Abstract

Many populous countries including India are burdened with a considerable backlog of legal cases. Development of automated systems that could process legal documents and augment and help legal practitioners can mitigate this. However, there is a dearth of high-quality corpora that is needed to develop such data-driven systems. The problem gets even more pronounced in the case of low resource languages such as Hindi. Additionally one of the most common and time sensitive cases handled by the courts are bail cases. In this Thesis, we first introduce the Hindi Legal Documents Corpus (HLDC), a corpus of more than 900K legal documents in Hindi. Documents are cleaned and structured to enable the development of downstream applications. We then introduce and tackle the task of bail prediction. We select the bail cases from our HLDC corpus and further extract the facts and arguments and judge’s summary. We experiment with a battery of models and propose a Multi-Task Learning (MTL) based model for the same. Our MTL model uses summarization as an auxiliary task along with bail prediction as the main task. The intermediate summarisation step is a novel introduction which serves dual purposes. First, it reduces the document size without compromising on the information. Since many transformer models have constraint on the input length, sending a summarised version of the documents allows us to overcome this barrier. Second, it builds towards explainable legal NLP systems as it allows us to identify salient sentences. This Thesis lays the foundation for research in Legal NLP for Hindi court documents. The multitude of legal NLP tasks and challenges are indicative of the need for further research in this area.

Abstract

Natural Language Processing (NLP) has evolved drastically in many areas due to the popularity and success of a variety of different Machine Learning and specifically Deep Learning techniques and their applications commercially as well as academcially in the last decade or so. Typically most of the work done in these fields globally has centred around solving tasks with the use of resource rich languages such as English as a base upon which models have been built. The large chasm this technological and academic revolution brings about, to a world that is divided in a multitude of regional cultures and languages that haven’t evolved at the same pace, creates a very real obstacle. We aim to play a part in bridging this gap that has been brought about with the creation of a rich resource for one such low resource language in Telugu, a Dravidian language. We aim to focus on curating and annotating a dialogue corpus as it allows us to approach a variety of tasks and holds various different real world applications including the rather pertinent current phenomenon of improving Human Computer Interactions especially in the context of low resource languages such as Telugu. It is important to also take note of the evolution of language and different means of communication as a function of time and one product of this observation is the increased usage of code-mixing in colloquial discourse. In order to develop robust models and accurately recreate Human-Human interactions with the use of computers, we need to be able work with and model on code-mixed data. Therefore another focus of our study is in the creation of a dialogue corpus that is heavily code-mixed with Telugu and English and the subsequent modelling of the dialogue system using deep learning techniques. In order to determine the resourcefulness of the Telugu-English code-mixed dialogue corpus that we have carefully curated, we put it to the test by performing two different widely studied NLP tasks in : Text-based Speaker Identification and Automatic Humour Recognition. We provide a comprehensive analysis and survey of modern deep learning techniques for the task of text-based speaker identification and present a detailed explanation of all the choices made while undertaking this study. The deep learning techniques implemented in this body of work include the likes of Convolutional Neural Networks, LSTMs, and Transformer based models.

Abstract

Social media has enabled individuals to easily communicate with others while freely sharing their ideas and thoughts by providing anonymity. Due to the substantial growth of user-generated web content in general and social media networks in particular, the amount of discriminatory content is also steadily increasing, leading to significant toxicity in abusive behaviours and online discourse. One type of such harmful content is hate speech, which directly attacks or promotes hate towards an individual member or a group based on their actual or perceived aspects of identity, such as ethnicity, religion, and discrimination. Developing models for this type of data can be difficult due to the fact that the data available on online media may be inadequate, unstructured, and highly imbalanced. In this thesis, we focus on discrimination, especially the form of discrimination perpetrated based on sex. Sexism, a form of oppression based on one’s sex, manifests itself in numerous ways and causes enormous suffering. During the last years, the role of women within online platforms has gained attention, unfortunately, because of the growing hatred and abuse against them. In view of the growing number of experiences of sexism reported online (accounts of sexism) and the increase in sexist posts in online conversations, automatically categorizing both sorts of posts can assist the fight against sexism, as it can facilitate effective analyses by gender studies researchers and government officials involved in policy making. In this thesis, we investigate the detection and categorization of both types of sexist posts. The existing work on sexism classification has certain limitations in terms of the categories of sexism used and/or whether they can co-occur. To the best of our knowledge, we work with considerably more categories of sexism than any prior work through our 23-class problem formulation. We present the first semi-supervised work for the multi-label classification of accounts describing any type(s) of sexism. We devise self-training based techniques tailor-made for the multi-label nature of the problem to utilize unlabeled samples for augmenting the labeled set. We identify high textual diversity with respect to the existing labeled set as a desirable quality for candidate unlabeled instances and develop methods for incorporating it into our approach. We also explore ways of infusing class imbalance alleviation for multi-label classification into our semi-supervised learning, independently and in conjunction with the method involving diversity. In addition to data augmentation methods, we develop a neural model which combines biLSTM and attention with a domain-adapted BERT model in an end-to-end trainable manner. Further, we formulate a multi-level training approach in which models are sequentially trained using categories of sexism of different levels of granularity. Moreover, we devise a loss function that exploits any label confidence scores associated with the data. We also propose a knowledge-based cascaded multi-task framework for fine-grained multi-label sexism classification that allows for shared learning across multiple tasks through common layers/weights and a combined loss function. We leverage several supporting tasks, including homogeneous and heterogeneous auxiliary tasks. Homogeneous tasks are set up without incurring any manual labeling cost and heterogeneous tasks are set up which have a high correlation with the accounts of sexism. Unlabeled accounts of sexism are utilized through unsupervised learning to help construct our multi-task setup. In addition, we incorporate a knowledge module within the framework to infuse domain-specific external knowledge features into the learning process. Further, we investigate transfer learning that employs weakly labeled accounts of sexism and transfers the learning to the multi-label sexism classification. We also devise objective functions that exploit label correlations in the training data explicitly. In order to evaluate the versatility of our approaches, we experiment with adapting them for the toxic comment classification dataset. Multiple proposed multi-task methods outperform the state-of-the-art multi-label sexism classification and toxic comment classification across five standard metrics. In addition to analysing the textual accounts describing sexism suffered or observed in a multi-label multi-class setup, we also analyze how sexism is expressed in online conversations, especially in the social networks, wherein the text perpetrates sexism. Moreover, we also aimed to address the problem of detecting sexism rather than only the categorization. Specifically, we explore the problem of detecting whether a tweet/gab post is sexist or not. We further discriminate the detected sexist post into one of the fine-grained sexism categories in a single-label multi-class manner since labeled data (text perpetrates sexism) is not available wherein

Abstract

Air pollution poses a threat to the lives of all living beings. Government authorities generally monitor pollution levels using a high-grade setup. The high-grade instruments are expensive and also require space for setup. A few low-cost sensors have been introduced for monitoring air quality. But, those sensors also have a few shortcomings. This thesis mainly focuses on the performance evaluation of low-cost PM sensors and an image processing-based technique to estimate air quality. Firstly, the performance of three new and popular low-cost particulate matter (PM) sensors, namely SDS011, Prana Air, and SPS30, for measuring PM2.5 and PM10 levels is evaluated against a standard reference Aeroqual Series-500. The test setup was exposed to PM concentrations ranging from 30 µg/cm3 to 600 µg/cm3 . The results were based on 1 min, 15 min, 30 min, and 1 hr average readings. The experiments were carried out in indoor as well as outdoor environments. The comparative evaluation was performed before and after calibration. The performance of these sensors is evaluated in terms of coefficient of determination (R2 ), coefficient of variation (Cv), and root mean square error (RMSE). A real-time Air Quality Index (AQI) estimation technique using images and weather sensors on Indian roads is also presented. A mixture of image features, i.e., traffic density, visibility, and sensor features, i.e., temperature and humidity, were used to predict the AQI. Object detection and localizationbased Deep Learning (DL) method and image processing techniques were used to extract image features. At the same time, an ML model was trained on those features to estimate the AQI. For this experiment, a dataset containing 5048 images was collected over four months from September-December 2021 using AQSense device that was developed in International Institute of Information Technology (IIIT), Hyderabad, and co-located AQI values across different seasons were collected by driving on the roads of Hyderabad city in India. The experimental results report an overall accuracy of 82% for AQI prediction. A few challenges faced during the measurement campaign are also discussed.

Abstract

Ever since secure multi-party computation was successfully showed [6], it has remained an interesting research field in the cryptography community. Multi-party computation is one of the most important primitive in modern cryptography, in which mutually distrusting parties want to compute a functionality over their private data without revealing any additional information about their data. It is well established that the information theoretic multi-party computation among n parties is possible if the number of corrupted nodes t is less than n 2 for semi-honest adversary and the number of corrupted nodes t is less than n 3 for malicious adversary. Despite the progress over past 3 decades, multi-party computation have many improvements to go through before it becomes viable for practical scenarios. Also, there happens to exist many areas which are little explored. In this dissertation we explore one such path where we rely on physical cryptographic assumption rather than the traditional assumptions. The traditional assumptions such as discrete log problem, RSA, etc. are always subject to non-existence of poly time algorithm for computational problems of certain nature. Also, these assumptions do not cover computationally unbounded adversary. The motivation to study such assumption is that the physical cryptographic assumptions can be characterized in a way that they are realistic as well as provide required security guarantees. In this thesis, we discover that channel asynchrony is one such assumption where the channel may not deliver the messages in the same order as they were sent. Usually we use solutions such as sequence number to order the messages in such case but for security such asynchrony happens to be a boon. It is also more efficient practically compared to the protocols in the information theoretic setting. This thesis also solves an open problem in the multi-party computation literature. “Is it possible to remove the dependency on the multiplicative depth of the circuit in the overall communication complexity of a secure MPC protocol while maintaining its efficiency and the restrictions on the number of corruptions such that 3ta + 2tp + tf < n?”. This work is based on the works of Goyal et al. and Hirt et al. Goyal et. al. solve a similar question for active adversarial setting instead of mixed adversary. Hirt et al. have also given the multiplicative depth dependent protocol for mixed adversarial setting. We combine these two works to answer the question stated above in affirmative. Middleboxes in a computer network system inspect and analyse network traffic to detect malicious communications, monitor system performance and provide operational services. However, encrypted traffic, which has become increasingly prevalent, hinders the ability of middleboxes to perform such services. A common practice in addressing this issue is by employing a “Man-in-the-Middle” (MitM)approach, wherein an encrypted traffic flow between two end-points is interrupted, decrypted and analyzed by the middleboxes. The MitM approach is straightforward and is used by many organisations, but there are both practical and privacy concerns. Practically, due to the cost of the MitM appliances and the latency incurred due to the encrypt-decrypt processes, enterprises continue to seek solutions that are less costly and less compute-intensive. There has also been discussion on the many efforts required to configure MitM. Besides, MitM violates end-to-end privacy guarantee, raising privacy concerns and potential issues on compliance especially with the rising awareness on user privacy. Furthermore, some of the MitM implementations were found to be flawed. Consequently, new practical and privacy-preserving techniques that enable inspection over encrypted traffic were proposed. In this thesis, we also systematically examine these techniques to compare their advantages, limitations and challenges. We categorise them into four main categories by defining a framework that consist of system architectures, use cases, trust and threat models. These are searchable encryption, access control, machine learning and trusted hardware. We first discuss the man-in-the-middle approach as a baseline, then discuss in details each of them, and provide an in-depth comparisons of their advantages and limitations. By doing so we describe practical constraints, advantages and pitfalls towards adopting the techniques. Following this, we give insights on the gaps between research work and practical implementation in the industries, which leads us to the discussion on the challenges and research directions.

Abstract

Natural Language Processing (NLP) systems have attracted substantial attention for their potential and propitious commercial value in recent years. Part-of-speech (POS) tagging is a well-studied task and is one of the essential and fundamental steps required for various NLP tasks. Numerous types of supervised, unsupervised, and semi-supervised POS taggers exist for different languages. Each of these POS taggers has its unique strengths. To leverage the combined benefits of these existing taggers, developing ensemble methods is a good option. Ensemble Methods are techniques that create multiple base models and then combine them to produce improved results and usually produce more accurate results than a single model. Our study investigates ensemble methods to exploit the complementary characteristics of Part-ofspeech taggers. We present an ensemble technique that enhances the performance of Supervised taggers with the help of Semi-supervised taggers. We propose a two-layer static ensemble with a decision tree classifier as the second layer. Outputs of the first level base POS taggers are chosen as new input features set to the second level decision tree. We use Semi-supervised context based lists (CBLs) tagger as one of the first layer taggers, which uses rich contextual information and helps in tagging both existing and unseen words and uses no domain knowledge, while supervised taggers give good performance for words present in the training model. Enhanced performance of our proposed ensemble approach over the base methods suggests that integrating these methods combines the strengths of base taggers. Later, the thesis focuses on developing a dynamic classifier selection approach suitable for POS tagging. In Dynamic Classifier Selection (DCS) techniques, the most appropriate set of classifiers is selected dynamically for each test sample. This selection is based on the performance of base classifiers on similar context words. In DCS methods, it is assumed that if a classifier performs well in the region of competence, that classifier is most likely to perform better for the given test instance. Word2Vec, a word Embedding algorithm trained over unlabeled data, is used to pick these similar context-based words from the validation set. Our proposed DCS approach on five languages outperforms the base POS taggers. These results suggest that we successfully selected the appropriate classifier for a given test word. Overall, we propose ensemble methods to address the Parts-of-speech tagging of resource-poor languages. These methods work with fewer annotated data by using unlabeled data without considering any domain-based knowledge. These ensemble methods can be an excellent choice to integrate and leverage the benefits of various classifiers and help the large number (6500+) of “resource-poor languages” that do not have much-annotated training data

Abstract

Machine Translation is the task where a machine generates a sentence in language T, given an input sentence in language S. The generated and input sentences must have similar semantics. In the current literature, there are various methods proposed to solve the task. Deep neural network based translation models (Neural Machine Translation (NMT)) have been shown to achieve state-of-the-art performance for multiple language pairs. However, neural methods struggle to perform well for lowresource languages. Low-resource languages are understudied, suffer from the data scarcity problem, and lack efficient language processing tools. Since neural methods require large amount of data for training effective models, less-complex statistical methods outperform these methods. Transformer-based NMT models have achieved state-of-the-art performance for various languages. Multiple parallel attention mechanisms that use multiple attention heads facilitate greater performance of the Transformer model for various applications e.g., NMT), text classification. However, transformer models do not perform well in low-resource conditions. In this thesis, we propose a novel Dynamic Head Importance Computation Mechanism (DHICM), which enhances the performance of the transformer model, especially, in low-resource conditions. In multi-head attention mechanism, different heads attend to different parts of the input. However, the limitation is that multiple heads might attend to the same part of the input, resulting in multiple heads being redundant. Thus, the model resources are underutilized. One approach to avoid this is to prune least important heads based on certain importance score. We focus on designing a head importance computation method to dynamically calculate the importance of a head with respect to the input. Our insight is to design an additional attention layer together with multi-head attention, and utilize the outputs of the multi-head attention along with the input, to compute the importance for each head. Additionally, we ad d an extra loss function to prevent the model from assigning same score to all heads, to identify more important heads and improvise performance. We analy zed performance of DHICM for NMT with different languages. Experiments on different datasets show that DHICM outperforms traditional Transformer-based approach by large margin, especially, when less training data is available. Code mixing is a phenomena of mixing two or more languages, prevalent in multilingual communities. In recent years, due to the prevalence of social media platforms, there has been a surge in the usage of code-mixed texts. Code-mixed texts are informal in nature, and do not necessarily follow pre-defined syntactic structures. Due to the informal nature, traditional NLP systems for monolingual languages, do not perform well for code-mixed inputs. Moreover, good quality code-mixed data is scarce, hence,

Abstract

Today human identification is a practice followed quite frequently in contexts ranging from forensic labs to high security organisations to almost every domestic household. One of the oldest and most accurate ways to establish the identity of an individual is through fingerprints. It is so because, biologically no two people (not even identical twins) have the same fingerprints. Hence it comes as no surprise that fingerprint biometrics has been a very active domain of research for a long time. Over the years, especially with the emergence of the deep learning era, the computational modules of a fingerprint recognition system have become bigger and bigger. Simultaneously, the actual hardware deploying fingerprint recognition systems has become more and more user friendly. These developments provide a contrast as one tries to incorporate the highly accurate but also extremely heavy modules of a fingerprint recognition system in the palm of our hands. In this thesis, we will perform a couple of studies which explore ways to optimize fingerprint recognition systems. This is done by trying to parallelize modules and eliminate semantically redundant computations. Along the way we will also obtain better performances as compared to the academic state of the art. Typical fingerprint recognition systems are comprised of a spoof detection module and a subsequent matching module, running one after the other. In the first study, we posit that both spoof detection and fingerprint matching are correlated tasks. Therefore, rather than performing the two tasks separately, we propose a joint model for spoof detection and matching to simultaneously perform both tasks without compromising the accuracy of either task. In practice, this reduces the time and memory requirements of the fingerprint recognition system by 50% and 40%, respectively. On the other hand, fingerprint matching task itself can be solved in multiple ways. Fingerprint feature extraction is a core step in the fingerprint matching task and it can be solved by using a global as well as a local approach. State of the art global approaches use huge deep learning models to process the full fingerprint image at once which makes the corresponding approach memory intensive. Whereas local approaches involve minutiae based patch extraction and multiple feature extraction steps which make the corresponding approach time-intensive. But both these approaches provide useful and sometimes exclusive insights for solving the problem. Hence using both kinds of approaches together for extracting fingerprint representations is semantically useful but quite inefficient. Hence, in the second study, we use a convolutional transformer based approach with an in-built minutiae extractor. This study provides an efficient solution for feature extraction that parallely provides a global as well as a local represen tation of a fingerprint. This not only provides a replacement for using a heavy combination of two independent local and global approaches but also provides a further 54.41% speedup over the local approach and a 57.93% conservation in memory as compared to the global approach. Both of these studies work towards the common goal of optimizing fingerprint recognition systems and hence provide a significant advantage in the context of such systems deployed on resource constrained devices like smartphones.

Abstract

Recent work in NLP has focused on applying powerful neural sequence models to various learning problems. These neural models excel at extracting statistical patterns from large amounts of data but struggle to learn patterns or reason about language from only a few examples. We ask the question: Can we learn explicit rules that generalize well from only a few examples? We explore this question by viewing linguistic rules as programs that operate on linguistic forms. This allows us to tackle the problem of learning linguistic rules using program synthesis – a method to learn rules in the form of programs in a domain-specific language (DSL). We develop a synthesis model to learn phonological rules as programs in a DSL. In addition to being highly sample-efficient, our approach generates human-readable programs and allows control over the generalizability of the learned programs. We test the ability of our models to generalise from only a few training examples using our new dataset of problems from the Linguistics Olympiad. These problems are tasks from contests for high school students around the world that require inferring linguistic patterns from a small number of given examples. These problems are a challenging set of tasks that require strong linguistic reasoning ability. Having shown that program synthesis can be used to learn phonological rules in highly data-constrained settings, we use the problem of phonological stress placement as a case to study how the design of the domain-specific language influences the generalisation ability when using the same learning algorithm. We find that encoding the distinction between consonants and vowels results in much better performance, and providing syllable-level information further improves generalization. Program synthesis, thus, provides a way to investigate how access to explicit linguistic information influences what can be learned from a small number of examples.

Abstract

Virtual Labs are a collection of college-level online experiments in the science and engineering disciplines. Virtual Labs provides a remote experimentation platform for hands-on learning. It is essential to analyze and improve the web performance of the Virtual Labs platform to enhance the user experience as well as to cope with the increased usage. This study identifies and implements approaches to improve the performance of Virtual Labs experiments. We gather the web performance data using two tools, Google Analytics and Webpagetest, analyze and implement the performance improvement approaches. The implemented approaches have reduced the load time of the Virtual Labs experiments by 80% and reduced the load time of the page’s main content from 6.1 seconds to under 1.5 seconds. The loading and rendering metrics have been optimized, and the efficiency of the experiment loading has increased.