Improving Automated Planning with Machine Learning

This chapter reports the last machine learning techniques for the assistance of automated planning. Recent discoveries in automated planning have opened the scope of planners, from toy problems to real-world applications, making new challenges come into focus. The planning community believes that machine learning can assist to address these new challenges. The chapter collects the last machine learning techniques for assisting automated planners classified in: techniques for the improvement of the planning search processes and techniques for the automatic definition of planning action models. For each technique, the chapter provides an in-depth analysis of their domain, advantages and disadvantages. Finally, the chapter draws the outline of the new promising avenues for research in learning for planning systems.

Counting the Hidden Defects in Software Documents

The author uses neural networks to estimate how many defects are hidden in a software document. Input for the models are metrics that get collected when effecting a standard quality assurance technique on the document, a software inspection. For inspections, the empirical data sets typically are small. The author identifies two key ingredients for a successful application of neural networks to small data sets: Adapting the size, complexity, and input dimension of the networks to the amount of information available for training; and using Bayesian techniques instead of cross-validation for determining model parameters and selecting the final model. For inspections, the machine learning approach is highly successful and outperforms the previously existing defect estimation methods in software engineering by a factor of 4 in accuracy on the standard benchmark. The author’s approach is well applicable in other contexts that are subject to small training data sets.

Machine Learning for Biometrics

Biometrics aims at reliable and robust identification of humans from their personal traits, mainly for security and authentication purposes, but also for identifying and tracking the users of smarter applications. Frequently considered modalities are fingerprint, face, iris, palmprint and voice, but there are many other possible biometrics, including gait, ear image, retina, DNA, and even behaviours. This chapter presents a survey of machine learning methods used for biometrics applications, and identifies relevant research issues. The author focuses on three areas of interest: offline methods for biometric template construction and recognition, information fusion methods for integrating multiple biometrics to obtain robust results, and methods for dealing with temporal information. By introducing exemplary and influential machine learning approaches in the context of specific biometrics applications, the author hopes to provide the reader with the means to create novel machine learning solutions to challenging biometrics problems.

Evolutionary Multi-Objective Optimization of Autonomous Mobile Robots in Neural-Based Cognition for Behavioural Robustness

The utilization of a multi-objective approach for evolving artificial neural networks that act as the controllers for phototaxis and radio frequency (RF) localization behaviors of a virtual Khepera robot simulated in a 3D, physics-based environment is discussed in this chapter. It explains the comparison performances among the elitism without archive and elitism with archive used in the evolutionary multi-objective optimization (EMO) algorithm in an evolutionary robotics study. Furthermore, the controllers’ moving performances, tracking ability and robustness also have been demonstrated and tested with four different levels of environments. The experimentation results showed the controllers allowed the robots to navigate successfully, hence demonstrating the EMO algorithm can be practically used to automatically generate controllers for phototaxis and RF-localization behaviors, respectively. Understanding the underlying assumptions and theoretical constructs through the utilization of EMO will allow the robotics researchers to better design autonomous robot controllers that require minimal levels of human-designed elements.

A Survey of Bayesian Techniques in Computer Vision

The Bayesian approach to classification is intended to solve questions concerning how to assign a class to an observed pattern using probability estimations. Red, green and blue (RGB) or hue, saturation and lightness (HSL) values of pixels in digital colour images can be considered as feature vectors to be classified, thus leading to Bayesian colour image segmentation. Bayesian classifiers are also used to sort objects but, in this case, reduction of the dimensionality of the feature vector is often required prior to the analysis. This chapter shows some applications of Bayesian learning techniques in computer vision in the agriculture and agri-food sectors. Inspection and classification of fruit and vegetables, robotics, insect identification and process automation are some of the examples shown. Problems related with the natural variability of colour, sizes and shapes of biological products, and natural illuminants are also discussed. Moreover, implementations that lead to real-time implementation are explained.

Applications of Evolutionary Neural Networks for Sales Forecasting of Fashionable Products

The evolutionary neural network (ENN), which is the hybrid combination of evolutionary computation and neural network, is a suitable candidate for topology design, and is widely adopted. An ENN approach with a direct binary representation to every single neural network connection is proposed in this chapter for sales forecasting of fashionable products. In this chapter, the authors will first explore the details on how an evolutionary computation approach can be applied in searching for a desirable network structure for establishing the appropriate sales forecasting system. The optimized ENN structure for sales forecasting is then developed. With the use of real sales data, the authors compare the performances of the proposed ENN forecasting scheme with several traditional methods which include artificial neural network (ANN) and SARIMA. The authors obtain the conditions in which their proposed ENN outperforms other methods. Insights regarding the applications of ENN for forecasting sales of fashionable products are generated. Finally, future research directions are outlined.

Application of Neural Networks in Animal Science

Stock breeding has been one of the most important sources of food and labour throughout human history. Every advance in this field has always led to important and beneficial impacts on human society. These innovations have mainly taken place in machines or genetics, but data analysis has been somewhat ignored. Most of the published works in data analysis use linear models, and there are few works in the literature that use non-linear methods for data processing in stock breeding where these methods have proven to obtain better results and performance than linear, classical methods. This chapter demonstrates the use of non-linear methods by presenting two practical applications: milk yield production in goats, and analysis of farming production systems.

Machine Learning and Financial Investing

This chapter presents the case for knowledge-based machine learning in financial investing. Machine learning here, while it will exploit knowledge, will also rely heavily on the evolutionary computation paradigm of learning, namely reproduction with change and selection of the fit. The chapter will begin with a model for financial investing and then review what has been reported in the literature as regards knowledge-based and machine-learning-based methods for financial investing. Finally, a design of a financial investing system is described which incorporates the key features identified through the literature review. The emerging trend of incorporating knowledge-based methods into evolutionary methods for financial investing suggests opportunities for future researchers.

Machine Learning in Natural Language Processing

This chapter presents applications of machine learning techniques to traditional problems in natural language processing, including part-of-speech tagging, entity recognition and word-sense disambiguation. People usually solve such problems without difficulty or at least do a very good job. Linguistics may suggest labour-intensive ways of manually constructing rule-based systems. It is, however, the easy availability of large collections of texts that has made machine learning a method of choice for processing volumes of data well above the human capacity. One of the main purposes of text processing is all manner of information extraction and knowledge extraction from such large text. Machine learning methods discussed in this chapter have stimulated wide-ranging research in natural language processing and helped build applications with serious deployment potential.

Machine Learning Applications in Mega-Text Processing

This chapter presents applications of machine learning techniques to problems in natural language processing that require work with very large amounts of text. Such problems came into focus after the Internet and other computer-based environments acquired the status of the prime medium for text delivery and exchange. In all cases which the authors discuss, an algorithm has ensured a meaningful result, be it the knowledge of consumer opinions, the protection of personal information or the selection of news reports. The chapter covers elements of opinion mining, news monitoring and privacy protection, and, in parallel, discusses text representation, feature selection, and word category and text classification problems. The applications presented here combine scientific interest and significant economic potential.