Abstract

Cloud storage is central to working of any data center. With the explosion of data it has further taken a center stage. The per gigabyte cost of cloud storage has drastically gone down over the last 5 years. But while the demand for compute power increases in additive manner, the accompanying provisioned storage increases in multiplicative manner. To add to that for provisioning of reliable data, any storage system would need to provide features such as scalability, reliability and de-duplication out of box. Providing such features on bare metal storage is expensive and would make any such system out of reach of most of the end users. The advent of Software Defined Storage has proved to be very helpful to solve this problem. Software Defined Storage frameworks typically decouple the task of managing and provisioning data from the underlying hardware. Hence the data plane can reside on cheap commercial hardware making it accessible to end users. But different Software Defined Storage provide different novelties out of box. The myriad of solutions provided makes it difficult for users to choose one over of the other. We try to solve this problem heterogeneity of storage systems by studying various ideas that can potentially help us to leverage novel features of different storage systems. Specifically we study four ideas which can potentially help us combine the heterogeneous storage systems. • As first case study, IO Workload assignment in a heterogeneous Software Defined Storage environment using offline statistical modelling to estimate performance measures such as Throughput, IOPS, et al. As a proof of concept we use support vector regression to estimate the performance of individual IO Workloads on each available Software Defined Storage(SDS) system for optimal assignment and build a working prototype comprising of HDFS, GlusterFS, and Ceph. • As second case study we tackle the same problem of IO Workload assignment using online statistically modelling. Finding ground truth for such a system can be cumbersome. To this end we use hierarchical clustering algorithms to down sample the data to find canonical data points and determine which data points are worthy of finding ground truth. • As third case study, we study MultiStack, which is a cloud orchestration platform for provisioning job in a hybrid cloud environment. We leverage such platform to provision storage in a hybrid cloud environment. As a POC we use the platform to provision storage across Amazon Web Service and Openstack. The performance measure we try to optimize here for is the job completion time of IO workloads. • As fourth case study, we try to go to a more basic key value pair domain. We try to leverage multiple hash rings to see if the load distribution characteristics of the DHTs change. We establish that overall it does not lead to substantial change in the load distribution characteristics but in the process we find it improves the load distribution in case of adversarial traffic. This work is essentially a collection of trying to find various avenues in the cloud storage ecosystem where heterogeneity can be leveraged to get betterment in terms of cloud performance metrics. In places where we establish such a gain is not substantial we see heterogeneity can help us gain betterment of other metrics like load distribution in case of adverserial traffic.

Abstract

Over the past two decades, as a part of data mining, several approaches and frameworks have been developed for extracting different kinds of interesting knowledge from multiple types of data. In the Internet era, the data mining algorithms are being employed to build massive information/knowledge driven systems such as search engines, recommendation systems and decision support systems. Due to inherent value of new kinds of knowledge in improving business performance, there is a significant thrust towards developing better knowledge extraction approaches from different data sets for building next-generation information- and knowledge-based systems. The term diversity arises repeatedly in economics, natural sciences, social sciences, and information science. Related to information science, recently, the research efforts are being made to develop approaches to improve the diversity performance of recommender system (RS), and web search query recommendations (QRs). The problem of improving diversity is being encountered in several real-life applications. In the domain of a computer network. Finding diverse node in a network which is connected to multiple clusters of nodes has been studied. In social networks, the diversity has characterized as how diverse a given node connects with its peers. The notion of diversity has been investigated within the context of text documents using Rao's measure. Normally, the data mining approaches extract the knowledge structures like significant patterns (frequent patterns, and association rules), clusters, and classes. In the area of pattern mining, the process of frequent pattern (FP) extracts the interesting information about the association among the items in transactional data sets. Several algorithms have been proposed to extract FPs efficiently from the transactional data sets. Several approaches have been proposed to mine the subset of interesting FPs according to the needs and demands of the application by defining new interestingness measures like periodicity, pattern-length, and cost (utility) of a FP. In this thesis, we propose a model of diverse frequent patterns (DFPs) and show that it can be employed to improve the performance of RS and web search QR systems. We have proposed a model to compute the diversity of pattern (set of items). Given a domain, set of items can be grouped into a category, and a pattern may contain the items which belong to multiple categories. In several applications, it may be useful to distinguish between the pattern having items belonging to multiple categories and the pattern having items belonging to one or a few categories. The existing FP mining approaches do not distinguish the patterns based on the categories of the items in the pattern. The notion of emph{diversity} captures the extent of the items in the pattern belong to different categories. We propose a framework to rank the pattern by analyzing the extent the items of the pattern belong to different categories in the corresponding concept hierarchy. We defined a measure to capture the extent of diversity to rank the patterns which is called as diverse rank (drank). A pattern which satisfies the threshold value of drank is called a diverse pattern (DP). Also, a pattern which satisfies the threshold values of drank and support are called diverse frequent pattern (DFP). We propose two frameworks to compute the diversity of the pattern: one is based on the balanced concept hierarchy of the items of the pattern and the other is based on the unbalanced concept hierarchy of the items of the pattern. Through experimental results, we show that the knowledge of DFPs is different from the knowledge of FPs. Based on the concept of DFP, we have proposed an improved association rule (AR) based RS approach by observing the fact that providing high accuracy of recommendation alone cannot meet user satisfaction in RS. In the literature, efforts have been made to improve the variety/diversity of recommendations for higher user satisfaction. In the existing approaches, the accuracy performance has been compromised for the sake of improving the variety of recommendations. The proposed approach is aimed to improve both the diversity as well as the accuracy of AR based RSs. The proposed RS approach computes the drank for the pattern of AR based on the concept hierarchy formed by the corresponding items. The recommendations are made based on ARs with high confidence and high drank. The experimental results on the real-world MovieLens data set show that the proposed RS approach improves the performance of the existing AR based RS with better diversity without compromising the accuracy. Further, by exploiting the concept of DFP, we have investigated an improved approach for search QRs. Search engines use a QR technique to recommend a different set of queries to improve the user satisfaction on search results. The existing approaches recommend queries that are similar to the user's query. Some

Abstract

Payload Transport is one of the newly explored areas, where a robot or a formation of robots are used to transport a payload from one point to another. Multiple approaches to payload transportation have been carried out using either ground robots or aerial vehicles like drones. A single robot or a drone has an advantage in terms of its low cost and less dependency on other robots but has a single point of failure. Hence various coordinated multi-robot systems came to existence to overcome the disadvantage. We focus on building a multi-robot system which moves in a formation carrying a load from one place to another. We support failure resiliency where if any robot fails, there is always a backup robot available to replace the failed robot. In this thesis, we present a mechanism to handle the failures in a multi-robot system by robot replace- ment. In our work, we use multiple non-holonomic differential drive robots which moves in a formation following a known trajectory. A formation here refers to a group of robots carrying a payload on them while maintaining the desired shape. A decentralized leader-follower based formation control law is used to move the robots in a formation. There are recharging stations present along the route taken by the robots. When the robots in formation reach a recharging station, a check is performed based on the remaining robots battery capacity. If any of the robots in the formation has low battery capacity, it therefore, cannot continue the task of transportation and is replaced by a charged robot present at the recharging station. In this work, we focus on building a custom robot that is capable of carrying a payload with a lifting mechanism and an accurate localization. The kinematics and dynamics are developed for a differential drive, non-holonomic robot. We also developed a friction and battery model and derived a relation to between the robot dynamics and the power capacity of the battery. Further, we presented a novel concept which ensures that the multi-robot system lasts much longer than the battery life of an individual robot. Each robot is considered having two operating states: Charg- ing and Discharging. A robot in charging state means that the robot is present at the recharging station or is fully charged, waiting the replacement to happen and does not contribute to payload transport, whereas a robot in discharging state means a robot in formation transporting a payload and therefore is under continuous discharge. A system-level optimization is presented to decide upon these operational state (charging or discharging) of each robot in the system. The optimization provides the ids of the robots that need replacement at a particular recharge hub. We also present an algorithm for robot replacement to increase the running life of the Multi-Robot Payload Transport System. We also introduce the concept of robot battery recharge hubs which are present along the trajectory. Robots in the formation can be replaced at these hub locations with charged robots using a replacement mechanism. The robot ids obtained by the optimization are considered for replacement. The replacement algorithm is used to replace the low battery robot with the charged robot at the nearest recharge hub. We compare the optimization approach with a baseline approach and show that the Multi-robot system carrying a payload lasts longer using the optimization approach than the baseline. In a baseline approach, a certain threshold is considered (30%, 40% of the total battery, in our work), and the robot below this threshold will be considered for replacement. We showcase the efficacy of the proposed approach over the baseline approach through simulations and experiments with real robots.

Abstract

With more than a trillion web pages, there is a plethora of content available for consumption. Search Engine queries invariably lead to overwhelming information, parts of it relevant and others irrelevant. Often the information provided can be conflicting, ambiguous, and inconsistent, which can have serious consequences for people who increasingly rely on web sources for information related to security, health, academia, etc. Prior Research stresses upon the idea that traditional Search Engine Optimization techniques tend to focus on making top-ranked results more and more relevant and mostly depend on the user’s personal information and site popularity. Regardless, people often use two divergent terminologies- credibility and popularity, interchangeably. Credibility, an important quality characteristic of web pages is questionable in many cases and tends to be non-uniform. Credibility refers to the degree to which a website could be relied upon. Principally, credibility can be thought of as a compass for guiding us safely through a world of uncertainty, risk and moral hazards. Novice users who use search engines do not know how to start and lack sufficient knowledge for finding the best possible results. For a novice user, surface features such as fonts, colour, images and other layouts of the web page create the first impression on credibility [19]. While for most of the regular users of the internet, content relevance, information source, evolution of content and other fine-grained features of credibility decide web page usage. In past, researchers have proposed approaches for credibility assessment and enumerated features influencing the credibility of web pages. Assessment of few of those features can be automated using existing literature and contemporary knowledge; while others still need human intelligence. Web have been expanding since its inception, with which various kinds of web pages are emerging, categorized as genre of web page, for example– Help, Article, Discussion, etc. Depending on the genre of web page, the importance of credibility features such as web page date time modified, grammar, image to text ratio, in and out links, and other web page features may differ for assessment. Therefore, credibility without factoring genre of web page can lead to incorrect assessment. We conducted a crowd-source survey over multiple channels, where we asked participants to mark individual importance of web page elements(features) across different web genres, on a Likert scale of 1 to 4. The surveyed results implied that the importance of each feature vary across genres, which therefore supported our argument about the need for genre-aware credibility assessment of a web page. In this work, we propose an automated approach for credibility assessment of web page, where genre is also identified within to give human experts alike assessment results. We design a framework (called W EBCred) based on our proposed approach which accommodates various individual structures like – crawling, genre classification, normalization, scoring, etc. and keep them independent from each other to facilitate further extensib lilty. The proposed framework allows the addition of new genres, features and alter weightages providing flexibility for user intervention. To validate our proposed approach, we developed an Open-Source tool, which is capable of genre identification along with extraction and normalization of selected feature instance values to calculate a credible score (called GCS) of every web page. Few of these features were new and their extraction methodologies are defined by ourselves, as they are not explicit. Our tool is fully automated, such that it assess the Genre Credibility Score (GCS) of a given web page without any human aid. The source code [9] of developed tool is available over Github for further extension, and is deployed [7] on web for testing. We carried out extensive experiments to establish the effectiveness of our approach. We experimented our approach with ‘Information Security’ dataset having 8,550 URLs with 171 features across 7 genres. This dataset has been used for crowdsourced survey, training genre classification model and normalizing extracted feature instance values. The supervised learning algorithm, Gradient Boosted Decision Tree classified genres with 88.75% testing accuracy over 10 fold cross-validation, which overcome the current benchmark (about 80%). The calculated GCS based on identified genres by our trained model, correlated 69% with crowdsourced Web Of Trust (W OT ) score and 13% with algorithm based Alexa ranking for selected ‘Information Security’ web pages. As a further validation of our trained model and overall approach, we tested our trained model on separate ‘Health’ domain web pages, which correctly classify genres with 82.26% accuracy. Further the calculated GCS for ‘Health’ web pages correlates 59% with W OT and 23% with Alexa ranking. C

Abstract

The thesis is a novel attempt to propose a methodology which could help create and profile any semi-open organisation using publicly available data about its members and how they communicate with each other. The thesis is divided into two parts. In the first part, we lay down 5 methods through which we can create a virtual image of any organisation. The second part gives out a practical application of the proposed methodology where we find insights into an organisation and the movement of its members. We use the data available about an organisation in the public domain, from their websites, newspapers and other open source data. To illustrate our methodology, we use the instance of the Sangh Parivar with its affiliates along with the parent body, the Rashtriya Swayamsevak Sangh to perform a comprehensive analysis. We performed Social Network Analysis on some of the sub-fronts of the network using the method of co-occurrence to deduce the relationship between different organs of the Sangh Parivar. The results show a significant change in the way many sub-organisations are placed in the network with varying influence. We used another step of our methodology to show a Path Dependency in the Bharatiya Janata Party by studying the trail formed by the elected Members of the Parliament in the last 30 years. Using the connections these elected MPs make with the members from outside the political fraternity along with the Path Dependency and their relationship with the other fronts of the Sangh Parivar, we attempt to give out a complete picture of the members of the people associated with the Sangh Parivar using basic computational tools and publicly available data.

Abstract

Deep learning models have evolved remarkably, and are pushing the state-of-the-art in various problems across domains. At the same time, the complexity and the amount of resources these DNNs consume has greatly increased. Today’s DNNs are computationally intensive to train and run, especially Convolutional Neural Networks (CNNs) used for vision applications. They also occupy a large amount of memory and consume a large amount of power during training. This poses a major roadblock to the deployment of such networks, especially in real-time applications or on resource-limited devices. Two methods have shown promise in compressing CNNs: (i) Binarization and (ii) Pruning. We explore these two methods in this thesis. The first method to achieve improvements in computational/spatial efficiency is to binarize (1-bit quantize) the weights and activations in a network. However, naive binarization results in accuracy drops for most tasks. In this work, we present a Distribution-Aware approach to Binarizing Networks (DABN) that allows us to retain the advantages of a binarized network, while improving accuracy over binary networks. We also develop efficient implementations of DABN across different architectures. We present a theoretical analysis of DABN to show the effective representational power of the resulting layers, and explore the forms of data they model best. Experiments on popular sketch datasets show that DABN offers better accuracies than naive binarization. We further investigate the question of where to binarize inputs at layer-level granularity and show that selectively binarizing the inputs to specific layers in the network could lead to significant improvements in accuracy while preserving most of the advantages of binarization. We analyze the binarization tradeoff using a metric that jointly models the input binarization error and computational cost. We introduce an efficient algorithm to select layers whose inputs are to be binarized. We discuss practical guidelines based on insights obtained from applying the algorithm to a variety of models. Experiments on the Imagenet dataset using AlexNet and ResNet-18 models show 3-4% improvement in accuracy over fully binarized networks with minimal impact on compression. The improvements are even more substantial on sketch datasets like TU-Berlin, where we match state-of-the-art accuracy, getting more than 8% increase in accuracies over binary networks. We further show that our approach can be applied in tandem with other forms of compression that deal with individual layers or overall model compression (e.g., SqueezeNets). In contrast to previous binarization approaches, we are able to binarize the weights in the last layers of a network, which enables us to compress a large fraction of additional parameters. The second method explored is pruning. We investigate pruning neural networks from a graph-theoretic perspective. Efficient CNN designs like ResNets and DenseNet were proposed to improve accuracy vs efficiency trade-offs. They essentially increased the connectivity, allowing efficient information flow across layers. Inspired by these techniques, we propose to model connections between filters of a CNN using graphs which are simultaneously sparse and well-connected. Sparsity results in efficiency while well-connectedness can preserve the expressive power of the CNNs. We use a well-studied class of graphs from theoretical computer science that satisfies these properties known as Expander graphs. Expander graphs are used to model connections between filters in CNNs to design networks called XNets. We present two guarantees on the connectivity of X-Nets: (i) Each node of a layer influences every node in a layer O(logn) steps away, where n is the number of layers between the two layers (ii) The number of paths between two sets of nodes is proportional to the product of their sizes. We also propose efficient training and inference algorithms, making it possible to train deeper and wider X-Nets effectively. Expander based models give a 4% improvement in accuracy on MobileNet over grouped convolutions, a popular technique which has the same sparsity but worse connectivity. X-Nets give better performance trade-offs than the original ResNet and DenseNet-BC architectures. We achieve model sizes comparable to state-of-the-art pruning techniques using our simple architecture design, without any pruning.

Abstract

With the advent of internet of things (IoT) technology, millions of smart devices will be connected to each other as well as to the internet. These connected devices will generate a huge amount of data and collaboratively dispense meaningful insights about the surrounding physical phenomenon by sensing, processing and actuating other devices to cater to various applications. In IoT networks, the sensor nodes need to continuously measure and transmit sensor data to the sink node or to the cloud directly. The communication cost in this transaction is usually high as compared to other operations of a sensor node. Keeping in mind that these battery-operated devices have a limited lifetime, efforts have been made in the wireless sensor networks (WSN) and IoT literature to improve the lifetime of the nodes. Of the many possible ways to increase the lifetime, reducing the sensor data communication to the sink node or other nodes is one of the most efficient ways. These data-transmission reduction techniques are often referred to as data reduction schemes. This thesis focuses on the popular data reduction schemes based on the classical least mean square (LMS) algorithm and the Shewhart test-based scheme. The algorithms have been implemented on a custom-designed low-cost sensor node using easily accessible hardware. The results of the proposed implementations have been evaluated upon real-time deployment of the nodes in our lab. Various results and measurements such as percentage reduction in data transmission, the current consumption of the node and estimation of the battery lifetime have been quantified based on the observations. Comparison of the LMS and Shewhart test-based data reduction schemes on different datasets encourages us to improve the accuracy of the estimated time series data at the sink node. To achieve this, a piggybacking and interpolation based scheme has been proposed along with the Shewhart test-based scheme to improve the mean square error (MSE) of the estimated data, without increasing the number of data transmissions to the sink. The proposed algorithm has also been implemented on the custom-designed sensor node. The current consumption and battery lifetime have been analyzed. It is seen that there is an inherent trade-off between the decrease in the MSE and the reduction in the battery-lifetime for the number of piggybacks chosen. However, the decrease in the battery lifetime is negligible as compared to the improvement in the accuracy. Later, the Shewhart test-based data reduction scheme is implemented on NodeMCU and integrated with oneM2M standard to show the proof-of-concept for using these algorithms for industrial and real-world IoT applications.

Abstract

Community-driven Question Answering (CQA) platforms such as Stack Overflow, Quora, Yammer, AnswerHub, etc., have gained immense popularity in recent years among the Internet users to seek, learn and share information. These platforms not only provide users with the means to discuss their queries with the community and fulfill their information needs, but also serve the purpose of knowledge-base creation. Each CQA platform encourages rich content by recognizing contributions of its users in terms of different metrics of reputation. Thus, peers on these platforms dissuade other users to post low- quality content by closing, disliking or not answering their questions. This is expected to keep the platform focused; however, the aggressive attitude of the community towards marking the posts as low- quality leads to negative impact on users’ experience of the platform (especially for new users) leading to reduced interest in the platform. Thus, to improve the user experience and maintain the content quality on the CQA platforms includes issues such as the increasing count of low-quality questions, aggressive communities and the editing process for the low-quality questions on the platforms. In this thesis, we address two issues related to the content quality management at the CQA platforms: (1) We note that editing closed questions can be time consuming, especially for new users: Often, the edited versions of the closed questions may still be found to be unsuitable by moderators and remain closed for a long time. This motivates the need of automated mechanisms that can potentially help users with appropriate feedback on their closed, disliked or unanswered questions, so that they can obtain rel-evant answers in a timely fashion without getting unnecessarily stuck waiting for moderators’ feedback.(2) We also note that another reason behind questions getting closed is duplicity. The community marks them as duplicate considering their semantic similarity with the related questions already available on the platform. Although users can search for related questions using the search & retrieval systems of these platforms before posting their questions. But most of these systems capture and return questions primarily based on syntactic similarity which may not meet the exact need of the user. To address the above-mentioned issues: First, we present a framework to assist the users in the re-opening process of closed questions on the platform. We build a predictive modeling framework that suggests to a user as to whether the edited version of his/her closed question will get successfully re-opened or not. This can assist users at large by retracting them from entering the review process with improper edits. To learn these models effectively, we analyze the closed questions of established & non-established users (determined by their reputation score). We note that the established users have higher odds of getting their closed questions reopened than non-established users. Thus, we present an approach to leverage the better editing skills of established users to train a binary classifier that determines whether the edited closed questions will get reopened or not. In order to train such a classi- fier, we identify a rich set of relevant and effective features based on exploratory analysis of a popular CQA platform, namely Stack Overflow. Through empirical evaluation of the proposed approach using publicly available data from Stack Overflow CQA platform, (i) we demonstrate that the proposed pre- dictive modeling framework can be used to estimate whether an edited version of a closed question will get reopened or not, (ii) we show that by leveraging such a framework, the users can potentially save substantial amount of time that currently goes waste while waiting for a response from the moderators. Second, to reduce the number of duplicate questions we present a mechanism to generate a set of factual questions from the accepted answers of the top-rated questions on the Stack Overflow platform. The mechanism is based on the SRL (Semantic Role Labels) parse of the sentences in the accepted answers of these questions. We demonstrate that the generated set of factual questions can potentially increase the syntactic coverage of the questions on the platform, thus yielding better results while using the search & retrieval system of the platform. Eventually, this can lead to lesser number of duplicate closed questions and improved content quality on the Stack Overflow platform.

Abstract

In this thesis we mainly focus on three major aspects of computational humour recognition. We start with categorizing humour based on the classical theories of humour along with features like theme, emotions and topics. We then look at the problem of recognizing humour in conversations and broadcasted speeches which are more complex and large than short jokes. Finally, we try to differentiate between different types of off-color humour and try to detect insulting remarks from off-colour humour in which dark humour is often misclassified as insulting humour. Most scholarly works in the field of computational detection of humour derive their inspiration from the incongruity theory. Incongruity is an indispensable facet in drawing a line between humorous and non-humorous occurrences but is immensely inadequate in shedding light on what actually made the particular occurrence a funny one. Classical theories like Script based Semantic Theory of Humour(SSTH) and General Verbal Theory of Humour(GVTH) try and achieve this feat to an adequate extent. We adhere to a more holistic approach towards classification of humour based on these classical theories with a few improvements and revisions. Through experiments based on our linear approach and performed on large data-sets of short jokes, we are able to demonstrate the adaptability and show componentizability of our model, and that a host of classification techniques can be used to overcome the challenging problem of distinguishing between various categories and sub-categories of jokes. Almost all the studies done in the field ofcomputational humour recognition has been done on datasets consisting of short jokes, tweets and puns. We try to detect humour in conversations and broadcasted speeches as they are complex and contains more contextual information when compared to short jokes. For the purpose of automatic humour detection in monologues we built a corpus containing humorous utterances of TED talks and for dialogues we analysed data from a popular TV-sitcom Friends whose canned laughter gives an indication of when the audience would react. We classified dialogues/monologues into humorous and non-humorous by using multiple deep learning methods. Our experiments on the data show that such deep learning methods outperform the baseline by 21 accuracy points respectively on the TED Talk dataset. Off colour humour is a category of humour which is considered by many to be in poor taste or overly vulgar. Most commonly, off-colour humour contains remarks on particular ethnic group or gender, violence, domestic abuse, acts concerned with sex, excessive swearing or profanity. Blue humour, black humour and insult humour are types of off-colour humour. Blue and black humour unlike insult humour are not outrightly insulting in nature but are often misclassified because of the presence of insults and harmful speech. We then provide an original data-set consisting of nearly 15,000 instances and a novel approach towards resolving the problem of separating black and blue humour from offensive humour which is essential so that free speech on the internet is not curtailed. Our experiments show that deep learning methods outperforms other n-grams based approaches like SVMs, Naive Bayes and Logistic Regression by a large margin.

Abstract

Autonomously navigating to a goal pose is central to several robotic applications such as inspection, surgery, object manipulation, transportation and more. Often navigation frameworks require the goal pose to be specified in Cartesian space and a localization module that estimates the current pose of the robot. However, in the absence of a metric-scale map and an accurate position estimation module, it is difficult to attain the correct goal pose. Visual servoing approaches, therefore represent the goal pose in the form of an image, which is more intuitive for humans and easier to obtain as compared to the 6-DoF coordinates. The image feedback from a robot's camera is further employed by visual servoing approaches for localizing the robot in image space and navigating it to the goal pose. Existing visual servoing approaches control the robot such that the overlap between the current image and the goal image is iteratively maximized. Therefore, existing approaches do not generalize while servoing to different objects or even instances from the same category. We introduce a novel problem of instance invariant visual servoing, where the robot is required to servo a given instance in the desired pose represented by a goal image of a different instance from the same category. This problem is more relevant in practical scenarios since the same robot can handle multiple object instances without changing the goal image. We initially tackle the problem of instance invariant visual servoing using a geometric approach. We propose a novel visual servoing control law based on a linear combination of 3D models representing the given object category. Our approach is capable of accommodating the deformation in shapes among object instances and the resulting controller is able to smoothly navigate the robot to the desired pose even though the instances are different. To tackle the large variation in appearance across object instances, we further propose part-based visual features that uniquely capture the locations of an object's parts in images. Advantages of using such part-aware semantics are two-fold. (i) It conceals the illumination and textural variations from the visual servoing algorithm. (ii) Semantic keypoints enable us to accurately match descriptors across instances. To circumvent the requirement of 3D models by the above-mentioned geometric solution, we propose a discriminative learning based framework for visual servoing across object instances. This approach is able to learn the desired pose from previously seen examples of goal images and is able to navigate the robot despite appearance and shape variations. Specifically, we learn a binary classifier that discriminates the goal image from images captured from other viewpoints for that object category. The classification error is then used to navigate the robot towards the desired pose. Furthermore, we design controllers for linear, kernel and exemplar Support Vector Machine (SVM) and empirically discuss their performance in the visual servoing context. To address large intra-category variation in appearance, we introduce Principal Oriented Glyph (POG) features that are easier to obtain as compared to part-based features. We next consider the problem of navigating a robot in unstructured environments using deep learning techniques. Motivated by recent breakthroughs in performance of data driven methods on recognition and detection tasks, we aim to learn visual feature representations suitable for servoing tasks in unstructured and unknown environments. In contrast to existing visual servoing approaches that require the knowledge of a scene geometry especially depth and camera parameters, we present an end-to-end learning based approach for visual servoing in diverse scenes that is agnostic to the knowledge of a scene's geometry and camera parameters. This is achieved by training a Convolutional Neural Network (CNN) over color images with synchronized camera poses. After assessing the capability of CNNs in classical visual servoing tasks. We next apply them to our instance invariant visual servoing task. However, catering to shape, appearance and viewpoints variations all together is a difficult task for a CNN. Thus, instead of directly regressing to the desired pose, we use CNNs to predict locations of part based keypoints in images so that appearance variations are decoupled. To tackle geometrical and viewpoint variations, we further present a pose induction strategy based on the part-based keypoints predicted by our network. Specifically, we reconstruct the locations of parts of the given instance in 3D, from a sequence of 2D keypoints estimated by the network and align this reconstruction with a template instance. This aligned reconstruction is then used to localize the robot and navigate it the desired pose. Finally, we employ our framework for the autonomous inspection of vehicles using Micro Aerial Vehicles (MAVs), which is vital