SAFE: Smart, Authenticated, Fast Exams
SAFE (Smart Authenticated Fast Exams) is a smart-phone app based system to conduct easy online-exams in proctored venues. Today, smart-phones or other electronic devices are explicitly disallowed in exam venues, due to the obvious possibilities of misuse for cheating. SAFE brings the smart-phone revolution into the exam hall.
Weakly supervised 3D shape analysis
Online repositories contain millions of 3D shapes, providing data for a wide range of data-driven 3D modeling interfaces. Such interfaces facilitate, accelerate and democratise computer-aided design and 3D content creation. By automatically learning design rules and structural principles from training data, these interfaces allow even novice and casual users to design complex and functional objects.
Physically based animation and rendering
Many real world phenomena, like fluid flows, combustion, and garment drapes demonstrate astonishing complexity of structure, movement and appearance. Efficient and accurate simulation and animation of these play a crucial part in many applications ranging from scientific visualisation to advertising, from computer aided design to retail and from entertainment to medicine.
Reinforcement learning
Our group’s research is motivated by the goal of creating intelligent agents, especially ones that can learn. In pursuit of this goal, we consider questions from a wide variety of topics. Central to our investigation is reinforcement learning (RL), which is a general paradigm for an agent, through trial and error, to discover actions that maximise its long-term gain. RL finds application in a variety of domains, including game-playing, stock-trading, medical decision-making, and environmental preservation.
Issues in analysis of images from space borne platforms
Earth observation image data are mainly characterised by (i) Spatial resolution – the ability of the imaging sensor to focus on very small areas and thereby distinguish between closely spaced features, and (ii) Spectral resolution – ability to observe the targets in a number of narrow wavelength bands of the electromagnetic spectrum.
High spatial resolution images allow image analysis based on objects (regions) in terms of their shape, size, and spectral homogeneity, examples which can be seen below:
Hindi Shabdamitra: A digital aid for language teaching and learning
िहंदीशब्दमित्र(Hindi Shabdamitra) is an e-learning product meant for Hindi language teaching and learning.It uses Hindi Wordnet which is further augmented with audio-visual features, grammatical properties and is presented in a learner-friendly layered format.
Integrated computational materials engineering
Integrated computational materials engineering (ICME) has two key components namely multi-scale modeling beginning with first principles and data informatics for better design of products. Both components are focused on shortercycle of product development and efficient use of materials and resources leading to lower cost of manufacturing.‘Integrated’ means integration of interdisciplinary technology to desired product via multi-scale qmodeling. In the ‘bottom-up’ approach of ICME one starts from first principles, designs a material for a given application.
Multiscale modeling of metal deformation under extreme conditions
Reactivity initiated accident (RIA) or loss of coolant accident (LOCA) in a nuclear reactor may lead to sudden temperature rise. Accidents caused by RIA or LOCA condition may lead to a dynamic expansion of fuel pallets. This results into a multi-axial state of deformation caused by high thermal loading (1000 o Cs -1 ) in presence of extreme conditions of irradiation. Low temperature in early stage transient, metal-water reaction and accumulated irradiation in a high burn-up clad may also lead to a brittle failure.
Three dimensional study of mechanisms of compressive deformation and failure in porous bulk metallic glasses
Bulk metallic glasses (BMGs) are amorphous alloys with enhanced properties such as high strength, large elastic strain, corrosion resistance, high fatigue and fracture toughness in comparison with their crystalline counterparts. This makes BMGs suitable for deployment in many applications. However, upon loading uniaxially beyond their yield point, most BMGs fail catastrophically, displaying very little plastic strain. This limits the use of BMGs in load bearing applications.
Phase field modeling of microstructural evolution
Patterns are ubiquitous in nature – such as the stripes on a cat or the dunes of sand on a beach. In many engineering materials, at length scales which can only be seen under a microscope, we often see patterns that form liked stripes or lamellar structures. In addition, many interesting shapes – such as, pyramids, cuboids, plates, and prisms – are often seen under the microscope.
Detecting and curating errors in Sanskrit OCR documents
Background
One of the major objectives of Science and Heritage Initiative (SandhI) at IIT Bombay is to create a large data-base of Sanskrit texts (to begin with that dealing with Science and Technology) in searchable format. To meet this end, we decided to resort to OCR techniques. Given the limitations of different tools available, some efforts were launched to synthesise the available tools and thereby increase efficiency of the process.
The problem
Automating reading comprehension by generating question and answer pair
Asking relevant and intelligent questions has always been an integral part of human learning, as it can help assess user understanding of a piece of text (a comprehension, an article, etc.). However, forming questions manually has always been an arduous task. Automated question generation (QG) systems can help alleviate this problem by learning to generate questions and answers on a large scale and in lesser time. Such a system has many applications in a myriad of other areas such as FAQ generation, intelligent tutoring systems, and virtual assistants.
An adaptive framework for end-to-end corrections in Indic-OCR
Optical character recognition (OCR) is the process of converting the document images into an editable electronic format. This has many advantages like data compression, enabling search or edit options in the images/text, and creating the database for other applications like machine translation, speech recognition, and enhancing dictionaries and language models. OCR in Indian Languages is quite challenging due to richness in inflections.
Video analytics for security
Surveillance cameras have emerged as very effective and important aspect of security and monitoring. Unfortunately however, they suffer from two serious challenges. First, their effectiveness in preventing a mishap is limited by the alertness levels of humans who are expected to monitor a grid of live feeds from many cameras 24X7. Since humans are not known for large attention spans, more often than not the mishap misses the eyes of the on-duty guards and the purpose is defeated. The recorded CCTV footages then at best serve to understand what happened, as a post-mortem analysis.
Video analytics for compliance and quality monitoring in MoRD skill development centres
Deen Dayal Upadhyaya Grameen Kaushalya Yojana (DDU-GKY) is a placement linked skill development scheme for rural poor youth by the Ministry of Rural Development (MoRD). It is an important component of the National Skill Development Policy. The DDU-GKY skilling ecosystem consists of MoRD; State missions; project implementing agencies or training partners; and technical support agencies. Both generic (soft skills, English and Information Technology) and trade-specific training is offered to youth at various training centers in partnership with the third-party training agencies.
Assessment of farmer producer organisations using Rubric methodology
Farmer producer organisation (FPO) is a type of producer organisation (PO) which is a legal entity formed by primary producers, viz. farmers, milk producers, fishermen, weavers, rural artisans, craftsmen and the alike. The primary objective of a FPO is to mobilise small and marginal farmers (~80% of total farmers in India) into member-owned producer organizations. It helps to foster technology penetration, improve productivity and access to the inputs, investments, markets and services, and increases farmer incomes, thereby strengthening their sustainable agriculture based livelihoods.
Search across knowledge sources: Components of building semantic search systems
Semantic search can be described as the effort to improve the accuracy of the search process by understanding the context and limiting the ambiguity. Semantic search engines are more likely to try to understand the meanings that are hidden in retrieved documents and users’ queries, by means of adding semantic tags into texts, in order to bring structure into and conceptualise the objects within documents. The primary components of the semantic web, ontologies and knowledge graphs (populated ontologies), are rich sources of domain knowledge.
IndoWordNet
WordNets are lexical structures composed of sets of synonyms called synsets, and semantic relations between these synsets. Wordnets help in various natural language processing (NLP) tasks such as word sense disambiguation, machine translation, etc. Unavailability of a crucial lexical resource like Wordnet has impeded the development of NLP technologies for Indian languages.
Machine translation
Machine translation (MT) deals with automatic translation of text from one natural language to another. It is one of the most challenging problems in natural language processing (NLP), requiring knowledge from all sub-areas of NLP. In an increasingly connected world, human interaction requires crossing language barriers in the government, business, social and cultural spheres. MT is a key technology to overcome these language barriers.
Cognitive NLP
Cognitive NLP research at the Centre for Indian Language Technology (CFILT), attempts to gain insights into the cognitive underpinnings of human language processing and understanding. The insights are then translated to methods and models that contribute to the field of NLP by achieving the following objectives: (1) Optimising human annotation effort for better annotation management, and (2) Improving existing NLP systems by introducing cognitive features.