Technocraft Center for Applied Artificial Intelligence
-
LOG IN
Projects
Round 1
The crowdsourcing platform is assumed to be divided into multiple classes, based on workers skill, experience, etc. We consider the problem of cost optimal utilization of a crowdsourcing platform for binary, unsupervised classification of a collection of items, given a prescribed error threshold.
Faculty associated
Prof. N. Hemachandra (IEOR) and Prof. Jayakrishnan Nair (EE)
Learning for Beam alignment in 5G
In this work, we propose to develop learning algorithms for beam alignment between base stations and various users to maintain high rates. One thus needs regular updates of sufficiently good estimates of the positions of all the mobiles.
Faculty associated
Prof. Veeraruna Kavita(IEOR) and Prof. Manjesh K. Hanawal(IEOR)
Automating Threat Detection and Response in Linux Endpoints
Endpoint Detection and Response EDR is an advanced technology for the detection and prevention of attacks on cyberinfrastructure. It is an integrated security solution that combines realtime continuous monitoring and collection of endpoint data with rules-based automated response and analysis capabilities.
Faculty associated
Prof. Manjesh K. Hanawal (IEOR)
3D Medical Image data synthesis for classification and segmentation using deep generative techniques abeling 3D medical images
Typical prediction tasks using 3D medical images like classification and segmentation using deep neural networks require large amounts of labeled training data. In this project, we will develop novel variants of deep networks e.g. generative adversarial networks GANs, variational encoders VAEs to synthesize 3D medical image data for classification and segmentation tasks.
Faculty associated
Prof. P. Balamurugan (IEOR)
Round 2
Assistive AI Technology Development for Tactical and Operational Planning of Supply Chains
Modern supply-chain networks are highly competitive among the entities who operate at the retailer level interfacing directly with the stochastic demands of the consumers. At the same time, the retailers/distributors are well connected to each other via a transportation network. Our project aims to develop and analyze an explainable framework for these two objectives using game theoretic modeling and by providing mechanism design solutions.
Faculty associated
Dr. Swaprava Nath, Assistant Professor, Dept of CSE, IIT Bombay
AI/ML Applications for Enhanced Smart Metering for Residential Electricity Consumption
Residential electricity consumption accounts for 24% of the total electricity demand in India and is the largest connected on the Indian power grid. Smart metering allows utilities to continuously monitor and potentially control this load. In this project, we are working with SustLabs to develop AI/ML-based algorithms for load disaggregation at the smart meter level.
Faculty associated
Dr. Anupama Kowli, Associate Professor, Dept of EE, IIT Bombay
Robust domain adaptation strategies for vibration condition monitoring of machines at the edge.
In this project, we are investigating compact machine learning algorithms implementable on microcontrollers for diagnosing faults and anomalies in machines based on vibration sensing. The focus is on development of light-weight algorithms that are immune to domain shift and concept drift (i.e. scalable to large number of various types of machines)
Faculty associated
Siddharth Tallur, Associate Professor, Electrical Engineering, IIT Bombay
Pattern Recognition-AI and NMR aided HOS analysis of Biological drugs.
Biosimilar refers to the biological drug that is highly similar to the available marketed drug such that there are no clinically meaningful differences in terms of safety, purity, and potency of the product. These drugs can only enter the market after completing the patent duration of the licensed first-generation drugs.
Faculty associated
Prof. Ashutosh Kumar, Professor, Department of Biosciences and Bioengineering, IIT Bombay
Categorization of Landfill Mined Residues-coarse fractions using AI and ML techniques.
A major fraction of the municipal solid waste generated, directed toward landfills, poses significant challenges like insufficient land, water contamination, and greenhouse emissions. To cater to these issues, landfill mining is being promoted worldwide. The current manual practices of landfill mined residues (LMRs) segregation being inefficient and hazardous to labor makes automation an inevitable alternative. This project aims the development of low-cost field implementable AI/ML-based techniques which integrate the infrared and visible response of LMR to ease the automation of their identification. Further, with the help of our industry partner, M/s Dalmia Cements, an attempt would be made to scale up the findings from this study for industrial implementation.
Faculty associated
Prof. Devendra Narain Singh, Department of Civil Engineering, IIT Bombay
