisions made by teams are generally better than those made by in...

In this paper, data equalisation is applied to output nodes of individual classifier in a multi-classifiers system such that the average difference of the output activation values is smaller. This helps in overall competitiveness of the output nodes of individual classifier. This will then improve the accuracy rate of a combined classifier which ag...

One way of obtaining better recognition result is to have multi-classifier systems. The problem of multi-classifier systems is the lack of competitiveness which degrade the performance of the final output. A data equalization method is proposed to increase the competitiveness of the output activation values of the individual classifier in a multi-c...

This article illustrates how current assessment practice can be modified and implemented to match course objectives and promote active learning, and the discussion is based on the second year In‐House Practical Training (IHPT) short course in the School of Mechanical and Production Engineering (MPE), Nanyang Technological University (NTU), Singapor...

Neural networks have been used in many problems such as character recognition, time series forecasting and image coding. The generalisation of the network depends on its internal structure. Network parameters should be set correctly so that data outside the class will not be overfitted. One mechanism to achieve an optimal neural network structure i...

This paper reports a study of the motivation and learning behaviour of first year computer engineering students at the School of Applied Science, Nanyang Technological University, Singapore. The Study Process Questionnaire (SPQ) (Biggs, 1987) is used to conduct the survey. The results of the survey are analyzed and recommendations are made for pros...

The objective of this article is to propose a highly effective computer-based teaching methods in an image-processing course. The image-processing course is heavily mathematically oriented. With the multimedia technology, it is easy to implement an innovative teaching methods using computer to attract student attention in the image-processing cours...

This article illustrates how current assessment practice can be modified and implemented to match course objectives and promote active learning, and the discussion is based on the second year In-House Practical Training (IHPT) short course in the School of Mechanical and Production Engineering (MPE), Nanyang Technological University (NTU), Singapor...

A new measure based on hidden-output node activation is proposed for measuring the relevance of hidden nodes in a neural network. The concept has been successfully applied for pruning in several classification problems. The experiments indicate that redundant nodes are pruned down resulting in optimal network topologies. The measure has been compar...

The Nanyang Technological University admits predominantly two large groups of entrants annually for its computer engineering undergraduate course, namely, junior college graduates with GCE ‘A’ levels and polytechnic graduates with Technician Diplomas. This study is aimed at assessing these engineering undergraduates’ common difficulties and identif...

A new sequential thinning algorithm, which uses both flag map and bitmap simultaneously to decide if a boundary pixel can be deleted, as well as the incorporation of smoothing templates to smooth the final skeleton, is proposed in this paper. Three performance measurements are proposed for an objective evaluation of this novel algorithm against a s...

Artificial Neural Networks (ANNs) have been used to perform classification for Automatic Speech Recognition (ASR). In this paper, we propose a new neural network, the Contenders' Network (CN) which requires little initial knowledge of the classification problem and lesser neurons than other ANNS.

We describe methods to analyse and obtain the optimal values of various factors that affect the performance of a signature verification system using a neural network approach. A modified model of backpropagation is used to reduce the learning time of the system. Various factors that are examined in this paper are the effect of learning rate of the...

Artificial Neural Networks (ANNs) have been used to perform classification for Automatic Speech Recognition (ASR). This paper proposes a new neural network, the Contenders' Network (CN) which requires little initial knowledge of the classification problem and lesser neurons than other ANNs

The authors wish to acknowledge the generous financial support from the Nanyang Technological University, School of Applied Science for funding the AF project. AF is an intelligent toolkit for anti-forgery. Abstract: A new sequential thinning algorithm, which uses both flag map and bitmap simultaneously to decide if a boundary pixel can be deleted,...

This project considers the problems of feature extraction, pose estimation and identification of spatial objects under different working environments, lighting conditions and camera projections. A model-based recognition system is proposed and implemented for identifying and estimating the pose of target objects comprising of straight and well-defi...

Despite the success of many pattern recognition problems in a constrained domain, the task of pattern recognition is "ill-defined" and difficult due to the noise and large variations in input data. A promising approach is to use several classifiers simultaneously, such that they can complement each other in correctness. This thesis tackles the reco...