In this paper I take a look at a classic problem in Spanish morphosyntax, namely the alternation between the forms -se and -ra in the Imperfect Subjunctive (Imperfecto de Subjuntivo). Research on this topic has mainly focused on sociolinguistic variation, and has been done almost exclusively with impressionistic data and speakers’ intuitions. I address the problem from a usage-based perspective, using corpus linguistics methods. The main claim is that the choice between -se and -ra correlates to a certain extent with morphosyntactic and discourse factors. Through collostructional analysis I also show that there exists repelled and attracted collexemes that distinguish and relate both forms.

This dissertation presents the development of a semi-supervised incremental learning framework with a multi-view perspective for image concept modeling. For reliable image concept characterization, having a large number of labeled images is crucial. However, the size of the training set is often limited due to the cost required for generating concept labels associated with objects in a large quantity of images. To address this issue, in this research, we propose to incrementally incorporate unlabeled samples into a learning process to enhance concept models originally learned with a small number of labeled samples. To tackle the sub-optimality problem of conventional techniques, the proposed incremental learning framework selects unlabeled samples based on an expected error reduction function that measures contributions of the unlabeled samples based on their ability to increase the modeling accuracy. To improve the convergence property of the proposed incremental learning framework, we further propose a multi-view learning approach that makes use of multiple features such as color, texture, etc., of images when including unlabeled samples. For robustness to mismatches between training and testing conditions, a discriminative learning algorithm, namely a kernelized maximal- figure-of-merit (kMFoM) learning approach is also developed. Combining individual techniques, we conduct a set of experiments on various image concept modeling problems, such as handwritten digit recognition, object recognition, and image spam detection to highlight the effectiveness of the proposed framework.

An important part of gene regulation is mediated by specific proteins, called transcription factors, which influence the transcription of a particular gene by binding to specific sites on DNA sequences, called transcription factor binding sites (TFBS) or, simply, motifs. Such binding sites are relatively short segments of DNA, normally 5 to 25 nucleotides long, over- represented in a set of co-regulated DNA sequences. There are two different problems in this setup: motif representation, accounting for the model that describes the TFBS's; and motif discovery, focusing in unravelling TFBS's from a set of co-regulated DNA sequences. This thesis proposes a discriminative scoring criterion that culminates in a discriminative mixture of Bayesian networks to distinguish TFBS's from the background DNA. This new probabilistic model supports further evidence in non-additivity among binding site positions, providing a superior discriminative power in TFBS's detection. On the other hand, extra knowledge carefully selected from the literature was incorporated in TFBS discovery in order to capture a variety of characteristics of the TFBS's patterns. This extra knowledge was combined during the process of motif discovery leading to results that are considerably more accurate than those achieved by methods that rely in the DNA sequence alone.

This dissertation presents segmental discriminative analysis techniques for American Sign Language (ASL) recognition and verification. ASL recognition is a sequence classification problem. One of the most successful techniques for recognizing ASL is the hidden Markov model (HMM) and its variants. This dissertation addresses two problems in sign recognition by HMMs. The first is discriminative feature selection for temporally-correlated data. Temporal correlation in sequences often causes difficulties in feature selection. To mitigate this problem, this dissertation proposes segmentally-boosted HMMs (SBHMMs), which construct the state-optimized features in a segmental and discriminative manner. The second problem is the decomposition of ASL signs for efficient and accurate recognition. For this problem, this dissertation proposes discriminative state-space clustering (DISC), a data-driven method of automatically extracting sub-sign units by state-tying from the results of feature selection. DISC and SBHMMs can jointly search for discriminative feature sets and representation units of ASL recognition. ASL verification, which determines whether an input signing sequence matches a pre-defined phrase, shares similarities with ASL recognition, but it has more prior knowledge and a higher expectation of accuracy. Therefore, ASL verification requires additional discriminative analysis not only in utilizing prior knowledge but also in actively selecting a set of phrases that have a high expectation of verification accuracy in the service of improving the experience of users. This dissertation describes ASL verification using CopyCat, an ASL game that helps deaf children acquire language abilities at an early age. It then presents the "probe" technique which automatically searches for an optimal threshold for verification using prior knowledge and BIG, a bi-gram error-ranking predictor which efficiently selects/creates phrases that, based on the previous performance of existing verification systems, should have high verification accuracy. This work demonstrates the utility of the described technologies in a series of experiments. SBHMMs are validated in ASL phrase recognition as well as various other applications such as lip reading and speech recognition. DISC-SBHMMs consistently produce fewer errors than traditional HMMs and SBHMMs in recognizing ASL phrases using an instrumented glove. Probe achieves verification efficacy comparable to the optimum obtained from manually exhaustive search. Finally, when verifying phrases in CopyCat, BIG predicts which CopyCat phrases, even unseen in training, will have the best verification accuracy with results comparable to much more computationally intensive methods.