Research Highlights

Members of AI@IISc Group conduct cutting-edge research on a variety of aspects of machine learning and related fields, ranging from theoretical foundations to new algorithms as well as several exciting applications; this research is regularly published in top-tier journals and conferences. The following are some examples of recent research topics explored in the group (further details can often be found on faculty members’ homepages):

  • Bioinformatics
  • Blockchain Technology
  • Clustering
  • Computational Social Choice
  • Computer Vision
  • Data Mining
  • Deep Learning
  • Game Theory and Mechanism Design
  • Graphical Models
  • High Performance Computing for AI
  • Image Processing
  • Knowledge Graphs
  • Machine Learning in Computer Systems
  • Multiarmed Bandit Mechanisms
  • Natural Language Understanding
  • Ranking
  • Reinforcement Learning
  • Robust Learning Methods for Uncertain Data
  • Spectral Algorithms
  • Statistical Learning Theory, Statistical Consistency of Learning Algorithms
  • Structured Prediction
  • Support Vector Machines and Kernel based Learning Methods

Symmetry in Scalar Fields

Vijay Natarajan

Automatic detection of symmetry is a challenging problem because both the segmentation of the domain into potential symmetric segments and the correspondence between segments that are symmetric need to be determined. Moreover, real life data sets never exhibit perfect symmetry. Since the search space for locating symmetric segments is quite large, it is also important to design an algorithm that is computationally efficient. Domain experts are interested in studying important features that provide insights about the underlying scientific phenomena that is being analyzed.

We have developed four different methods to detect symmetry in scalar fields. We have also demonstrated applications to symmetry-aware transfer function design, symmetry-aware isosurface extraction, and symmetry-aware editing and rendering enhance traditional visualization methods. We believe that the methods we propose for symmetry detection will open new frontiers in analyzing structural similarity of scalar fields and more applications of symmetry detection will emerge.


Task agnostic Universal Adversarial Perturbations

Mopuri Reddy, Aditya Ganeshan and R. Venkatesh Babu

Machine learning models are susceptible to adversarial perturbations: small changes to input that can cause large changes in output. It is also demonstrated that there exist input-agnostic perturbations, called universal adversarial perturbations, which can change the inference of target model on most of the data samples. However, existing methods to craft universal perturbations are (i) task specific, (ii) require samples from the training data distribution, and (iii) perform complex optimizations. Additionally, because of the data dependence, fooling ability of the crafted perturbations is proportional to the available training data. In this project, we present a novel, generalizable and data-free approach for crafting universal adversarial perturbations. Independent of the underlying task, our objective achieves fooling via corrupting the extracted features at multiple layers. Therefore, the proposed objective is generalizable to craft image-agnostic perturbations across multiple vision tasks such as object recognition, semantic segmentation, and depth estimation. In the practical setting of black-box attack scenario (when the attacker does not have access to the target model and it’s training data), we show that our objective outperforms the data dependent objectives to fool the learned models. Further, via exploiting simple priors related to the data distribution, our objective remarkably boosts the fooling ability of the crafted perturbations. Significant fooling rates achieved by our objective emphasize that the current deep learning models are now at an increased risk, since our objective generalizes across multiple tasks without the requirement of training data for crafting the perturbations.

Cross-Modal Retrieval

Soma Biswas

Due to increase in the number of sources of data, research in cross-modal matching is becoming an increasingly important area of research. It has several applications like match- ing text with image, matching near infra-red images with visible images (eg, for matching face images captured during night-time or low-light conditions to standard visible light images in the database), matching sketch images with pictures for forensic applications, etc. We are developing novel algorithms for this problem, which is extremely challenging, due to significant differences between data from different modalities. Specifically, we have developed a generalized semantic preserving hashing technique for cross-modal retrieval algorithms, which can work seamlessly for single and multi-label data, as well as in paired and unpaired scenarios. The algorithm obtained state-of-the-art performance in different applications.

Using Statistical Mechanics to understand depth in Deep Networks

Chiranjib Bhattacharyya

Understanding the representational power of Restricted Boltzmann Machines (RBMs) with multiple layers is an ill-understood problem and is an area of active research. Motivated from the approach of Inherent Structure formalism (Stillinger & Weber, 1982), extensively used in analysing Spin Glasses, we propose a novel measure called Inherent Structure Capacity (ISC), which characterizes the representation capacity of a fixed architecture RBM by the expected number of modes of distributions emanating from the RBM with parameters drawn from a prior distribution. We introduce Lean RBMs, which are multi-layer RBMs where each layer can have at-most O(n) units with the number of visible units being n. We show that for every single layer RBM with Omega(n^{2+r}), r >= 0, hidden units there exists a two-layered lean RBM with Theta(n^2) parameters with the same ISC, establishing that 2 layer RBMs can achieve the same representational power as single-layer RBMs but using far fewer number of parameters. To the best of our knowledge, this is the first result which quantitatively establishes the need for layering.

Fairness in an Algorithmic World

Siddharth Barman

Siddharth’s recent work considers fairness from a computation lens. For example, one of his recent works shows that one can compute outcomes, in economic systems, which are efficient and fair at the same time, i.e., the seeming incompatible properties of efficiency and fairness can be achieved together. Given that fairness is a fundamental consideration in many resource-allocation problems, these results can potentially influence allocation policies in practical settings. Siddharth is also interested in developing fairness guarantees in machine-learning contexts such as clustering and classification.

Network Consistent Data Association

Anirban Chakraborty

Existing data association techniques mostly focus on matching pairs of data-point sets and then repeating this process along space-time to achieve long term correspondences. However, in many problems such as person re-identification, a set of data-points may be observed at multiple spatio-temporal locations and/or by multiple agents in a network and simply combining the local pairwise association results between sets of data-points often leads to inconsistencies over the global space-time horizons. In this paper, we propose a Novel Network Consistent Data Association (NCDA) framework formulated as an optimization problem that not only maintains consistency in association results across the network, but also improves the pairwise data association accuracies. The proposed NCDA can be solved as a binary integer program leading to a globally optimal solution and is capable of handling the challenging data-association scenario where the number of data-points varies across different sets of instances in the network. We also present an online implementation of NCDA method that can dynamically associate new observations to already observed data-points in an iterative fashion, while maintaining network consistency. We have tested both the batch and the online NCDA in two application areas—person re-identification and spatio-temporal cell tracking and observed consistent and highly accurate data association results in all the cases.

Neural programming and program analysis

Aditya Kanade, Shirish Shevade

Automatically synthesising programs and analysing their behaviour is considered a holy grail of Artificial Intelligence. We work developing neural network based solutions towards automating these tasks. Our recent work has been published in venues such as FSE’16, AAAI’17 and AIED’18.

Modelling vehicle dynamics using Indian road traffic videos

Deep Neural Networks were employed to identify different vehicles in videos obtained from traffic junctions

AI/ML for Next Generation Communication Systems

Chandra R. Murthy

We are investigating the use of AI/ML techniques for communications related applications. An immediate use case seems to be to use these techniques to solve otherwise-hard problems by training a neural network. Issues such as limited training size, model generalizability, etc are the key ones under investigation.