Using the EM Algorithm to Train Neural Networks: Misconceptions and a New Algorithm for Multiclass Classification

The expectation-maximization (EM) algorithm has been of considerable interest in recent years as the basis for various algorithms in application areas of neural networks such as pattern recognition. However, there exists some misconceptions concerning its application to neural networks. In this paper, we clarify these misconceptions and consider how the EM algorithm can be adopted to train multilayer perceptron (MLP) and mixture of experts (ME) networks in applications to multiclass classification. We identify some situations where the application of the EM algorithm to train MLP networks may be of limited value and discuss some ways of handling the difficulties. For ME networks, it is reported in the literature that networks trained by the EM algorithm using iteratively reweighted least squares (IRLS) algorithm in the inner loop of the M-step, often performed poorly in multiclass classification. However, we found that the convergence of the IRLS algorithm is stable and that the log likelihood is monotonic increasing when a learning rate smaller than one is adopted. Also, we propose the use of an expectation-conditional maximization (ECM) algorithm to train ME networks. Its performance is demonstrated to be superior to the IRLS algorithm on some simulated and real data sets.

Genetic algorithms with applications in wireless communications

Genetic algorithms (GAs) are known to locate the global optimal solution provided sufficient population and/or generation is used. Practically, a near-optimal satisfactory result can be found by Gas with a limited number of generations. In wireless communications, the exhaustive searching approach is widely applied to many techniques, such as maximum likelihood decoding (MLD) and distance spectrum (DS) techniques. The complexity of the exhaustive searching approach in the MLD or the DS technique is exponential in the number of transmit antennas and the size of the signal constellation for the multiple-input multiple-output (MIMO) communication systems. If a large number of antennas and a large size of signal constellations, e.g. PSK and QAM, are employed in the MIMO systems, the exhaustive searching approach becomes impractical and time consuming. In this paper, the GAs are applied to the MLD and DS techniques to provide a near-optimal performance with a reduced computational complexity for the MIMO systems. Two different GA-based efficient searching approaches are proposed for the MLD and DS techniques, respectively. The first proposed approach is based on a GA with sharing function method, which is employed to locate the multiple solutions of the distance spectrum for the Space-time Trellis Coded Orthogonal Frequency Division Multiplexing (STTC-OFDM) systems. The second approach is the GA-based MLD that attempts to find the closest point to the transmitted signal. The proposed approach can return a satisfactory result with a good initial signal vector provided to the GA. Through simulation results, it is shown that the proposed GA-based efficient searching approaches can achieve near-optimal performance, but with a lower searching complexity comparing with the original MLD and DS techniques for the MIMO systems.

System for foreign exchange trading using genetic algorithms and reinforcement learning

Foreign exchange trading has emerged recently as a significant activity in many countries. As with most forms of trading, the activity is influenced by many random parameters so that the creation of a system that effectively emulates the trading process will be very helpful. A major issue for traders in the deregulated Foreign Exchange Market is when to sell and when to buy a particular currency in order to maximize profit. This paper presents novel trading strategies based on the machine learning methods of genetic algorithms and reinforcement learning.

GANN: Genetic algorithm neural networks for the detection of conserved combinations of features in DNA

Background: The multitude of motif detection algorithms developed to date have largely focused on the detection of patterns in primary sequence. Since sequence-dependent DNA structure and flexibility may also play a role in protein-DNA interactions, the simultaneous exploration of sequence-and structure-based hypotheses about the composition of binding sites and the ordering of features in a regulatory region should be considered as well. The consideration of structural features requires the development of new detection tools that can deal with data types other than primary sequence. Results: GANN ( available at http://bioinformatics.org.au/gann) is a machine learning tool for the detection of conserved features in DNA. The software suite contains programs to extract different regions of genomic DNA from flat files and convert these sequences to indices that reflect sequence and structural composition or the presence of specific protein binding sites. The machine learning component allows the classification of different types of sequences based on subsamples of these indices, and can identify the best combinations of indices and machine learning architecture for sequence discrimination. Another key feature of GANN is the replicated splitting of data into training and test sets, and the implementation of negative controls. In validation experiments, GANN successfully merged important sequence and structural features to yield good predictive models for synthetic and real regulatory regions. Conclusion: GANN is a flexible tool that can search through large sets of sequence and structural feature combinations to identify those that best characterize a set of sequences.

The applicability of recurrent neural networks for biological sequence analysis

Selection of machine learning techniques requires a certain sensitivity to the requirements of the problem. In particular, the problem can be made more tractable by deliberately using algorithms that are biased toward solutions of the requisite kind. In this paper, we argue that recurrent neural networks have a natural bias toward a problem domain of which biological sequence analysis tasks are a subset. We use experiments with synthetic data to illustrate this bias. We then demonstrate that this bias can be exploitable using a data set of protein sequences containing several classes of subcellular localization targeting peptides. The results show that, compared with feed forward, recurrent neural networks will generally perform better on sequence analysis tasks. Furthermore, as the patterns within the sequence become more ambiguous, the choice of specific recurrent architecture becomes more critical.

A class of self-stabilizing MCA learning algorithms

In this letter, we propose a class of self-stabilizing learning algorithms for minor component analysis (MCA), which includes a few well-known MCA learning algorithms. Self-stabilizing means that the sign of the weight vector length change is independent of the presented input vector. For these algorithms, rigorous global convergence proof is given and the convergence rate is also discussed. By combining the positive properties of these algorithms, a new learning algorithm is proposed which can improve the performance. Simulations are employed to confirm our theoretical results.

Statistical racing techniques for improved empirical evaluation of evolutionary algorithms

In empirical studies of Evolutionary Algorithms, it is usually desirable to evaluate and compare algorithms using as many different parameter settings and test problems as possible, in border to have a clear and detailed picture of their performance. Unfortunately, the total number of experiments required may be very large, which often makes such research work computationally prohibitive. In this paper, the application of a statistical method called racing is proposed as a general-purpose tool to reduce the computational requirements of large-scale experimental studies in evolutionary algorithms. Experimental results are presented that show that racing typically requires only a small fraction of the cost of an exhaustive experimental study.

On building a principled framework for evaluating and testing evolutionary algorithms: A continuous landscape generator

In this paper, we address some issue related to evaluating and testing evolutionary algorithms. A landscape generator based on Gaussian functions is proposed for generating a variety of continuous landscapes as fitness functions. Through some initial experiments, we illustrate the usefulness of this landscape generator in testing evolutionary algorithms.

An implementation of genetic algorithms as a basis for a trading system on the foreign exchange market

Foreign exchange trading has emerged in recent times as a significant activity in many countries. As with most forms of trading, the activity is influenced by many random parameters so that the creation of a system that effectively emulates the trading process is very helpful. In this paper, we try to create such a system with a genetic algorithm engine to emulate trader behaviour on the foreign exchange market and to find the most profitable trading strategy.

Governing coordination: Behavioural principles and neural correlates

The coordination of movement is governed by a coalition of constraints. The expression of these constraints ranges from the concrete—the restricted range of motion offered by the mechanical configuration of our muscles and joints; to the abstract—the difficulty that we experience in combining simple movements into complex rhythms. We seek to illustrate that the various constraints on coordination are complementary and inclusive, and the means by which their expression and interaction are mediated systematically by the integrative action of the central nervous system (CNS). Beyond identifying the general principles at the behavioural level that govern the mutual interplay of constraints, we attempt to demonstrate that these principles have as their foundation specific functional properties of the cortical motor systems. We propose that regions of the brain upstream of the motor cortex may play a significant role in mediating interactions between the functional representations of muscles engaged in sensorimotor coordination tasks. We also argue that activity in these ldquosupramotorrdquo regions may mediate the stabilising role of augmented sensory feedback.

The sites of neural adaptation induced by resistance training in humans

Although it has long been supposed that resistance training causes adaptive changes in the CNS, the sites and nature of these adaptations have not previously been identified. In order to determine whether the neural adaptations to resistance training occur to a greater extent at cortical or subcortical sites in the CNS, we compared the effects of resistance training on the electromyographic (EMG) responses to transcranial magnetic (TMS) and electrical (TES) stimulation. Motor evoked potentials (MEPs) were recorded from the first dorsal interosseous muscle of 16 individuals before and after 4 weeks of resistance training for the index finger abductors (n = 8), or training involving finger abduction-adduction without external resistance (n = 8). TMS was delivered at rest at intensities from 5 % below the passive threshold to the maximal output of the stimulator. TMS and TES were also delivered at the active threshold intensity while the participants exerted torques ranging from 5 to 60 % of their maximum voluntary contraction (MVC) torque. The average latency of MEPs elicited by TES was significantly shorter than that of TMS MEPs (TES latency = 21.5 ± 1.4 ms; TMS latency = 23.4 ± 1.4 ms; P < 0.05), which indicates that the site of activation differed between the two forms of stimulation. Training resulted in a significant increase in MVC torque for the resistance-training group, but not the control group. There were no statistically significant changes in the corticospinal properties measured at rest for either group. For the active trials involving both TMS and TES, however, the slope of the relationship between MEP size and the torque exerted was significantly lower after training for the resistance-training group (P < 0.05). Thus, for a specific level of muscle activity, the magnitude of the EMG responses to both forms of transcranial stimulation were smaller following resistance training. These results suggest that resistance training changes the functional properties of spinal cord circuitry in humans, but does not substantially affect the organisation of the motor cortex.

Algorithms for sequence analysis via mutagenesis

Despite many successes of conventional DNA sequencing methods, some DNAs remain difficult or impossible to sequence. Unsequenceable regions occur in the genomes of many biologically important organisms, including the human genome. Such regions range in length from tens to millions of bases, and may contain valuable information such as the sequences of important genes. The authors have recently developed a technique that renders a wide range of problematic DNAs amenable to sequencing. The technique is known as sequence analysis via mutagenesis (SAM). This paper presents a number of algorithms for analysing and interpreting data generated by this technique.

Special issue on advances in data mining and its applications

Data mining is the process to identify valid, implicit, previously unknown, potentially useful and understandable information from large databases. It is an important step in the process of knowledge discovery in databases, (Olaru & Wehenkel, 1999). In a data mining process, input data can be structured, seme-structured, or unstructured. Data can be in text, categorical or numerical values. One of the important characteristics of data mining is its ability to deal data with large volume, distributed, time variant, noisy, and high dimensionality. A large number of data mining algorithms have been developed for different applications. For example, association rules mining can be useful for market basket problems, clustering algorithms can be used to discover trends in unsupervised learning problems, classification algorithms can be applied in decision-making problems, and sequential and time series mining algorithms can be used in predicting events, fault detection, and other supervised learning problems (Vapnik, 1999). Classification is among the most important tasks in the data mining, particularly for data mining applications into engineering fields. Together with regression, classification is mainly for predictive modelling. So far, there have been a number of classification algorithms in practice. According to (Sebastiani, 2002), the main classification algorithms can be categorized as: decision tree and rule based approach such as C4.5 (Quinlan, 1996); probability methods such as Bayesian classifier (Lewis, 1998); on-line methods such as Winnow (Littlestone, 1988) and CVFDT (Hulten 2001), neural networks methods (Rumelhart, Hinton & Wiliams, 1986); example-based methods such as k-nearest neighbors (Duda & Hart, 1973), and SVM (Cortes & Vapnik, 1995). Other important techniques for classification tasks include Associative Classification (Liu et al, 1998) and Ensemble Classification (Tumer, 1996).