Deep generative models are essential to Natural Language Processing (NLP) due to their outstanding ability to use unlabelled data, to incorporate abundant linguistic features, and to learn interpretable dependencies among data. As the structure becomes deeper and more complex, having an effective and efficient inference method becomes increasingly important. In this thesis, neural variational inference is applied to carry out inference for deep generative models. While traditional variational methods derive an analytic approximation for the intractable distributions over latent variables, here we construct an inference network conditioned on the discrete text input to provide the variational distribution. The powerful neural networks are able to approximate complicated non-linear distributions and grant the possibilities for more interesting and complicated generative models. Therefore, we develop the potential of neural variational inference and apply it to a variety of models for NLP with continuous or discrete latent variables. This thesis is divided into three parts. Part I introduces a <b>generic variational inference framework</b> for generative and conditional models of text. For continuous or discrete latent variables, we apply a continuous reparameterisation trick or the REINFORCE algorithm to build low-variance gradient estimators. To further explore Bayesian non-parametrics in deep neural networks, we propose a family of neural networks that parameterise categorical distributions with continuous latent variables. Using the stick-breaking construction, an unbounded categorical distribution is incorporated into our deep generative models which can be optimised by stochastic gradient back-propagation with a continuous reparameterisation. Part II explores <b>continuous latent variable models for NLP</b>. Chapter 3 discusses the Neural Variational Document Model (NVDM): an unsupervised generative model of text which aims to extract a continuous semantic latent variable for each document. In Chapter 4, the neural topic models modify the neural document models by parameterising categorical distributions with continuous latent variables, where the topics are explicitly modelled by discrete latent variables. The models are further extended to neural unbounded topic models with the help of stick-breaking construction, and a truncation-free variational inference method is proposed based on a Recurrent Stick-breaking construction (RSB). Chapter 5 describes the Neural Answer Selection Model (NASM) for learning a latent stochastic attention mechanism to model the semantics of question-answer pairs and predict their relatedness. Part III discusses <b>discrete latent variable models</b>. Chapter 6 introduces latent sentence compression models. The Auto-encoding Sentence Compression Model (ASC), as a discrete variational auto-encoder, generates a sentence by a sequence of discrete latent variables representing explicit words. The Forced Attention Sentence Compression Model (FSC) incorporates a combined pointer network biased towards the usage of words from source sentence, which significantly improves the performance when jointly trained with the ASC model in a semi-supervised learning fashion. Chapter 7 describes the Latent Intention Dialogue Models (LIDM) that employ a discrete latent variable to learn underlying dialogue intentions. Additionally, the latent intentions can be interpreted as actions guiding the generation of machine responses, which could be further refined autonomously by reinforcement learning. Finally, Chapter 8 summarizes our findings and directions for future work.

This thesis focuses on four main research themes namely shot boundary detection, fast frame alignment, activity-driven video summarisation, and highlights based video annotation and retrieval. A number of novel algorithms have been proposed to address these issues, which can be highlighted as follows. Firstly, accurate and robust shot boundary detection is achieved through modelling of cuts into sub-categories and appearance based modelling of several gradual transitions, along with some novel features extracted from compressed video. Secondly, fast and robust frame alignment is achieved via the proposed subspace phase correlation (SPC) and an improved sub-pixel strategy. The SPC is proved to be insensitive to zero-mean-noise, and its gradient-based extension is even robust to non-zero-mean noise and can be used to deal with non-overlapped regions for robust image registration. Thirdly, hierarchical modelling of rush videos using formal language techniques is proposed, which can guide the modelling and removal of several kinds of junk frames as well as adaptive clustering of retakes. With an extracted activity level measurement, shot and sub-shot are detected for content-adaptive video summarisation. Fourthly, highlights based video annotation and retrieval is achieved, in which statistical modelling of skin pixel colours, knowledge-based shot detection, and improved determination of camera motion patterns are employed. Within these proposed techniques, one important principle is to integrate various kinds of feature evidence and to incorporate prior knowledge in modelling the given problems. High-level hierarchical representation is extracted from the original linear structure for effective management and content-based retrieval of video data. As most of the work is implemented in the compressed domain, one additional benefit is the achieved high efficiency, which will be useful for many online applications. / EU IST FP6 Project

This thesis focuses on four main research themes namely shot boundary detection, fast frame alignment, activity-driven video summarisation, and highlights based video annotation and retrieval. A number of novel algorithms have been proposed to address these issues, which can be highlighted as follows. Firstly, accurate and robust shot boundary detection is achieved through modelling of cuts into sub-categories and appearance based modelling of several gradual transitions, along with some novel features extracted from compressed video. Secondly, fast and robust frame alignment is achieved via the proposed subspace phase correlation (SPC) and an improved sub-pixel strategy. The SPC is proved to be insensitive to zero-mean-noise, and its gradient-based extension is even robust to non-zero-mean noise and can be used to deal with non-overlapped regions for robust image registration. Thirdly, hierarchical modelling of rush videos using formal language techniques is proposed, which can guide the modelling and removal of several kinds of junk frames as well as adaptive clustering of retakes. With an extracted activity level measurement, shot and sub-shot are detected for content-adaptive video summarisation. Fourthly, highlights based video annotation and retrieval is achieved, in which statistical modelling of skin pixel colours, knowledge-based shot detection, and improved determination of camera motion patterns are employed. Within these proposed techniques, one important principle is to integrate various kinds of feature evidence and to incorporate prior knowledge in modelling the given problems. High-level hierarchical representation is extracted from the original linear structure for effective management and content-based retrieval of video data. As most of the work is implemented in the compressed domain, one additional benefit is the achieved high efficiency, which will be useful for many online applications.

Video is a large part of today’s society where surveillance cameras represent the biggest source of big data, and real-time entertainment is the largest network traffic category. There is currently a large interest in analysing the contents of video where video analysis is mainly conducted by people. This increase in video has for instance made it difficult for professional editors to analyse movies and series in a scalable way, and alternative solutions are needed. The media technology company June, want to explore scalable alternatives for extracting metadata from video. With recent advances in Machine Learning and the rise of machine-learning-asa-service platforms, June wished more specifically to explore how these Machine Learning services can be utilised for extracting metadata from videos, and from it construct a summary regarding its contents. This work examined Machine Learning as an option for scalable video summarisation which resulted in developing and evaluating an application that utilised transcription, summarisation, and translation services to produce a text based summarisation of video. Furthermore to examine the services current state of affairs, multiple services from different providers were tested, evaluated and compared to each other. Lastly, in order to evaluate the summarisation services an evaluation model was developed. The test results showed that the translation services were the only service that produced good results. Transcription and summarisation performed poorly in the tests which renders the suggested solution of combining the three services for video summarisation as impractical. / Video är en stor del av dagens samhälle där bland annat övervakningskameror är den största källan av data och underhållning i realtid är den kategori som står för mest nätverkstrafik. Det finns i dagsläget ett stort intresse i att analysera innehållet av video, denna videoanalys utförs även främst av människor. Ökningen av video har gjort det svårt för exempelvis professionella redaktörer att hinna analysera filmer och serier och mer skalbara alternativ behövs. Mediaföretaget June vill utforska alternativ för att extrahera metadata från video på ett skalbart sätt. Med de senaste framstegen inom maskininlärning och framväxten av machine-learningas-a-service plattformar, önskar June mer specifikt att utforska hur maskininlärning kan nyttjas för att extrahera metadata från video och med det konstruera en sammanfattning av innehållet. Det utförda arbetet undersökte maskininlärning som skalbart alternativ för att kunna sammanfatta videos innehåll. Arbetet resulterade i utvecklandet samt utvärderingen av en applikation som nyttjade maskininlärningstjänster för transkribering, sammanfattning samt översättning för att producera en textbaserad sammanfattning av videos innehåll. För att utvärdera tjänsternas nuvarande tillstånd så testades samt utvärderades tjänster från olika leverantörer för att sedan jämföras mot varandra. Slutligen framtogs en egenutvecklad modell för att kunna utvärdera tjänsterna för sammanfattning. Testresultaten visade att tjänsterna för översättning var de enda tjänsterna som gav bra resultat. Tjänsterna för transkribering och sammanfattning gav dåliga resultat vilket gör den föreslagna lösningen av att kombinera de tre tjänsterna för att sammanfatta videoinnehåll som opraktisk.

Students who study through the medium of a second language often have reading/listening comprehension and general study problems. This study focuses on particular aspects of these problems only, namely, identification of main ideas, summarisation and note-taking. The aim of this study was w determine the effect of teaching L2 students a combination of metacognitive and cognitive strategies for reading and listening comprehension (the main idea, summarising and note-taking by means of dictation). An intervention programme was designed in order to teach students these skills. There were ten students in both the experimental and the control groups. Both groups were assessed before and after the intervention programme. The findings reveal that the intervention was successful, with the experimental group showing greater improvement than the control group. The findings of this study have implications for second language tertiary learning and teaching theory and practice / Linguistics and Modern Languages / M.A. (Linguistics)

Students who study through the medium of a second language often have reading/listening comprehension and general study problems. This study focuses on particular aspects of these problems only, namely, identification of main ideas, summarisation and note-taking. The aim of this study was w determine the effect of teaching L2 students a combination of metacognitive and cognitive strategies for reading and listening comprehension (the main idea, summarising and note-taking by means of dictation). An intervention programme was designed in order to teach students these skills. There were ten students in both the experimental and the control groups. Both groups were assessed before and after the intervention programme. The findings reveal that the intervention was successful, with the experimental group showing greater improvement than the control group. The findings of this study have implications for second language tertiary learning and teaching theory and practice / Linguistics and Modern Languages / M.A. (Linguistics)