Deep learning has in recent years come to dominate the previously separate fields of research in machine learning, computer vision, natural language understanding and speech recognition. Despite breakthroughs in training deep networks, there remains a lack of understanding of both the optimization and structure of deep networks. The approach advocated by many researchers in the field has been to train monolithic networks with excess complexity, and strong regularization --- an approach that leaves much to desire in efficiency. Instead we propose that carefully designing networks in consideration of our prior knowledge of the task and learned representation can improve the memory and compute efficiency of state-of-the art networks, and even improve generalization --- what we propose to denote as structural priors. We present two such novel structural priors for convolutional neural networks, and evaluate them in state-of-the-art image classification CNN architectures. The first of these methods proposes to exploit our knowledge of the low-rank nature of most filters learned for natural images by structuring a deep network to learn a collection of mostly small, low-rank, filters. The second addresses the filter/channel extents of convolutional filters, by learning filters with limited channel extents. The size of these channel-wise basis filters increases with the depth of the model, giving a novel sparse connection structure that resembles a tree root. Both methods are found to improve the generalization of these architectures while also decreasing the size and increasing the efficiency of their training and test-time computation. Finally, we present work towards conditional computation in deep neural networks, moving towards a method of automatically learning structural priors in deep networks. We propose a new discriminative learning model, conditional networks, that jointly exploit the accurate representation learning capabilities of deep neural networks with the efficient conditional computation of decision trees. Conditional networks yield smaller models, and offer test-time flexibility in the trade-off of computation vs. accuracy.

Statistical parametric speech synthesis (SPSS) has seen improvements over recent years, especially in terms of intelligibility. Synthetic speech is often clear and understandable, but it can also be bland and monotonous. Proper generation of natural speech prosody is still a largely unsolved problem. This is relevant especially in the context of expressive audiobook speech synthesis, where speech is expected to be fluid and captivating. In general, prosody can be seen as a layer that is superimposed on the segmental (phone) sequence. Listeners can perceive the same melody or rhythm in different utterances, and the same segmental sequence can be uttered with a different prosodic layer to convey a different message. For this reason, prosody is commonly accepted to be inherently suprasegmental. It is governed by longer units within the utterance (e.g. syllables, words, phrases) and beyond the utterance (e.g. discourse). However, common techniques for the modeling of speech prosody - and speech in general - operate mainly on very short intervals, either at the state or frame level, in both hidden Markov model (HMM) and deep neural network (DNN) based speech synthesis. This thesis presents contributions supporting the claim that stronger representations of suprasegmental variation are essential for the natural generation of fundamental frequency for statistical parametric speech synthesis. We conceptualize the problem by dividing it into three sub-problems: (1) representations of acoustic signals, (2) representations of linguistic contexts, and (3) the mapping of one representation to another. The contributions of this thesis provide novel methods and insights relating to these three sub-problems. In terms of sub-problem 1, we propose a multi-level representation of f0 using the continuous wavelet transform and the discrete cosine transform, as well as a wavelet-based decomposition strategy that is linguistically and perceptually motivated. In terms of sub-problem 2, we investigate additional linguistic features such as text-derived word embeddings and syllable bag-of-phones and we propose a novel method for learning word vector representations based on acoustic counts. Finally, considering sub-problem 3, insights are given regarding hierarchical models such as parallel and cascaded deep neural networks.

The appearance of the world varies dramatically not only from place to place but also from hour to hour and month to month. Billions of images that capture this complex relationship are uploaded to social-media websites every day and often are associated with precise time and location metadata. This rich source of data can be beneficial to improve our understanding of the globe. In this work, we propose a general framework that uses these publicly available images for constructing dense maps of different ground-level attributes from overhead imagery. In particular, we use well-defined probabilistic models and a weakly-supervised, multi-task training strategy to provide an estimate of the expected visual and auditory ground-level attributes consisting of the type of scenes, objects, and sounds a person can experience at a location. Through a large-scale evaluation on real data, we show that our learned models can be used for applications including mapping, image localization, image retrieval, and metadata verification.

[pt] Esta tese apresenta e examina contribuições inovadoras no front-end dos sistemas de reconhecimento automático de voz (RAV) para o realce e reconhecimento de voz em ambientes adversos. A primeira proposta consiste em aplicar um filtro de mediana sobre a função de distribuição de probabilidade de cada coeficiente cepstral antes de utilizar uma transformação para um domínio invariante às distorções, com o objetivo de adaptar a voz ruidosa ao ambiente limpo de referência através da modificação de histogramas. Fundamentadas nos resultados de estudos psicofísicos do sistema auditivo humano, que utiliza como princípio o fato de que o som que atinge o ouvido é sujeito a um processo chamado Análise de Cena Auditiva (ASA), o qual examina como o sistema auditivo separa as fontes de som que compõem a entrada acústica, três novas abordagens aplicadas independentemente foram propostas para realce e reconhecimento de voz. A primeira aplica a estimativa de uma nova máscara no domínio espectral usando o conceito da transformada de Fourier de tempo curto (STFT). A máscara proposta aplica a técnica Local Binary Pattern (LBP) à relação sinal ruído (SNR) de cada unidade de tempo-frequência (T-F) para estimar uma máscara de vizinhança ideal (INM). Continuando com essa abordagem, propõe-se em seguida nesta tese o mascaramento usando as transformadas wavelet com base nos LBP para realçar os espectros temporais dos coeficientes wavelet nas altas frequências. Finalmente, é proposto um novo método de estimação da máscara INM, utilizando um algoritmo de aprendizagem supervisionado das Deep Neural Networks (DNN) com o objetivo de realizar a classificação de unidades T-F obtidas da saída dos bancos de filtros pertencentes a uma mesma fonte de som (ou predominantemente voz ou predominantemente ruído). O desempenho é comparado com as técnicas de máscara tradicionais IBM e IRM, tanto em termos de qualidade objetiva da voz, como através de taxas de erro de palavra. Os resultados das técnicas propostas evidenciam as melhoras obtidas em ambientes ruidosos, com diferenças significativamente superiores às abordagens convencionais. / [en] This thesis presents and examines innovative contributions in frontend of the automatic speech recognition systems (ASR) for enhancement and speech recognition in adverse environments. The first proposal applies a median filter on the probability distribution function of each cepstral coefficient before using a transformation to a distortion-invariant domain, to adapt the corrupted voice to the clean reference environment by modifying histograms. Based on the results of psychophysical studies of the human auditory system, which uses as a principle the fact that sound reaching the ear is subjected to a process called Auditory Scene Analysis (ASA), which examines how the auditory system separates the sound sources that make up the acoustic input, three new approaches independently applied were proposed for enhancement and speech recognition. The first applies the estimation of a new mask in the spectral domain using the short-time Fourier Transform (STFT) concept. The proposed mask applies the Local Binary Pattern (LBP) technique to the Signal-to-Noise Ratio (SNR) of each time-frequency unit (T-F) to estimate an Ideal Neighborhood Mask (INM). Continuing with this approach, the masking using LBP-based wavelet transforms to highlight the temporal spectra of wavelet coefficients at high frequencies is proposed in this thesis. Finally, a new method of estimation of the INM mask is proposed, using a supervised learning algorithm of Deep Neural Network (DNN) to classify the T-F units obtained from the output of the filter banks belonging to a same source of sound (or predominantly voice or predominantly noise). The performance is compared with traditional IBM and IRM mask techniques, both regarding objective voice quality and through word error rates. The results of the proposed methods show the improvements obtained in noisy environments, with differences significantly superior to the conventional approaches.

Les recherches en Traitement Automatique des Langues (TAL) ont identifié différents niveaux d'analyse lexicale, syntaxique et sémantique. Il en découle un découpage hiérarchique des différentes tâches à réaliser afin d'analyser un énoncé. Les systèmes classiques du TAL reposent sur des analyseurs indépendants disposés en cascade au sein de chaînes de traitement (pipelines). Cette approche présente un certain nombre de limitations : la dépendance des modèles à la sélection empirique des traits, le cumul des erreurs dans le pipeline et la sensibilité au changement de domaine. Ces limitations peuvent conduire à des pertes de performances particulièrement importantes lorsqu'il existe un décalage entre les conditions d'apprentissage des modèles et celles d'utilisation. Un tel décalage existe lors de l'analyse de transcriptions automatiques de parole spontanée comme par exemple les conversations téléphoniques enregistrées dans des centres d'appels. En effet l'analyse d'une langue non-canonique pour laquelle il existe peu de données d'apprentissage, la présence de disfluences et de constructions syntaxiques spécifiques à l'oral ainsi que la présence d'erreurs de reconnaissance dans les transcriptions automatiques mènent à une détérioration importante des performances des systèmes d'analyse. C'est dans ce cadre que se déroule cette thèse, en visant à mettre au point des systèmes d'analyse à la fois robustes et flexibles permettant de dépasser les limitations des systèmes actuels à l'aide de modèles issus de l'apprentissage par réseaux de neurones profonds. / NLP researchers has identified different levels of linguistic analysis. This lead to a hierarchical division of the various tasks performed in order to analyze a text statement. The traditional approach considers task-specific models which are subsequently arranged in cascade within processing chains (pipelines). This approach has a number of limitations: the empirical selection of models features, the errors accumulation in the pipeline and the lack of robusteness to domain changes. These limitations lead to particularly high performance losses in the case of non-canonical language with limited data available such as transcriptions of conversations over phone. Disfluencies and speech-specific syntactic schemes, as well as transcription errors in automatic speech recognition systems, lead to a significant drop of performances. It is therefore necessary to develop robust and flexible systems. We intend to perform a syntactic and semantic analysis using a deep neural network multitask model while taking into account the variations of domain and/or language registers within the data.

abstract: Deep neural networks (DNN) have shown tremendous success in various cognitive tasks, such as image classification, speech recognition, etc. However, their usage on resource-constrained edge devices has been limited due to high computation and large memory requirement. To overcome these challenges, recent works have extensively investigated model compression techniques such as element-wise sparsity, structured sparsity and quantization. While most of these works have applied these compression techniques in isolation, there have been very few studies on application of quantization and structured sparsity together on a DNN model. This thesis co-optimizes structured sparsity and quantization constraints on DNN models during training. Specifically, it obtains optimal setting of 2-bit weight and 2-bit activation coupled with 4X structured compression by performing combined exploration of quantization and structured compression settings. The optimal DNN model achieves 50X weight memory reduction compared to floating-point uncompressed DNN. This memory saving is significant since applying only structured sparsity constraints achieves 2X memory savings and only quantization constraints achieves 16X memory savings. The algorithm has been validated on both high and low capacity DNNs and on wide-sparse and deep-sparse DNN models. Experiments demonstrated that deep-sparse DNN outperforms shallow-dense DNN with varying level of memory savings depending on DNN precision and sparsity levels. This work further proposed a Pareto-optimal approach to systematically extract optimal DNN models from a huge set of sparse and dense DNN models. The resulting 11 optimal designs were further evaluated by considering overall DNN memory which includes activation memory and weight memory. It was found that there is only a small change in the memory footprint of the optimal designs corresponding to the low sparsity DNNs. However, activation memory cannot be ignored for high sparsity DNNs. / Dissertation/Thesis / Masters Thesis Computer Engineering 2018

La thèse porte sur le développement d'architectures neuronales profondes permettant d'analyser des contenus textuels ou visuels, ou la combinaison des deux. De manière générale, le travail tire parti de la capacité des réseaux de neurones à apprendre des représentations abstraites. Les principales contributions de la thèse sont les suivantes: 1) Réseaux récurrents pour la compréhension de la parole: différentes architectures de réseaux sont comparées pour cette tâche sur leurs facultés à modéliser les observations ainsi que les dépendances sur les étiquettes à prédire. 2) Prédiction d’image et de mouvement : nous proposons une architecture permettant d'apprendre une représentation d'une image représentant une action humaine afin de prédire l'évolution du mouvement dans une vidéo ; l'originalité du modèle proposé réside dans sa capacité à prédire des images à une distance arbitraire dans une vidéo. 3) Encodeurs bidirectionnels multimodaux : le résultat majeur de la thèse concerne la proposition d'un réseau bidirectionnel permettant de traduire une modalité en une autre, offrant ainsi la possibilité de représenter conjointement plusieurs modalités. L'approche été étudiée principalement en structuration de collections de vidéos, dons le cadre d'évaluations internationales où l'approche proposée s'est imposée comme l'état de l'art. 4) Réseaux adverses pour la fusion multimodale: la thèse propose d'utiliser les architectures génératives adverses pour apprendre des représentations multimodales en offrant la possibilité de visualiser les représentations dans l'espace des images. / In this dissertation, the thesis that deep neural networks are suited for analysis of visual, textual and fused visual and textual content is discussed. This work evaluates the ability of deep neural networks to learn automatic multimodal representations in either unsupervised or supervised manners and brings the following main contributions:1) Recurrent neural networks for spoken language understanding (slot filling): different architectures are compared for this task with the aim of modeling both the input context and output label dependencies.2) Action prediction from single images: we propose an architecture that allow us to predict human actions from a single image. The architecture is evaluated on videos, by utilizing solely one frame as input.3) Bidirectional multimodal encoders: the main contribution of this thesis consists of neural architecture that translates from one modality to the other and conversely and offers and improved multimodal representation space where the initially disjoint representations can translated and fused. This enables for improved multimodal fusion of multiple modalities. The architecture was extensively studied an evaluated in international benchmarks within the task of video hyperlinking where it defined the state of the art today.4) Generative adversarial networks for multimodal fusion: continuing on the topic of multimodal fusion, we evaluate the possibility of using conditional generative adversarial networks to lean multimodal representations in addition to providing multimodal representations, generative adversarial networks permit to visualize the learned model directly in the image domain.

Deep artificial neural networks is a type of machine learning which can be used to find and utilize patterns in data. One of their many applications is as method for regression analysis. In this thesis deep artificial neural networks were implemented in the application of estimating the error of surface temperature forecasts as produced by a numerical weather prediction model. An ability to estimate the error of forecasts is synonymous with the ability to reduce forecast errors as the estimated error can be offset from the actual forecast. Six years of forecast data from the period 2010--2015 produced by the European Centre for Medium-Range Weather Forecasts' (ECMWF) numerical weather prediction model together with data from fourteen meteorological observational stations were used to train and evaluate error-predicting deep neural networks. The neural networks were able to reduce the forecast errors for all the locations that were tested to a varying extent. The largest reduction in error was by 83.0\% of the original error or a 16.7\degcs decrease in the mean-square error. The performance of the neural networks' error reduction ability was compared with that of a contemporary Kalman filter as implemented by the Swedish Meteorological and Hydrological Institute (SMHI). It was shown that the neural network implementation had superior performance for six out of seven of the evaluated stations where the Kalman filter had marginally better performance at one station.

Deep neural networks with millions of parameters are at the heart of many state of the art computer vision models. However, recent works have shown that models with much smaller number of parameters can often perform just as well. A smaller model has the advantage of being faster to evaluate and easier to store - both of which are crucial for real-time and embedded applications. While prior work on compressing neural networks have looked at methods based on sparsity, quantization and factorization of neural network layers, we look at the alternate approach of pruning neurons. Training Neural Networks is often described as a kind of `black magic', as successful training requires setting the right hyper-parameter values (such as the number of neurons in a layer, depth of the network, etc ). It is often not clear what these values should be, and these decisions often end up being either ad-hoc or driven through extensive experimentation. It would be desirable to automatically set some of these hyper-parameters for the user so as to minimize trial-and-error. Combining this objective with our earlier preference for smaller models, we ask the following question - for a given task, is it possible to come up with small neural network architectures automatically? In this thesis, we propose methods to achieve the same. The work is divided into four parts. First, given a neural network, we look at the problem of identifying important and unimportant neurons. We look at this problem in a data-free setting, i.e; assuming that the data the neural network was trained on, is not available. We propose two rules for identifying wasteful neurons and show that these suffice in such a data-free setting. By removing neurons based on these rules, we are able to reduce model size without significantly affecting accuracy. Second, we propose an automated learning procedure to remove neurons during the process of training. We call this procedure ‘Architecture-Learning’, as this automatically discovers the optimal width and depth of neural networks. We empirically show that this procedure is preferable to trial-and-error based Bayesian Optimization procedures for selecting neural network architectures. Third, we connect ‘Architecture-Learning’ to a popular regularize called ‘Dropout’, and propose a novel regularized which we call ‘Generalized Dropout’. From a Bayesian viewpoint, this method corresponds to a hierarchical extension of the Dropout algorithm. Empirically, we observe that Generalized Dropout corresponds to a more flexible version of Dropout, and works in scenarios where Dropout fails. Finally, we apply our procedure for removing neurons to the problem of removing weights in a neural network, and achieve state-of-the-art results in scarifying neural networks.

Understanding biology paves the way for discovering drugs targeting deadly diseases like cancer, and microscopy imaging is one of the most informative ways to study biology. However, analysis of large numbers of samples is often required to draw statistically verifiable conclusions. Automated approaches for analysis of microscopy image data makes it possible to handle large data sets, and at the same time reduce the risk of bias. Quantitative microscopy refers to computational methods for extracting measurements from microscopy images, enabling detection and comparison of subtle changes in morphology or behavior induced by varying experimental conditions. This thesis covers computational methods for segmentation and classification of biological samples imaged by microscopy. Recent increase in computational power has enabled the development of deep neural networks (DNNs) that perform well in solving real world problems. This thesis compares classical image analysis algorithms for segmentation of bacteria cells and introduces a novel method that combines classical image analysis and DNNs for improved cell segmentation and detection of rare phenotypes. This thesis also demonstrates a novel DNN for segmentation of clusters of cells (spheroid), with varying sizes, shapes and textures imaged by phase contrast microscopy. DNNs typically require large amounts of training data. This problem is addressed by proposing an automated approach for creating ground truths by utilizing multiple imaging modalities and classical image analysis. The resulting DNNs are applied to segment unstained cells from bright field microscopy images. In DNNs, it is often difficult to understand what image features have the largest influence on the final classification results. This is addressed in an experiment where DNNs are applied to classify zebrafish embryos based on phenotypic changes induced by drug treatment. The response of the trained DNN is tested by ablation studies, which revealed that the networks do not necessarily learn the features most obvious at visual examination. Finally, DNNs are explored for classification of cervical and oral cell samples collected for cancer screening. Initial results show that the DNNs can respond to very subtle malignancy associated changes. All the presented methods are developed using open-source tools and validated on real microscopy images.