Abstract

An autonomous personal assistive robot should be able to interact with humans and objects at high level, by understanding the semantic constructs in the environment. Robot should be able to fetch, place, move objects in the environment. This involves detection, recognition and manipulation of the object across the semantic construct. These semantic constructs can be defined on wide range of categories, from building, room, furniture to object level. This semantic understanding can be acheived by perceiving the environment with lot of details and then represent it in the form of a map. Obstacle location details should be available to the metric level for robust navigation of the robot. There are many mapping paradigms in literature and semantic mapping is recently focused for this purpose. Assuming that robot is able to perceive the environment in terms of semantic constructs like lab, room, table and chairs, the performance of recognition of any object is entailed to the position or the view of the robot’s camera which is under motion. Object search also depends on the efficiency and robustness of the recognition algorithms for a moving camera sensor. This thesis explores two problems: 1) How to place/move the robot in a indoor environment in order to efficiently search for an object? 2) How to use visual saliency models for practical indoor settings? Assuming that the map of the environment with semantic constructs is available to the robot, an exploration strategy has been formulated to search for an object in optimal way. Having done this for an ideal setting, visual attention models have been analysed for the practical indoor settings. In the first part, manual way of extraction of the semantic constructs using Kinect sensor in the form of 3D map is explained. Since these semantic constructs reduce the search space for the robot from the entire map to certain locations, they are defined as ”potential locations” for the search. This information is used to explore for the object in the environment. Viewpoint of the Kinect sensor is used as the primary ingredient in this formulation, so as to place the robot to view potential locations as highly as possible. At every instant a next best viewpoint is calculated based on the explored and un-explored semantic contructs in the map. Exploration is continued till the object is found or till all semantic constructs are explored. Comparisons of this exploration is shown with two other methods and their timings have been captured to substantiate our claim. Secondly, visual attention models have been explored to provide cognitive advantage to solve the above problem for a practical setting. Existing visual saliency models that mimic the attention regions, have been used to localize the single/less clutter regions in the scene and cannot be directly used to indoor settings. Influence of depth in the salient region detections have been experiments and cases have been derived. A new saliency model called 3D saliency has been proposed in this work and fused with the existing saliency models to derive RGBD-saliency. This improves the performance for indoor settings with an average improvement of 9% across all the methods. Fusion process uses SVM learning. It is explored on how this saliency can help in object search formulated earlier and pave ways of improvements in the future work.

Abstract

No Results Found

Abstract

The interplay of three biomolecules, DNA, RNA and proteins provide complexity and richness to the biological world. Proteins have long been the primary research focus among various groups across the world. The nature of their internal dynamics, apart from the biochemical and functional aspects, has attracted enormous attention. Investigation of the factors governing the dynamics can give valuable insights into protein activity or functions. Advanced theoretical and experimental techniques contribute to the better understanding of protein structures, stabilities and internal dynamics. Computer simulations provide with an excellent alternative to interpret and complement the experimental investigations and provide a molecular level insight into protein structure, dynamics and functions. This atomistic-level information obtained from biomolecular simulations can be used to derive various experimentally measurable macroscopic observables, including heat capacity, free energy, by using statistical mechanics. Like harmonic solids, the protein atoms execute harmonic vibrations around their equilibrium positions, at low temperatures. Upon heating, the amplitude of these harmonic motions increases linearly with the increase in temperature up to a critical temperature (Tg) beyond which a transition from the glass-like phase to liquid-like phase is observed. Atomistic MD simulations calculate dynamical properties relevant to harmonic-to-anharmonic transition which are attributed to flexibility in proteins and onset of biological activity. The glass transitions have been reported in proteins, DNA and RNA by various experimental and simulation studies. The glass-transition temperature (Tg) is observed at about 220 K. Recently, a low-temperature anharmonic transition has been observed in many proteins at around 150 K, that is much below the Tg. Reasons for the same have not clearly been stated. This work is an attempt to understand the underlying factors which contribute significantly to this behavior. The present work emphasizes on the role of MD simulations to understand the factors that contribute to the anharmonic transitions of proteins. Four homomeric polypeptides, polyglycine, polyalanine, polyisoleucine and polyphenylalanine have been analyzed for the dynamical behavior of each kind of side-chains to the anharmonic transitions over a range of temperatures (between 10K and 300K in steps of 10K). The dynamical properties of proteins are largely attributed to the amino acid side-chain groups as is exemplified by the system under consideration. Polyglycine, representing backbone-only polypeptide shows only solvent-dependent dynamical transition, while as polyalanine, polyisoleucine and polyphenylalanine shows a low-temperature onset of anharmonicity along with the dynamical transition at the critical temperature (Tg). The results obtained from our simulation studies correspond exactly with the experimental data from elastic incoherent neutron scattering (EINS) studies.

Abstract

Identifying subjective text and extracting opinion information about different entities from News, Blogs and other forms of user generated content like social media has a lot of applications. We wish to address this challenging task of mining opinions about different entities like organizations, people and places from different languages. But, it is known that resources and tools for mining opinions are scarce in many languages. In our research, we try to address resource scarcity problem by designing approaches which leverages resources from rich resource language like English for opinion mining. Named Entities, adjectives and verbs treated as possible opinion words are extracted from English in a comparable corpora. Possible opinions on targets (mainly people) represented by adjectives, verbs are translated and transliterated to resource-scarce languages. Initially, these transformed words are used to identify subjective sentences in resource-scarce languages. Though this approach is unsupervised, subjective and objective text classification at sentence-level can be improved using a supervised method. Accuracy of sentence-level subjectivity detection is further improved using language independent feature weighting and selection methods which are consistent across different languages. Once the subjective sentences are identified, they are used to create a subjective language model (SLM). While transliterated and translated words are used to form structured opinion queries (OQs) using inference network (IN) for retrieval to confirm the opinions about opinion targets in the documents. Experiments are performed at sentence-level and fine-grain level using Hindi comparable corpora to evaluate the effectiveness of approaches in finding opinions. Even though experiments are performed on Hindi, approach can be easily extended to other resource-scarce languages. Results have shown that for fine-grained opinion mining, OQs and SLM with IN framework are effective for opinion mining tasks in minimal resource languages when compared to other retrieval approaches. While Entropy based category coverage difference criterion (ECCD) feature selection method with our proposed feature weighting methods along with Naive Bayes Multinomial classifer is consistent for sentence-level subjective classification in Hindi, Telugu and other languages including English and European languages.

Abstract

Understanding the process of protein folding/unfolding has been a challenge for more than 50 years. Atomic level understanding requires huge computational power to gain insights in the process of unfolding. The emergence of single-molecule force measurement experiments has facilitated a better understanding of protein folding pathways and the thermodynamics involved. Experimentally, techniques like atomic force microscopy and optical tweezers are used to study the mechanical properties of single molecule. A computational method that closely mimics these experiments is steered molecular dynamics (SMD) simulations, which can provide atomistic level insight into the unfolding pathways. Recent experimental studies have shown that combinations of single-molecule experiments with traditional methods such as chemical and/or thermal denaturation yield additional insights into the folding phenomenon. In this study, results from extensive computations (a total of about 60 SMD simulations with a total length of about 0.4 μs) that address the effect of thermal perturbation on the mechanical stability of the I27 domain of the protein titin are reported. A wide range of temperatures (280 – 340 K) were considered for the pulling, which was done at both constant velocity and constant force using SMD simulations. The distance between the N- and C- terminal Cα atom was considered as the reaction coordinate for the present study. For I27 domain, maximum force of unfolding decreases with respect to increase in temperature, but the extension corresponding to this force remains similar, indicating the validity of end-to-end distance as reaction coordinate. Similar trends were observed for the most probable unfolding force and are in good agreement with the available experimental data. It is observed that the rate of unfolding increases with an increase in temperature after the initial barrier is crossed in the constant force simulations. This study identifies two competing pathways for the mechanical unfolding of I27 both at constant velocity and constant force SMD simulations, and illustrates the significance of combining various techniques to examine protein folding.

Abstract

Due to the digital and information revolution we are witnessing presently, there are a huge and continously increasing number of images present on the Internet. For example, a query for ``Eiffel Tower" on Google Images returns more than two million images. The easy accessiblity of this data provides us with unique opportunities to mine the contents of these images not only to do automatic organization, but also for providing interactive interfaces to browse, explore and query. This task is challenging given the massive size and the continuous growth of the collection. To add to this, these collections are taken in varying imaging conditions, with different cameras, at different resolutions, from different perspectives and have different degrees of occlusions present in them. Hence, for image collections even the simplest of tasks such as finding matching images turn out to be hard. The Computer Vision community has been actively designing and redesigning algorithms to overcome these challenges. One of the most widespread and noticable idea employed is that of extracting robust, invariant and repeatable local features in the images, followed by the subsequent quantization of the feature space as visual words. The similarity of images is gauged by the correspondence and similarity of thier local features. Verifications of the matchings is done to eliminate spurious matches. Building a data structure such as an inverted index over these visual words can catalyse the process of discovery of matching features. This mining of similar images by matching features, forms the basis of all high level algorithms such as clustering, skeletonization, summarization etc. which help in the organization, exploration and querying of these image collections. This thesis presents two novel algorithms which help in achieving this goal. First, we introduce a novel indexing scheme that makes it possible to do exhaustive pairwise matching in large image collections. The quantization of image features and thier indexing provide on a limited amount of leverage for speeding up the image matching process which depends upon the sparsity the posting lists. This sparsity is controlled by the number of visual words used which after a point cannot be increased arbitrarily without affecting recall. Our scheme, generates higher order features by pairing up nearby features and encoding their affine geometry. This provides a much larger feature space to index which can be subsequently reprojected to any desired size by defining appropriate hash functions. We implement our indexing scheme by providing an analogy with Bloom filters. The higher order features extracted in the images are inserted into their respective equally sized Bloom filters using a single hash function. This unformity in Bloom filters allows for only a single inverted index to be able to index the hash buckets of all the Bloom filters, and thus providing a simplified interface to implicity query all the Bloom filters. We choose the size of these Bloom filters to be in proportion to the size of the database. This enables us to do querying in constant time, since the average size of the posting lists becomes constant. Also, the use of such large implicit Bloom filters is able to sufficiently mitigate the negative effects of using a single hash functions. As a result, we are able to do exhaustive pairwise matching over large databases of upto 100K images in linear time complexity. Second, we present a fast and easy to implement framework for browsing large image collections of landmarks and monumental sites. The existing framework ``Phototourism" would require doing a reconstruction of the whole scene by employing Structure from Motion package called Bundler. This requires pairwise matching required to generate tracks of matching features across images. Next an incremental approach is applied, starting with a seed reconstruction and adding more matching images into the reconstruction. This, however, requires continous refinement of the whole reconstruction using a computationally expensive procedure called bundle adjustment. The pairwise matching and bundle adjustment become the limiting factors in scaling this technique to large image collections. To overcome the issues faced with ``Phototourism", our framework employs independent partial reconstructions of the scene. We use standard Bag of words model and indexing techniques to determine closest neighbours of each image in the collection, and do a local reconstruction corresponding to each image using only the neighbouring images. This requires us to only solve multiple simple reconstructions problems instead of one large reconstruction problem, making it computationally more tractable. Our browsing interface hops from one reconstruction to another to give the user an illusion of browsing a global reconstruction. Our approach also makes it easy to adapt to growing image collections, as adding an image only incurs a

Abstract

Mobile robots have the potential to be widely used. They can be used in complex indoor and outdoor environments accomplishing jobs which typically are earmarked for humans. A basic requirement for these kind of robots is ability to navigate autonomously. For autonomous navigation to be possible the robot needs to be able to perceive the environment around; specifically localize itself in it, and create a map which it can use to plan and navigate. Localization and mapping are inter-dependent on each other, that is to say localization quality determines mapping quality and mapping quality determines further localization quality. Furthermore, it will determine how reliably we avoid obstacles, search for objects, determine the best path for a goal etc. Quality of the map which is created is of enormous importance, as it inherently dictates merit of the task robot is expected to execute. Accurate mapping of environment becomes more challenging when it is attempted using a monocular camera. Monocular camera posses it’s host of difficulties, but is very cost effective compared to expensive laser scanners and stereo cameras. As a sensor, monocular camera provides plethora of information like depth, color, texture which can be used object identification, face recognition, lip reading, traffic signs, and and 3D structure, most of which is still not being harnessed, and hence presents itself as a sensor for the future and worth pursuing. Therefore, we have used a Monocular camera as the quintessential sensor for mapping the surroundings. We are a long way from realizing monocular camera’s potential. As of now it is incapable of handling violent motions. A fast moving robot or a sharp turn leads to breaking of the mapping process. Sideways motion for the camera is an ideal movement for VSLAM systems, but this luxury is not afforded to mobile robots. Looking straight ahead and moving in front (along the principle axis) is a feasible motion for them. For such motion, moving straight in a line at slow speed(not too slow) produces better result, and when robot needs to turn it is done so moderately. Hence, we have proposed a planning framework which devices a path ensuring motion with aforementioned constraints, that is we have optimized heading change, and velocity with respect to distance from the goal. We model our robot as a non-holonomic vehicle which allows us to define feasible motion states for the robot. We than parametrize it’s states, namely it’s position and orientation into polynomial functions, which serve as easy inputs for optimization methods. We use optimization at two stages, first a non-linear optimization to generate path for a given start and goal position, and second a relatively naive quadratic optimization method to estimate initial guess for the non-linear optimizer. A good initial guess from the quadratic optimization method will be reflected in quality of final output. We show that by following the path generated by aforementioned framework VSLAM is able to localize camera successfully, which is to say estimated camera trajectory has similar curves to the generated trajectory. We also show following the generated constrained trajectory tracking performance improves in comparison to when robot follows unconstrained trajectory generated by RRT, one of the most popular algorithms for path planning. We add dense reconstruction of planar objects after the path planning framework in our attempts towards automation. Algorithms used to create 3D maps from stream of images have primarily focused on sparse point maps. Sparse maps are incomplete representation of the environment, making comprehension of available data very hard. They are also not able to cater complete information to modules down the pipeline of a full fledged robotic system. We have proposed a framework for dense reconstruction of 3D planar objects. Most objects in an indoor environment are planar, hence we have focused our attention towards reconstructing such surfaces. Targeting planar objects allows us to perform dense tracking in real time on commodity laptop. We have used Homography and segmentation techniques to identify and track planar objects. Post-reconstruction we present a modified Bundle Adjustment, an optimization technique used to correct position of 3D points. Traditionally VSLAM based systems have required human intervention, dense model of an environment will provide us with information such as span of obstacles and ground required for autonomous navigation. In results section we qualitatively demonstrate superiority of dense models over sparse point clouds. We also, both quantitatively and qualitatively show performance enhancement achieved by proposed bundle adjustment over original bundle adjustment. In conclusion, we present a path planning framework which allows robot to move by on it’s own accord if answer to “Where to go?” is known. We also present a reconstruction framework which improves

Abstract

Several natural phenomena manifest themselves as spatio-temporally evolving processes, for example climate patterns, disease spread and urban sprawl. The study of these processes, which aims to increase our understanding of the spatio-temporal phenomena for their prediction and control, requires analysis tools to infer models and their parameters from collected data. Whilst several studies exist on how to model the characteristics of spatio-temporal processes from highly complex patterns, a generic, statistically valid approach which may be readily employed in a wide range of scenarios, such as with systems with different forms of observation processes or time-varying systems, is lacking. This work attempts to fill this void by providing a systems approach to spatiotemporal modeling which can be used with continuous observations, point process observations, systems exhibiting spatially varying dynamics and time-varying systems.
Even though many works have been successful in mining spatio-temporal patterns from a Data Mining (DM) and Knowledge Discovery (KD) approach, they suffer from numerous limitations - they are tightly dependent on the ability to hypothesize, to determine the relevant factors and to deduce a conclusion, which require large thematic data and understanding of the subject to even define the primary hypothesis of the analysis. Hence, are heavily coupled to the domain. Moreover, they tend to neglect the continuous nature of the phenomenon they are trying to observe by discretizing the space or the time in which the co-location and co-occurrence pattern mining is attempted. Lastly, scale factors bring in more challenges which limit the model design and applicability, for instance, the occurrence of natural break lines can sub-divide or alter the choice of the appropriate scale of analysis. Therefore, there is an increasing need of developing an approach that can produce valid analyses of variations in dynamic phenomena, with particular focus on utilization of the inherent spatio-temporal characteristics of the data to address the issues mentioned above.
The developed set of methods act as a framework that leverages the observed event data - both spatial and temporal characterizations of observed event occurrences, making use of inherent spatio-temporal dependency in such data. A risk map is generated that provides valuable insights into its geographical spread and to help quantify the spatial risk factor associated with it, acting as a basis for dynamic spatiotemporal modeling which can easily cater for spatially varying dynamics. The work finally leads to the development of a novel method to model meso-scale processes by capturing spatio-temporal correlation leveraging both the potential dependencies among the local coefficients, and the potential incongruities within global models for prediction of trends. In the context of disease management, the proposed method is able to infer causality, and simulate the spread given the initial sources derived primarily from the spatio-temporal history of the disease.
The methods are seen to compare favorably with other conventional approaches and to work well on real temporal data sets. The potential of adopting the proposed approach to spatio-temporal modeling is seen in the illustrative case studies of Salmonellosis disease in three states of USA, which have been presented to demonstrate the utility of the method, where it is found that reliable predictions are possible even in complex scenarios. The encouraging results mimicing the observed trends are a strong indication that the adopted approach, per se may be used for large-scale spatio-temporal systems across several disciplines that exhibit similar spatio-temporal dynamic behavior and thus, provide a mechanism by which stochastic models are made available for spatio-temporal decision-making processes. It is evident that such a framework will help prioritize decision-making process for better risk assessment and management including disease outbreak, as shown in the illustrative case studies.

Abstract

Sentiment Analysis is an area of focus over the last decade. Increase in user-generated content provide an important aspect for the researchers, industries and government(s) to mine this information. The user-generated content is one important source for various organizations to know/learn/identify the general expression/sentiment of different users on the product. In this work, we focus on mining sentiments and analyzing them for Hindi language. Hindi is the 4th commonly spoken language1 in the world. With the increase in the amount of information being communicated via regional languages like Hindi, comes a promising opportunity of mining this information. Mining sentiments in Hindi comes with their share of issues and challenges. Hindi is morphologically rich and is a free order language as compared to English, which adds complexity while handling the user-generated content. The scarcity of resources for the Hindi language brings challenges ranging from collection and generation of datasets. We take up this challenge and work towards building resources- reviews, blogs annotated corpora and subjective lexicon for Hindi language. We propose a technique to build a subjective lexicon given a pre-annotated seed list for a language and its wordnet representing the network/connectivity of words using synonyms and antonyms relations. One of the salient features of this technique is that the method can be applied for any language which has the wordnet available. To show the applicability of the technique on other languages, we experiment our technique on English language in addition to the Hindi language. The lexicon generated by our algorithm is evaluated using the following different metrics. 1. Comparing against Manual Annotation 2. Comparing against similar Existing Resources 3. Classification Performance In addition to resource creation, we take up the task of sentiment classification in Hindi Language. We work on two different genres of Hindi User- Generated Web Content- 1. Reviews 2. Blogs For both of these genres we present three different approaches for performing sentiment classification such as 1. Using Subjective Lexicon 2. N-Gram Method 3. Weighed N-Gram We aim at analysing the merits and demerits of each of the above approaches across the different genres for the sentiment classification task. We discuss in detail the problems and the issues while working with the user-generated content (reviews and blogs) in Hindi language. This research work, throws some light on the main differences between the User-Generated Content in English and Hindi language at linguistic and its representation level and the approaches followed to address the same. English language provides the option of leveraging the abundant resources and tools that have been developed in the past, the work for Indian Languages has just began since last decade and it is in early stage of research and development, so our focus has been on- “To effectively mine the subjective information from the user-generated content in Indian languages, overcoming the data scarcity challenges associated with such problems.”.

Abstract

Stroke is a chronic disease which often leads to death. Different medical imaging modalities enable diagnosis for stroke after the onset of symptoms. Time is of the essence during stroke analysis since the window of therapy is very small (< 3 hrs after the onset of symptoms). Recent clinical studies have shown the usefulness and significance of diagnosing stroke on the Diffusion Weighted Magnetic Resonance Imaging (DWI) scans of the brain in the early stages. Visual inspection of the DWI scans is difficult since multiple scans are acquired for a patient with varied contrast and the scans depict complementary information about the diffusion process in the brain. To make matters worse, the DWI scans are acquired at a very low resolution with poor signal to noise ratio (SNR) since the time of acquisition is significantly less (< 1 min) and are confounded by artifacts that mimic stroke lesions. Thus, an automated framework which can accurately capture the stroke lesions in the DWI data would assist the clinicians in a better diagnosis. This is focus of the thesis. Varying the acquisition parameter (b-value) generates different DWI scans with varied contrast. DWI with higher b-values provide improved sensitivity, conspicuity of stroke lesions and reduced artifacts at the cost of lower SNR. Along with the DWI scans, the Apparent Diffusion Coefficients (ADC) maps are also derived which give a measure of the true diffusion process in the brain irrespective of the acquisition artifacts that resemble stroke. In this thesis, we argue that integrating information from multiple sources, namely, low and high b-value data along with the ADC maps, can aid better characterization of stroke lesions in the data. Accordingly, we propose a novel approach for detecting and segmenting stroke regions from DWI data.