<p>The concept of sparseness is harnessed to learn a low dimensional representation of high dimensional data. This sparseness assumption is exploited in multiple ways. In the Bayesian Elastic Net, a small number of correlated features are identified for the response variable. In the sparse Factor Analysis for biomarker trajectories, the high dimensional gene expression data is reduced to a small number of latent factors, each with a prototypical dynamic trajectory. In the Bayesian Graphical LASSO, the inverse covariance matrix of the data distribution is assumed to be sparse, inducing a sparsely connected Gaussian graph. In the nonparametric Mixture of Factor Analyzers, the covariance matrices in the Gaussian Mixture Model are forced to be low-rank, which is closely related to the concept of block sparsity. </p><p>Finally in the information-theoretic projection design, a linear projection matrix is explicitly sought for information-preserving dimensionality reduction. All the methods mentioned above prove to be effective in learning both simulated and real high dimensional datasets.</p> / Dissertation

Uncertainty is a feature of many AI applications. While there are polynomial-time algorithms for planning in stochastic systems, planning is still slow, in part because most algorithms plan for all eventualities. Algorithms such as LAO* are able to find good or optimal policies more quickly when the starting state of the system is known. In this thesis we present an extension to LAO*, called BLAO*. BLAO* is an extension of the LAO* algorithm to a bidirectional search. We show that BLAO* finds optimal or E-optimal solutions for goal-directed MDPs without necessarily evaluating the entire state space. BLAO* converges much faster than LAO* or RTDP on our benchmarks.

As wireless networks continue to flourish worldwide and play an increasingly prominent role, it has become crucial to provide effective solutions to the inherent security issues associated with a wireless transmission medium. Unlike traditional solutions, which usually handle security at the application layer, the primary concern of this thesis is to analyze and develop solutions based on coding techniques at the physical layer. First, an information-theoretically secure communication protocol for quasi-static fading channels was developed and its performance with respect to theoretical limits was analyzed. A key element of the protocol is a reconciliation scheme for secret-key agreement based on low-density parity-check codes, which is specifically designed to operate on non-binary random variables and offers high reconciliation efficiency. Second, the fundamental trade-offs between cooperation and security were analyzed by investigating the transmission of confidential messages to cooperative relays. This information-theoretic study highlighted the importance of jamming as a means to increase secrecy and confirmed the importance of carefully chosen relaying strategies. Third, other applications of physical-layer security were investigated. Specifically, the use of secret-key agreement techniques for alternative cryptographic purposes was analyzed, and a framework for the design of practical information-theoretic commitment protocols over noisy channels was proposed. Finally, the benefit of using physical-layer coding techniques beyond the physical layer was illustrated by studying security issues in client-server networks. A coding scheme exploiting packet losses at the network layer was proposed to ensure reliable communication between clients and servers and security against colluding attackers.

Child-directed speech (CDS) differs from adult-directed speech (ADS) in several respects whose possible facilitating effects for language acquisition are still being studied. One such difference concerns articulation rate --- the number of linguistic units by the number of time units, excluding pauses --- which has been shown to be generally lower than in ADS. However, while it is well-established that ADS exhibits an inverse relation between articulation rate and information-theoretic surprisal --- the amount of information encoded in a linguistic unit --- this measure has been conspicuously absent in the study of articulation rate in CDS. Another issue is if the lower articulation rate in CDS is stable across utterances or an effect of local variation, such as final lengthening. The aim of this work is to arrive at a more comprehensive model of articulation rate in CDS by including surprisal and final lengthening. In particular, one-word utterances were studied, also in relation to word-length effects (the phenomenon that longer words generally have a higher articulation rate). To this end, a methodology for large-scale automatic phoneme-alignment was developed, which was applied to two longitudinal corpora of Swedish CDS. It was investigated i) how articulation rate in CDS varied with respect to child age, ii) whether there was a negative relation between articulation rate and surprisal in CDS, and iii) to what extent articulation rate was lower in CDS than in ADS. The results showed i) a weak positive effectof child age on articulation rate, ii) a negative relation between articulation rate and surprisal, and iii) that there was a lower articulation rate in CDS but that the difference could almost exclusively be attributed to one-word utterances and final lengthening. In other words, adults seem to adapt how fast they speak to their children's age, speaking faster to children is correlated with a reduced amount of information, and the difference in articulation rate between CDS and ADS is most prominent in isolated words and final lengthening. More generally, the results suggest that CDS is well-suited for word segmentation, since lower articulation rate in one-word utterances provides an additional cue.

Submitted by Maurício da Silva Medeiros Júnior (mauriciojr.df@gmail.com) on 2016-04-06T21:06:49Z No. of bitstreams: 1 Dissertação - Maurício Medeiros Jr. - FGV-EPGE.pdf: 2837403 bytes, checksum: dc338c6b56e600b8b9194b1c27abb080 (MD5) / Approved for entry into archive by BRUNA BARROS (bruna.barros@fgv.br) on 2016-04-19T18:34:15Z (GMT) No. of bitstreams: 1 Dissertação - Maurício Medeiros Jr. - FGV-EPGE.pdf: 2837403 bytes, checksum: dc338c6b56e600b8b9194b1c27abb080 (MD5) / Approved for entry into archive by Marcia Bacha (marcia.bacha@fgv.br) on 2016-04-27T18:16:31Z (GMT) No. of bitstreams: 1 Dissertação - Maurício Medeiros Jr. - FGV-EPGE.pdf: 2837403 bytes, checksum: dc338c6b56e600b8b9194b1c27abb080 (MD5) / Made available in DSpace on 2016-04-27T18:21:05Z (GMT). No. of bitstreams: 1 Dissertação - Maurício Medeiros Jr. - FGV-EPGE.pdf: 2837403 bytes, checksum: dc338c6b56e600b8b9194b1c27abb080 (MD5) Previous issue date: 2016-03-22 / We aim to provide a review of the stochastic discount factor bounds usually applied to diagnose asset pricing models. In particular, we mainly discuss the bounds used to analyze the disaster model of Barro (2006). Our attention is focused in this disaster model since the stochastic discount factor bounds that are applied to study the performance of disaster models usually consider the approach of Barro (2006). We first present the entropy bounds that provide a diagnosis of the analyzed disaster model which are the methods of Almeida and Garcia (2012, 2016); Ghosh et al. (2016). Then, we discuss how their results according to the disaster model are related to each other and also present the findings of other methodologies that are similar to these bounds but provide different evidence about the performance of the framework developed by Barro (2006).

abstract: Context – Urbanization can have negative effects on bat habitat use through the loss and isolation of habitat even for volant bats. Yet, how bats respond to the changing landscape composition and configuration of urban environments remains poorly understood. Objective – This study examines the relationship between bat habitat use and landscape pattern across multiple scales in the Phoenix metropolitan region. My research explores how landscape composition and configuration affects bat activity, foraging activity, and species richness (response variables), and the distinct habitats that they use. Methods – I used a multi-scale landscape approach and acoustic monitoring data to create predictive models that identified the key predictor variables across multiple scales within the study area. I selected three scales with the intent of capturing the landscape, home range, and site scales, which may all be relevant for understanding bat habitat use. Results – Overall, class-level metrics and configuration metrics best explained bat habitat use for bat species associated with this urban setting. The extent and extensiveness of water (corresponding to small water bodies and watercourses) were the most important predictor variables across all response variables. Bat activity was predicted to be high in native vegetation remnants, and low in native vegetation at the city periphery. Foraging activity was predicted to be high in fine-scale land cover heterogeneity. Species richness was predicted to be high in golf courses, and low in commercial areas. Bat habitat use was affected by urban landscape pattern mainly at the landscape and site scale. Conclusions – My results suggested in hot arid urban landscapes water is a limiting factor for bats, even in urban landscapes where the availability of water may be greater than in outlying native desert habitat. Golf courses had the highest species richness, and included the detection of the uncommon pocketed free-tailed bat (Nyctinomops femorosaccus). Water cover types had the second highest species richness. Golf courses may serve as important stop-overs or refuges for rare or elusive bats. Urban waterways and golf courses are novel urban cover types that can serve as compliments to urban preserves, and other green spaces for bat conservation. / Dissertation/Thesis / Masters Thesis Biology 2016

Breaking the anonymity of different wireless users with the purpose of decreasing internet crime rates is addressed in this thesis by considering radiometric identification techniques. Minute imperfections and non-idealities in the different transmitter components, especially the inherent nonlinearity in power amplifiers, result in variations in their Volterra series representations which could be utilized as a signature. For a two user scenario, signal processing algorithms based on generalized likelihood ratio test(GLRT) and the classical likelihood ratio test are introduced and the resulting receiver decision rules and performance curves are presented. These algorithms consider the high signal to noise ratio(SNR) case where we have available the input samples completely at the receiver which is a practical assumption for most cases. Volterra series are widely used in behavioral modeling of power amplifiers. To validate the existence of these variations in the Volterra series representation of power amplifiers, process variations are introduced as major sources. The plausibility of our techniques are justified by deriving and comparing the Volterra coefficients for the fast and slow process corners. Finally,an information theoretic framework is presented where the amount of mutual information of the output about the Volterra coefficients represents the amount of anonymity taken from users. Here, some results for the low SNR case are presented to prove the achievability of some information about individual systems using our hardware anonymity breaking techniques.

This paper compares two approaches for deriving subsumption algorithms for the description logic ALN: structural subsumption and an automata-theoretic characterization of subsumption. It turns out that structural subsumption algorithms can be seen as special implementations of the automata-theoretic characterization.

The representation of terminological knowledge may naturally lead to terminological cycles. In addition to descriptive semantics, the meaning of cyclic terminologies can also be captured by fixed-point semantics, namely, greatest and least fixed-point semantics. To gain a more profound understanding of these semantics and to obtain inference algorithms as well as complexity results for inconsistency, subsumption, and related inference tasks, this paper provides automata theoretic characterizations of these semantics. More precisely, the already existing results for FL₀ are extended to the language ALN, which additionally allows for primitive negation and number-restrictions. Unlike FL₀, the language ALN can express inconsistent concepts, which makes non-trivial extensions of the characterizations and algorithms necessary. Nevertheless, the complexity of reasoning does not increase when going from FL₀ to ALN. This distinguishes ALN from the very expressive languages with fixed-point operators proposed in the literature. It will be shown, however, that cyclic ALN-terminologies are expressive enough to capture schemata in certain semantic data models.

No description available.