The Support Vector Machine has gained significant popularity over recent years as a kernel-based supervised learning technique. However, choosing the appropriate kernel function and its associated parameters is not a trivial task. The kernel is often chosen from several widely-used and general-purpose functions, and the parameters are then empirically tuned for the best results on a specific data set. This paper explores the use of Particle Swarm Optimization and Genetic Programming as evolutionary approaches to evolve effective kernel functions for a given dataset. Rather than using expert knowledge, we evolve kernel functions without human-guided knowledge or intuition. Our results show consistently better SVM performance with evolved kernels over a variety of traditional kernels on several datasets.

Decision making has proposed multiple methods to help the decision maker in his analysis, by suggesting ways of formalization of the preferences as well as the assessment of the uncertainties. Although these techniques are established and proven to be mathematically sound, experience has shown that in certain situations we tend to avoid the formal approach by acting intuitively. Especially, when the decision involves a large number of attributes and outcomes, and where we need to use pragmatic and heuristic simplifications such as considering only the most important attributes and omitting the others. In this paper, we provide a model for decision making in situations subject to a large predictive uncertainty with a small learning sample. The high predictive uncertainty is concretized by a countably infinite number of prospects, making the preferences assessment more difficult. Our main result is an extension of the Maximum Entropy utility (MEU) principle into an asymptotic maximum entropy utility principle for preferences elicitation. This will allow us to overcome the limits of the existing MEU method to the extend that we focus on utility assessment when the set of the available discrete prospects is countably infinite. Furthermore, our proposed model can be used to analyze situations of high-cognitive load as well as to understand how humans handle these problems under Ceteris Paribus assumption.

Many studies have been conducted on how to detect emotion classes or magnitudes from multimedia information such as text, audio, and images. However, the methods that can use predicted emotion classes and magnitudes to render emotion expressions in Embodied Conversational Agents (ECA) are still unclear. This paper proposes a computer graphics methodology that uses predicted non-linear regression values to render facial expressions using mesh morphing techniques. Results of the rendering technique are presented and discussed.

In the area of student knowledge assessment, knowledge tracing is a model that has been used for over a decade to predict student knowledge and performance. Many modifications to this model have been proposed and evaluated, however, the modifications are often based on a combination of intuition and experience in the domain. This method of model improvement can be difficult for researchers without high level of domain experience and furthermore, the best improvements to the model could be unintuitive ones. Therefore, we propose a completely data driven approach to model improvement. This alternative allows for researchers to evaluate which aspects of a model are most likely to result in model performance improvement. Our results suggest a variety of different improvements to knowledge tracing many of which have not been explored.

Computational indices related to n-gram production were developed in order to assess the potential for n-gram indices to predict human scores of essay quality. A regression analyses was conducted on a corpus of 313 argumentative essays. The analyses demonstrated that a variety of n-gram indices were highly correlated to essay quality, but were also highly correlated to the number of words in the text (although many of the n-gram indices were stronger predictors of writing quality than the number of words in a text). A second regression analysis was conducted on a corpus of 88 argumentative essays that were controlled for text length differences. This analysis demonstrated that n-gram indices were still strong predictors of essay quality when text length was not a factor.

Are the words that people use indicative of their personality type preferences? In this paper, it is hypothesized that word-usage is not independent of personality type, as measured by the Myers-Briggs Type Indicator (MBTI) personality assessment tool. In-class writing samples were taken from 40 graduate students along with the MBTI. The experiment utilizes naïve Bayes classifiers and Support Vector Machines (SVMs) in an attempt to guess an individual’s personality type based on their word-choice. Classification is also attempted using emotional, social, cognitive, and psychological dimensions elicited by the analysis software, Linguistic Inquiry and Word Count (LIWC). The classifiers are evaluated with 40 distinct trials (leave-one-out cross validation), and parameters are chosen using leave-one-out cross validation of each trial’s training set. The experiment showed that the naïve Bayes classifiers (word-based and LIWC-based) outperformed the SVMs when guessing Sensing-Intuition (S-N) and Thinking-Feeling (T-F).

In this thesis I develop a variety of techniques to train, evaluate, and sample from intractable and high dimensional probabilistic models. Abstract exceeds arXiv space limitations -- see PDF.

Machine learning research is often conducted in vitro, divorced from motivating practical applications. A researcher might develop a new method for the general task of classification, then assess its utility by comparing its performance (such as accuracy or AUC) to that of existing classification models on publicly available datasets. In terms of advancing machine learning as an academic discipline, this approach has thus far proven quite fruitful. However, it is our view that the most interesting open problems in machine learning are those that arise during its application to real-world problems. We illustrate this point by reviewing two of our interdisciplinary collaborations, both of which have posed unique machine learning problems, providing fertile ground for novel research.

Collaborative filtering is a rapidly advancing research area. Every year several new techniques are proposed and yet it is not clear which of the techniques work best and under what conditions. In this paper we conduct a study comparing several collaborative filtering techniques -- both classic and recent state-of-the-art -- in a variety of experimental contexts. Specifically, we report conclusions controlling for number of items, number of users, sparsity level, performance criteria, and computational complexity. Our conclusions identify what algorithms work well and in what conditions, and contribute to both industrial deployment collaborative filtering algorithms and to the research community.

This paper studies parallelization schemes for stochastic Vector Quantization algorithms in order to obtain time speed-ups using distributed resources. We show that the most intuitive parallelization scheme does not lead to better performances than the sequential algorithm. Another distributed scheme is therefore introduced which obtains the expected speed-ups. Then, it is improved to fit implementation on distributed architectures where communications are slow and inter-machines synchronization too costly. The schemes are tested with simulated distributed architectures and, for the last one, with Microsoft Windows Azure platform obtaining speed-ups up to 32 Virtual Machines.