Abstract

The environmental impact on the atmosphere due to the anthropogenic processes by the human activity has been consistently increasing since recent decades. Due to which, global environmental problems such as climate change, depletion of stratospheric ozone and increase in regional air pollution are witnessed. Currently, these are major issues even for international agenda and agreement like Kyoto Protocol and Climate Conference (total number of COP’s 26 so far across globe) which came to a conclusion to regulate the emissions of climate destructors such as Chlorofluorocarbons (CFCs) and other greenhouse gases namely CO2 and CH4. Hence, monitoring of these gases globally with a distributed dense observational network of CO2 and CH4 more accurately since global carbon cycle plays a major role in earth’s climate system. Ground-based measurements are representative of only small locations around observation locations and is difficult to extrapolate this information over regional areas. Due to various constraints, it is not possible to have many observations in any given area. Satellite remote sensing on the other hand is able to give information over regional areas and with relatively low cost. Satellite retrievals add synoptic and geospatial extent to ground based measurements. Due to the advantage of high spatial coverage, remote sensing of atmospheric greenhouse gases (GHGs) is needed to play a vital role to understand the climate change in which CO2 and CH4 are of major concern. ISRO started National Carbon Project under the CAP-IGBP which focused on robust observational network across the country. Ground-based Fourier Transform infrared (FTIR) spectrometers are used around the globe for the measurement of atmospheric trace gases by solar absorption spectroscopy. Precise measurements of global columnar concentrations of GHGs are essential to investigate the major regions of sources and sinks on the Earth. Columnar measurements are especially important in the tropics, as convection is consistently strong and as a result flux signal are only weakly seen in surface measurements. Satellite based measurements complement ground-based observations by producing frequent global measurements of atmospheric constituents. Though the surface fluxes on CO2 and CH4 are strengthened in the recent decade, precise ground -based columnar GHGs measurements were found to be a gap area. Hence, first of its kind FTIR (Model: IFS 125M) in the country is established at National Remote Sensing Centre (NRSC), Shadnagar to generate long-term records and to study the columnar GHGs variability along with their vertical transport induced changes on surface fluxes. Thus, this thesis covers the retrieval of dry columnar CO2 and CH4 concentrations from the ground based FTIR data. Observations from ground based FTIR instrument have been collected at NRSC, Shadnagar, Telangana since 2016. We get direct solar radiation spectra from FTIR at a high spectral resolution of 0.01 cm-1 during clear sky days. With the combination of InSb detector and CaF2 beam splitter, the observations were made in the 0.90- 5.50µm during 2016 and 2019. The solar spectra data are collected at 5 min interval during pre-monsoon season. The hourly mean and 5 min XCO2 is mostly ranging from 405 ppmv to 412 ppmv with maximum deviation of 7 ppmv during the study period. The diurnal variation of XCO2, XCH4 are clearly observed with diurnal amplitude of ~2 ppmv and ~10 ppbv respectively. With the present retrieval scheme, precision of the FTIR achieved are 0.17 % to 0.52 % for CO2 and 0.30 % to 0.77 % for CH4 while meeting the global standards. A correlation coefficient of ‘r’ of 0.80 is observed between FTIR retrieved XCO2 against Orbiting Carbon Observatory-2 (OCO-2) satellite retrieved XCO2. The potential reasons for obtaining ‘r’ of 0.80 are satellite footprint which is chosen optimally due to lack of co-located points in finer grid size and global retrieval algorithm is different from regional retrieval algorithm. Further this thesis also has attempted to retrieve the columnar CH4 from the hyperspectral imaging sensor, the Airborne Visible/Infrared Imaging Spectrometer-Next Generation (AVIRIS-NG) covering spectral range from visible (380 nm) to shortwave infrared (2510 nm) with a spectral resolution of 5±0.5 nm. Line-by-line radiative transfer algorithm (LBLRTA) is applied on the AVIRIS-NG airborne data collected during December 2015-January 2016. Using LBLRTA the columnar density of methane was estimated from AVIRIS-NG transmittance data. For Shadnagar site, comparative analysis is done with the ground based FTIR observations and retrieved XCH4 concentration during the same period. Thus, this study demonstrated the potentiality of the AVIRIS-NG hyper spectral imaging sensor data application in the atmospheric species retrievals. With this, the present thesis covered the importance of columnar GHGs obs

Abstract

Conversational humor (CH) is a sub-domain of humor where the participants (speakers or listeners) engage in different types of humor such as retorts or teasing for various purposes. There are shared complexities between this phenomenon and the larger domain of humor, but several features are unique to CH. To overcome these complexities, we focus on two aspects of comprehending CH. First, we formulate a schema for a methodological approach to annotating CH using a famous Telugu stage play, Kanyasulkam, as a medium of analysis. We then collect data on humorous and non-humorous conversations and perform experiments using the state-of-the-art NLP models to detect the occurrence of CH. Contemporary work that focuses on shedding light on the purposes of CH by interlocutors considers only a few techniques or types. These analyses show the speakers’ intention, but it does not construct CH’s structure. In our work, we devise a schema that includes a hierarchical approach with different levels such as monologue/dialogue, benign/non-benign, and identifies the types and techniques involved in CH as well. In the pursuit of teaching a machine a language, we must bear in mind that for a model to perform reasonably well in tasks of any domain, we must have ample data to feed the model. However, for low resource languages such as Telugu, this becomes a difficult task. For this reason, annotators are employed to help with providing metadata to a dataset of a specific domain so that the model can learn the patterns belonging to it and produce satisfactory results. To this end, a work of literature is fully annotated by A1 and A2 in our study. We then calculate the agreement between their annotations to give an objective measure of the validity of the hierarchical schema. Many researchers in NLP have provided profound insights with the aim of detecting humor in various forms of text (tweets, movie scripts, jokes). Nevertheless, few studies have focused on text classification of conversational humor due to the complexity of the domain. We attempt to facilitate research in this direction by using several popular models to classify humorous and non-humorous conversations automatically. Recurrent Neural Networks (RNNs) such as Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (BiLSTM) learn sentence embeddings and are then used to classify text. In contrast, we use Text Graphical Convolutional Networks (GCN) that simultaneously learns the class a conversation belongs to and word embeddings based on word co-occurrence and document-word relations. In order to make the most out of the well-acclaimed pre-trained models, we fine-tune FastText word embeddings and different BERT (Bidirectional Encoder Representations from Transformers) models to generate sentence embeddings. We further use these models to classify text and compare their performance based on standard evaluation metrics.

Abstract

Modern visual-inertial navigation systems (VINS) are faced with a critical challenge in real-world deployment: they need to operate reliably and robustly in highly dynamic environments. Current best solutions merely filter dynamic objects as outliers based on the semantics of the object category. Such an approach does not scale as it requires semantic classifiers to encompass all possibly-moving object classes; this is hard to define, let alone deploy. On the other hand, many real-world environments exhibit strong structural regularities in the form of planes such as walls and ground surfaces, which are also crucially static. We present RP-VIO, a monocular visual-inertial odometry system that leverages the simple geometry of these planes for improved robustness and accuracy in challenging dynamic environments. Since existing datasets have a limited number of dynamic elements, we also present a highly-dynamic, photorealistic synthetic dataset for a more effective evaluation of the capabilities of modern VINS systems. We evaluate our approach on this dataset, and three diverse sequences from standard datasets including two real-world dynamic sequences and show a significant improvement in robustness and accuracy over a state-of-the-art monocular visual-inertial odometry system. We also show in simulation an improvement over a simple dynamic-features masking approach. Our code and dataset are publicly available1.

Abstract

Risk due to induced ground deformations has been observed from past seismic events. From the geographical statistics, it is evident that 60% of the Indian land mass is vulnerable to earthquakes of MSK intensity of VII or higher. Ground motion intensity due to an earthquake changes as it disseminates through the soil media from bedrock to the surface. The ground response because of an earthquake mainly depends on the magnitude of the event, geological details of the location, surface topography, fault mechanism, propagation path length between source and the site, dynamic properties of the soil medium (Abrahamson and Shedlock, 1997) and other local site conditions. Ground characteristics may result in amplification (causing resonance) or ground motion attenuation. Frequency of the ground motion is controlled by amplification mechanism. Source, path and site characterization were considered as the deciding factors of seismic ground motion intensity. Local site conditions are also considered as triggering factors of liquefaction. Therefore, estimation of local site effects can be substantial in seismic hazard assessment of an area. Probabilistic seismic hazard assessment, dynamic site characterization and ground response analysis provide an indispensable base for quantifying the seismic hazard associated with an area. Though design engineers are provided with macro level solutions for geotechnical and geological problems such as hazard maps, earthquake catalogs etc., from the past literature as well as practical experiences it was evident that they are not reliable in predicting the exact strong ground motion. In addition to this rapid infrastructure development with poor construction practices in terms of quality, mushrooming of structures and increased population growth rate also makes the country very vulnerable for earthquake damage. Hence it is very important to carry out the dynamic site characterization and site specific hazard assessment considering geotechnical, geological, geophysical and seismotectonic features. In this thesis, an attempt has been made to carry out seismic hazard assessment studies in the study area i.e., Vishakhapatnam. Also as an application of the results from hazard assessment studies, earthquake resistant design of pile supported wharf has been carried out considering the local site conditions in the study area. Vishakhapatnam lies in east coast region of southern India, falls under seismic zone II (IS: 1893, 2016) according to seismic zonation map of India. It has two operational ports making it the financial capital of the state and prominent in terms of trade and economy. Addressing the seismicity in the study area, seismic activity in the Eastern Ghats mobile belt region of southern part of India and Bay of Bengal has increased due to subduction of Burma tectonic plate towards the Bay of Bengal, which resulted in reactivation of older inactive faults and also new faults development (Reddy and Chandrakala, 2004). Therefore, increasing seismic risk and importance of the study area has motivated the researchers to conduct detailed ground response analysis and estimation of local site effects. Probabilistic seismic hazard assessment (PSHA) has been carried out to quantify the seismicity associated with the study area. Extensive geotechnical data has been collected from various private and public organizations. Using the available geotechnical data 1D and 2D soil profiles have been generated. Rock outcrops and landfilled sites have been identified from the soil profiles. Equivalent linear ground response analysis has been carried out to estimate the peak ground acceleration at various locations in the study area using DEEPSOIL (Youssef, 2009). The surface PGA values obtained have been used in liquefaction hazard assessment using stress based method (SBM) and energy based method (EBM). Hazard maps were generated from the estimated values of PGA and factor of safety against liquefaction (FL) from both the approaches. The city has been characterized into different zones of varying hazard with respect to PGA and FL. Soft soils may amplify or de-amplify the ground motion resulting in increased or decreased ground shaking intensity and remains as one of the deciding factor in seismic hazard intensity (Neelima and Towhata, 2016). Hence to estimate the effect of local site conditions, microtremor testing has been carried out at 75 locations and predominant frequency map has been developed for the study area using H/V spectral ratio method (Nakamura, 1989). The dynamic parameters are further used as input in seismic design and analysis of pile supported wharf structure in the study area. As mentioned earlier, the study area is having two operational ports contributing to the economy of the state and the country. Therefore designing the port structures considering the seismicity and local conditions will be of great importance. As

Abstract

In this work, we propose Spatio-Temporal Laban Features: a novel motion descriptor based on Laban Theory and provide a platform to document, analyse and annotate performed dance. Dance is an indispensable part of our life. It has accompanied social and religious events, ceremonies, rituals, celebrations, etc throughout our history. However, dance evolves, it is very susceptible to creative changes by practitioners. Being a part of our cultural heritage, especially classical and folk dance, it needs to be preserved to track it’s evolution through time. With the boom in popularity of media platforms, multimedia content has permeated virtually every aspect of our lives. The volume of usergenerated content on the web has skyrocketed, and unlike before, dance is also leaving behind more trail than ever. However, the majority of such content is undocumented primarily due to high costs associated with the time and the resources required annotate such data manually or through sensors. We leverage the advancements in Machine Learning to reduce these costs and automate motion indexing. We create a semantically searchable dance database with support for Labanotation generation, automatic annotation, and retrieval. Our approach is built on a thorough review of Laban Theory which is a very popular annotation system based on movement analysis. We apply a pose estimation module to retrieve pose and generate Labanotation over recorded videos. Labanotation provides us with a comprehensive annotation system that we use to generate our novel motion descriptor and it can be further exploited to build an ontology as well. It is also very relevant for the preservation and digitization of online resources, especially for dance cultural heritage, which has been our focus in this work. We built a semi-automatic annotation model that generates semantic annotations over a big video database. We combine two publicly available ballet datasets and test our model on it. High-level concepts such as ballet pose and steps are used to make the semantic library. These also act as descriptive meta-tags for any ballet video making the videos retrievable using a semantic text or even a video query. The final platform thus allows a user to generate and study the form of any dance sequence using Labanotation as well as search for particular steps to learn from a video database.

Abstract

The increasing popularity of semantic web has led to the creation and enrichment of ontologies and other linked datasets using automated algorithms. Alongside these algorithms, quality evaluation of the enriched ontologies is necessary to ensure reliability and reuse. Current quality evaluation approaches seek to evaluate ontologies in either syntactic (degree of following ontology development guidelines) or semantic (degree of semantic validity of enriched concepts/relations) aspects. Current syntactic quality evaluation approaches often use pre-defined rules and priorities for evaluation, which provide limited flexibility and customisability to the end user. Any extension to such works in an attempt to adapt it for a specific task or a different goal also requires significant programming, further posing hindrances in extensibility, especially for non-programming ontology engineers. to create or update rules at run time. Semantic approaches, on the other hand, which often involve crowdsourcing, suffer in output quality with increase in size and diversity of crowd validators. Most current approaches use naive majority voting, with some using spam-filtering techniques. While these methods are useful for eliminating spammers, the accuracy of these works are debatable in more general scenarios where only all validators have some degree of knowledge but only few are absolute experts regarding the ontology domain. In these scenarios, it would be useful to have an algorithm that grades validators according to their expertise and weights their decisions proportionally to their experience. This work thus seeks to propose a semi-automated framework that seeks to achieve both syntactic and semantic quality evaluation in an integrated framework. The proposed framework consists of: a) SynEvaluator and b) SemValidator for evaluating the syntactic and semantic aspects of ontology quality respectively. SynEvaluator uses a novel theoretical rule-creation framework that aims to improve customisability and flexibility for the end user. It does so by allowing for dynamic, task-specific creation and updation of syntactic rules at runtime, all without any need for programming. SemValidator uses crowdsourcing for semantic validation. It also uses a novel algorithm, TweetExpert, to estimate the Twitter-based expertise of validators for weighing their decisions as a means of quality control. The efficacy and validity of the framework is shown empirically on multiple ontologies.

Abstract

In machine learning, tasks like making predictions using a model and learning model parameters can often be formulated as optimization problems. The feasibility of using a machine learning model depends on the efficiency with which the corresponding optimization problems can be solved. As such, the area of machine learning throws up many challenges and interesting problems for research in the field of optimization. While in some cases, it is possible to directly apply off-the-shelf optimization methods for problems in machine learning, in many other cases, it becomes necessary to develop optimization algorithms that are tailor-made for specific problems. On the other hand, developing optimization algorithms for specific problem domains can itself be helped by machine learning techniques. Learning optimization algorithms from data can help relieve tedious effort required to develop optimization methods for new problem domains. The challenge here is to appropriately parameterize the space of algorithms for different optimization problems. In this context, we explore the interplay between the areas of optimization and machine learning and make contributions in specific problems of interest that lie in the overlap of these fields. We look into the optimization challenge presented by multi-class classification with a large number of output classes and propose a partial-linearization based approach that intuitively generalizes over several popular optimization algorithms applicable to this task. It is popular to use measures like average precision (AP) and normalized discounted cumulative gain (NDCG) to evaluate a model for a classification task or a ranked retrieval task, however, they are not as popularly used for learning the parameters of these models. This is because of the difficulty in optimizing these non-decomposable loss functions. We propose a useful characterization of a large class of such loss functions that we show are amenable to efficient optimization and present an algorithm that can be used to efficiently optimize this class of loss functions. On the other hand, we explore the possible application of machine learning techniques for developing optimization algorithms, specifically, for combinatorial optimization problems. A wide class of approximate algorithms for NP-hard combinatorial optimization problems solve a continuousrelaxation of the original problem and then round the fractional solution to obtain an approximate discrete solution. The rounding procedures involved in such methods are non-trivial and generally have to be ingeniously designed for each task. We propose a framework that allows the learning of such rounding procedures from unsupervised data. Such a learning based approach permits rapid development of algorithms for novel optimization problems.

Abstract

The present-day smart homes are proliferated with innumerable smart home devices. The community is accepting smart home devices. More than 70% of homes in North America have an IoT device. The growing number of smart home devices has led to its variegated usage patterns. Unlike mobile phones, IoT devices are usually static, have less computation power and are shared among the members and visitors of a house. Several studies have shown different usage patterns of smart devices based on the relationships among the people in the house. However, devices in a house can be used to cause dangerous consequences to the property or its members. Smart homes do not usually appear to be places where security and privacy could be compromised. We challenge this premise by proving that when individuals in a smart home have malicious intentions, they can cause maximum damage to the people and devices of the house. Though the issue is not commonly addressed, one can use smart home devices to hamper the security or privacy of the house members. Since those who get access to these devices are trusted, it is imperative to manage the rights of these users. A strict access control policy is required to grant only required privileges to the users of the devices and monitor their activities to prevent abuse. Our work aims to regulate the usage of devices in a shared multi-user, multi-device setup by laying out an access control system for the home. Designing such a system has been previously attempted. However, those works failed to acknowledge the practicality of every device being used differently by each user. We assign each user a different role for every device based on the user’s suitability with that of the device. Existing methods to manage privileges in smart home systems have not considered allocating privileges to users based on (i) the relationship of the user with the device, (ii) the location and risk of the device and (iii) the current environment. In this work, we take a multi-perspective view on the problem of sharing fine-grained privileges of IoT devices among multiple users in a smart home. We propose the concepts of user role (subset of privileges specific to each device), tasks and security levels (level for each privilege) to allot the proper privileges to users. We limit the exploitation of privileges assigned to legitimate insiders of the house. Thus, our work matches the aspirations of previous surveys on building a comprehensive access control system to manage privileges in a shared smart home. We deployed our solution as a tool on a python-based Flask server that receives requests from users in the home through a REST API.Administrators in a smart home are often overwhelmed by the number of options, controls and environments given to them to manage. Since devices are varied, managing a smart home becomes a non-trivial task. To aid administrators to manage the environment in a hassle-free manner, we have also designed Vigile. This system verifies whether the aspirations of privilege distribution to users in the smart home matches with the actual allocation of privileges.

Abstract

The use of local detectors and descriptors in typical computer vision pipelines work well until variations in viewpoint and appearance change become extreme. Past research in this area has typically focused on one of two approaches to this challenge: the use of projections into spaces more suitable for feature matching under extreme viewpoint changes, and attempting to learn features that are inherently more robust to viewpoint change. In this paper we present a novel framework that combines learning of invariant descriptors through data augmentation and orthographic viewpoint projection. We propose rotation-robust local descriptors, learnt through training data augmentation based on rotation homographies, and a correspondence ensemble technique that combines vanilla feature correspondences with those obtained through rotation-robust features. Using a range of benchmark datasets as well as contributing a new bespoke dataset for this research domain, we evaluate the effectiveness of the proposed approach on key tasks including pose estimation and visual place recognition. Our system outperforms a range of baseline and state-of-the-art techniques, including enabling higher levels of place recognition precision across opposing place viewpoints, and achieves practically-useful performance levels even under extreme viewpoint changes. We reduce pose estimation error by 86.72% relative to state of the art. In particular, our integration of Visual Place Recognition (VPR) with SLAM by leveraging the robustness of deep-learnt features and our homography-based extreme viewpoint invariance significantly boosts the performance of VPR, feature correspondence and pose graph sub-modules of the SLAM pipeline. We demonstrate a localization system capable of state-of-the-art performance despite perceptual aliasing and extreme 180-degree-rotated viewpoint change in a range of real-world and simulated experiments. Our system is able to achieve early loop closures that prevent significant drifts in SLAM trajectories.

Abstract

With the number of protein sequences increasing rapidly, it becomes imperative to have a basic idea of the function and structure of a protein before the three-dimensional structure becomes available. Protein-drug interactions play essential roles in many biological processes and therapeutics. Prediction of the active binding site of a protein helps discover and optimise these interactions leading to the design of better ligand molecules. The secondary structure provides clues to the shape that the protein can be expected to take. It tells us whether an amino acid belongs to a coil turn, alpha-helix or beta-sheet structure. Deep Learning is a class of machine learning algorithms that progressively uses multiple layers to extract higher-level features from raw input. Deep learning methods eliminate feature engineering for supervised learning tasks by translating the raw inputs into intermediate representations that capture the more abstract and composite information, removing redundancies in the original input. The rapid adoption and success of deep learning algorithms in various sections of structural biology beckon deep learning algorithms for accurate binding site detection and secondary structure prediction. Protein-drug interactions play essential roles in many biological processes and therapeutics. Prediction of the active binding site of a protein helps discover and optimise these interactions leading to the design of better ligand molecules. The tertiary structure of a protein determines the binding sites available to the drug molecule. To quickly and accurately predict the binding site from sequence alone without utilising the three-dimensional structure is challenging. In the first study, a Residual Neural Network (leveraging skip connections) [1] is implemented to predict a protein’s most active binding site. An Annotated Database of Druggable Binding Sites from the Protein DataBank, sc-PDB [2], is used for training the network. Features extracted from the Multiple Sequence Alignments (MSAs) of the protein generated using DeepMSA, such as Position-Specific Scoring Matrix (PSSM), Secondary Structure (SS3), and Relative Solvent Accessibility (RSA), are provided as input to the network. A weighted binary cross-entropy loss function is used to counter the substantial imbalance in the two classes of binding and non-binding residues. The network performs very well on single-chain proteins, providing a pocket that has good interactions with a ligand. Secondary structure predictions predict three classes: C (Coil Turn), H (Alpha Helix) and E(Beta Sheet). In the second study, a Transformer (based on the multi-attention mechanism) network is used to train on the TR4590 dataset, which contains 4590 proteins with a sequence similarity cut off 25% and Xray resolution better than 2.0A. Ten models are trained across 10-fold cross-validation, and a weighted ˚ cross-entropy loss function is used. The ten trained models are run on the test set containing 1199 sequences. The mean of the probabilities of each class is taken, and then the class with the maximum probability is considered the class to which the amino acid belongs. The model achieves an accuracy of 82.51%.