In this thesis, I look at three problems with important applications in data processing. Incorporating side information, provided by the user or derived from data, is a main theme of each of these problems. <br /><br /> This thesis makes a number of contributions. The first is a technique for combining different embedding objectives, which is then exploited to incorporate side information expressed in terms of transformation invariants known to hold in the data. It also introduces two different ways of incorporating transformation invariants in order to make new similarity measures. Two algorithms are proposed which learn metrics based on different types of side information. These learned metrics can then be used in subsequent embedding methods. Finally, it introduces a manifold learning algorithm that is useful when applied to sequential decision problems. In this case we are given action labels in addition to data points. Actions in the manifold learned by this algorithm have meaningful representations in that they are represented as simple transformations.

In this thesis, I look at three problems with important applications in data processing. Incorporating side information, provided by the user or derived from data, is a main theme of each of these problems. <br /><br /> This thesis makes a number of contributions. The first is a technique for combining different embedding objectives, which is then exploited to incorporate side information expressed in terms of transformation invariants known to hold in the data. It also introduces two different ways of incorporating transformation invariants in order to make new similarity measures. Two algorithms are proposed which learn metrics based on different types of side information. These learned metrics can then be used in subsequent embedding methods. Finally, it introduces a manifold learning algorithm that is useful when applied to sequential decision problems. In this case we are given action labels in addition to data points. Actions in the manifold learned by this algorithm have meaningful representations in that they are represented as simple transformations.

In this work, we present a set of methods aimed at improving the discriminative power of time-varying features of signals that contain noise. These methods use properties of noise signals as well as information theoretic techniques to factor types of noise and support signal inference for electroencephalographic (EEG) based brain-machine interfaces (BMI). EEG data were collected over two studies aimed at addressing Psychophysiological issues involving symmetry and mental rotation processing. The Psychophysiological data gathered in the mental rotation study also tested the feasibility of using dissociations of mental rotation tasks correlated with rotation angle in a BMI. We show the feasibility of mental rotation for BMI by showing comparable bitrates and recognition accuracy to state-of-the-art BMIs. The conclusion is that by using the feature selection methods introduced in this work to dissociate mental rotation tasks, we produce bitrates and recognition rates comparable to current BMIs.

Biochemical reactions make up most of the activity in a cell. There is inherent stochasticity in the kinetic behavior of biochemical reactions which in turn governs the fate of various cellular processes. In this work, the precision of a method for dimensionality reduction for stochastic modeling of biochemical reactions is evaluated. Further, a method of stochastic simulation of reaction kinetics is implemented in case of a specific biochemical network involved in maintenance of long-term potentiation (LTP), the basic substrate for learning and memory formation. The dimensionality reduction method diverges significantly from a full stochastic model in prediction the variance of the fluctuations. The application of the stochastic simulation method to LTP modeling was used to find qualitative dependence of stochastic fluctuations on reaction volume and model parameters. / text

In many important real world applications the initial representation of the data is inconvenient, or even prohibitive for further analysis. For example, in image analysis, text analysis and computational genetics high-dimensional, massive, structural, incomplete, and noisy data sets are common. Therefore, feature extraction, or revelation of informative features from the raw data is one of fundamental machine learning problems. Efficient feature extraction helps to understand data and the process that generates it, reduce costs for future measurements and data analysis. The representation of the structured data as a compact set of informative numeric features allows applying well studied machine learning techniques instead of developing new ones.. The dissertation focuses on supervised and semi-supervised feature extraction methods, which optimize the dependence structure of features. The dependence is measured using the kernel estimator of Hilbert-Schmidt norm of covariance operator (HSIC measure). Two dependence structures are investigated: in the first case we seek features which maximize the dependence on the dependent variable, and in the second one, we additionally minimize the mutual dependence of features. Linear and kernel formulations of HBFE and HSCA are provided. Using Laplacian regularization framework we construct semi-supervised variants of HBFE and HSCA. Suggested algorithms were investigated experimentally using conventional and multilabel classification data... [to full text] / Daugelis praktiškai reikšmingu sistemu mokymo uždaviniu reikalauja gebeti panaudoti didelio matavimo, strukturizuotus, netiesinius duomenis. Vaizdu, teksto, socialiniu bei verslo ryšiu analize, ivairus bioinformatikos uždaviniai galetu buti tokiu uždaviniu pavyzdžiais. Todel požymiu išskyrimas dažnai yra pirmasis žingsnis, kuriuo pradedama duomenu analize ir nuo kurio priklauso galutinio rezultato sekme. Šio disertacinio darbo tyrimo objektas yra požymiu išskyrimo algoritmai, besiremiantys priklausomumo savoka. Darbe nagrinejamas priklausomumas, nusakytas kovariacinio operatoriaus Hilberto-Šmidto normos (HSIC mato) branduoliniu ivertiniu. Pasiulyti šiuo ivertiniu besiremiantys HBFE ir HSCA algoritmai leidžia dirbti su bet kokios strukturos duomenimis, bei yra formuluojami tikriniu vektoriu terminais (tai leidžia optimizavimui naudoti standartinius paketus), bei taikytini ne tik prižiurimo, bet ir dalinai prižiurimo mokymo imtims. Pastaruoju atveju HBFE ir HSCA modifikacijos remiasi Laplaso reguliarizacija. Eksperimentais su klasifikavimo bei daugiažymio klasifikavimo duomenimis parodyta, jog pasiulyti algoritmai leidžia pagerinti klasifikavimo efektyvuma lyginant su PCA ar LDA.

Daugelis praktiškai reikšmingu sistemu mokymo uždaviniu reikalauja gebeti panaudoti didelio matavimo, strukturizuotus, netiesinius duomenis. Vaizdu, teksto, socialiniu bei verslo ryšiu analize, ivairus bioinformatikos uždaviniai galetu buti tokiu uždaviniu pavyzdžiais. Todel požymiu išskyrimas dažnai yra pirmasis žingsnis, kuriuo pradedama duomenu analize ir nuo kurio priklauso galutinio rezultato sekme. Šio disertacinio darbo tyrimo objektas yra požymiu išskyrimo algoritmai, besiremiantys priklausomumo savoka. Darbe nagrinejamas priklausomumas, nusakytas kovariacinio operatoriaus Hilberto-Šmidto normos (HSIC mato) branduoliniu ivertiniu. Pasiulyti šiuo ivertiniu besiremiantys HBFE ir HSCA algoritmai leidžia dirbti su bet kokios strukturos duomenimis, bei yra formuluojami tikriniu vektoriu terminais (tai leidžia optimizavimui naudoti standartinius paketus), bei taikytini ne tik prižiurimo, bet ir dalinai prižiurimo mokymo imtims. Pastaruoju atveju HBFE ir HSCA modifikacijos remiasi Laplaso reguliarizacija. Eksperimentais su klasifikavimo bei daugiažymio klasifikavimo duomenimis parodyta, jog pasiulyti algoritmai leidžia pagerinti klasifikavimo efektyvuma lyginant su PCA ar LDA. / In many important real world applications the initial representation of the data is inconvenient, or even prohibitive for further analysis. For example, in image analysis, text analysis and computational genetics high-dimensional, massive, structural, incomplete, and noisy data sets are common. Therefore, feature extraction, or revelation of informative features from the raw data is one of fundamental machine learning problems. Efficient feature extraction helps to understand data and the process that generates it, reduce costs for future measurements and data analysis. The representation of the structured data as a compact set of informative numeric features allows applying well studied machine learning techniques instead of developing new ones.. The dissertation focuses on supervised and semi-supervised feature extraction methods, which optimize the dependence structure of features. The dependence is measured using the kernel estimator of Hilbert-Schmidt norm of covariance operator (HSIC measure). Two dependence structures are investigated: in the first case we seek features which maximize the dependence on the dependent variable, and in the second one, we additionally minimize the mutual dependence of features. Linear and kernel formulations of HBFE and HSCA are provided. Using Laplacian regularization framework we construct semi-supervised variants of HBFE and HSCA. Suggested algorithms were investigated experimentally using conventional and multilabel classification data... [to full text]

Single view reconstruction is fundamentally an under-constrained problem. We aim to develop new approaches to model human face and motion with model priors that restrict the space of possible solutions. First, we develop a novel approach to recover the 3D shape from a single view image under challenging conditions, such as large variations in illumination and pose. The problem is addressed by employing the techniques of non-linear manifold embedding and alignment. Specifically, the local image models for each patch of facial images and the local surface models for each patch of 3D shape are learned using a non-linear dimensionality reduction technique, and the correspondences between these local models are then learned by a manifold alignment method. Local models successfully remove the dependency of large training databases for human face modeling. By combining the local shapes, the global shape of a face can be reconstructed directly from a single linear system of equations via least square. Unfortunately, this learning-based approach cannot be successfully applied to the problem of human motion modeling due to the internal and external variations in single view video-based marker-less motion capture. Therefore, we introduce a new model-based approach for capturing human motion using a stream of depth images from a single depth sensor. While a depth sensor provides metric 3D information, using a single sensor, instead of a camera array, results in a view-dependent and incomplete measurement of object motion. We develop a novel two-stage template fitting algorithm that is invariant to subject size and view-point variations, and robust to occlusions. Starting from a known pose, our algorithm first estimates a body configuration through temporal registration, which is used to search the template motion database for a best match. The best match body configuration as well as its corresponding surface mesh model are deformed to fit the input depth map, filling in the part that is occluded from the input and compensating for differences in pose and body-size between the input image and the template. Our approach does not require any makers, user-interaction, or appearance-based tracking. Experiments show that our approaches can achieve good modeling results for human face and motion, and are capable of dealing with variety of challenges in single view reconstruction, e.g., occlusion.

In many real-world applications of supervised learning, only a limited number of labeled examples are available because the cost of obtaining high-quality examples is high. Even with a relatively large number of labeled examples, the learning problem may still suffer from limited supervision as the complexity of the prediction function increases. Therefore, learning with limited supervision presents a major challenge to machine learning. With the goal of supervision reduction, this thesis studies the representation, discovery and incorporation of extra input and output information in learning. Information about the input space can be encoded by regularization. We first design a semi-supervised learning method for text classification that encodes the correlation of words inferred from seemingly irrelevant unlabeled text. We then propose a multi-task learning framework with a matrix-normal penalty, which compactly encodes the covariance structure of the joint input space of multiple tasks. To capture structure information that is more general than covariance and correlation, we study a class of regularization penalties on model compressibility. Then we design the projection penalty, which encodes the structure information from a dimension reduction while controlling the risk of information loss. Information about the output space can be exploited by error correcting output codes. Using the composite likelihood view, we propose an improved pairwise coding for multi-label classification, which encodes pairwise label density (as opposed to label comparisons) and decodes using variational methods. We then investigate problemdependent codes, where the encoding is learned from data instead of being predefined. We first propose a multi-label output code using canonical correlation analysis, where predictability of the code is optimized. We then argue that both discriminability and predictability are critical for output coding, and propose a max-margin formulation that promotes both discriminative and predictable codes. We empirically study our methods in a wide spectrum of applications, including document categorization, landmine detection, face recognition, brain signal classification, handwritten digit recognition, house price forecasting, music emotion prediction, medical decision, email analysis, gene function classification, outdoor scene recognition, and so forth. In all these applications, our proposed methods for encoding input and output information lead to significantly improved prediction performance.

Doctor of Philosophy (PhD) / This thesis presents a computational method for the inductive inference of explicit and implicit semantic design knowledge from the symbolic-mathematical syntax of design formulations using an unsupervised pattern recognition and extraction approach. Existing research shows that AI / machine learning based design computation approaches either require high levels of knowledge engineering or large training databases to acquire problem reformulation knowledge. The method presented in this thesis addresses these methodological limitations. The thesis develops, tests, and evaluates ways in which the method may be employed for design problem reformulation. The method is based on the linear algebra based factorization method Singular Value Decomposition (SVD), dimensionality reduction and similarity measurement through unsupervised clustering. The method calculates linear approximations of the associative patterns of symbol cooccurrences in a design problem representation to infer induced coupling strengths between variables, constraints and system components. Unsupervised clustering of these approximations is used to identify useful reformulations. These two components of the method automate a range of reformulation tasks that have traditionally required different solution algorithms. Example reformulation tasks that it performs include selection of linked design variables, parameters and constraints, design decomposition, modularity and integrative systems analysis, heuristically aiding design “case” identification, topology modeling and layout planning. The relationship between the syntax of design representation and the encoded semantic meaning is an open design theory research question. Based on the results of the method, the thesis presents a set of theoretical postulates on computable relationships between design syntax and semantics. The postulates relate the performance of the method with empirical findings and theoretical insights provided by cognitive neuroscience and cognitive science on how the human mind engages in symbol processing and the resulting capacities inherent in symbolic representational systems to encode “meaning”. The performance of the method suggests that semantic “meaning” is a higher order, global phenomenon that lies distributed in the design representation in explicit and implicit ways. A one-to-one local mapping between a design symbol and its meaning, a largely prevalent approach adopted by many AI and learning algorithms, may not be sufficient to capture and represent this meaning. By changing the theoretical standpoint on how a “symbol” is defined in design representations, it was possible to use a simple set of mathematical ideas to perform unsupervised inductive inference of knowledge in a knowledge-lean and training-lean manner, for a knowledge domain that traditionally relies on “giving” the system complex design domain and task knowledge for performing the same set of tasks.

Doctor of Philosophy (PhD) / This thesis presents a computational method for the inductive inference of explicit and implicit semantic design knowledge from the symbolic-mathematical syntax of design formulations using an unsupervised pattern recognition and extraction approach. Existing research shows that AI / machine learning based design computation approaches either require high levels of knowledge engineering or large training databases to acquire problem reformulation knowledge. The method presented in this thesis addresses these methodological limitations. The thesis develops, tests, and evaluates ways in which the method may be employed for design problem reformulation. The method is based on the linear algebra based factorization method Singular Value Decomposition (SVD), dimensionality reduction and similarity measurement through unsupervised clustering. The method calculates linear approximations of the associative patterns of symbol cooccurrences in a design problem representation to infer induced coupling strengths between variables, constraints and system components. Unsupervised clustering of these approximations is used to identify useful reformulations. These two components of the method automate a range of reformulation tasks that have traditionally required different solution algorithms. Example reformulation tasks that it performs include selection of linked design variables, parameters and constraints, design decomposition, modularity and integrative systems analysis, heuristically aiding design “case” identification, topology modeling and layout planning. The relationship between the syntax of design representation and the encoded semantic meaning is an open design theory research question. Based on the results of the method, the thesis presents a set of theoretical postulates on computable relationships between design syntax and semantics. The postulates relate the performance of the method with empirical findings and theoretical insights provided by cognitive neuroscience and cognitive science on how the human mind engages in symbol processing and the resulting capacities inherent in symbolic representational systems to encode “meaning”. The performance of the method suggests that semantic “meaning” is a higher order, global phenomenon that lies distributed in the design representation in explicit and implicit ways. A one-to-one local mapping between a design symbol and its meaning, a largely prevalent approach adopted by many AI and learning algorithms, may not be sufficient to capture and represent this meaning. By changing the theoretical standpoint on how a “symbol” is defined in design representations, it was possible to use a simple set of mathematical ideas to perform unsupervised inductive inference of knowledge in a knowledge-lean and training-lean manner, for a knowledge domain that traditionally relies on “giving” the system complex design domain and task knowledge for performing the same set of tasks.