Causality in Microbiomes: Inferring causality is the process of connecting a cause with an effect. Identifying even a single causal relationship from data is more valuable than observing dozens of correlations in a data set. The study of causality is not new in many areas of science, but in recent years with the advances in artificial intelligence, Bayesian networks, causal calculus, data science, and machine learning, the question has become “how to draw a causal conclusion in a data-driven way?”. Given a sufficiently large and rich data set, the theoretical foundations of causality allows us to go well beyond merely discovering statistical associations in data, but to infer causal relationships in a quantitative manner and to even explore “what-if” questions, which can have a profound impact on data-driven decision making in any domain. Learning causal inference has been compared to human level intelligence. In my dissertation, I am developing a causal framework for the highly dynamic, interdependent, and complex data sets generated from microbiome studies. A microbiome is a community of microbes including bacteria, archaea, protists, fungi and viruses that share an environmental niche. Microbiomes have been referred to as a social network because of the complex set of potential interactions between its various taxonomic members. Microbiome studies have recently been augmented with the collection of multiomics data that allows a glimpse into different aspects of the microbial community. This dissertation will explore the application of the causality to data from microbiome studies. Data from these studies also include multiomics data sets and data from longitudinal studies. This work will improve our understanding of microbial communities inside human bodies and in environmental settings, and will help develop therapies when imbalances arise in these communities.
Facial Key-Point Detection: In my early days of graduate school, I worked on computer vision and pattern recognition. Facial key-points detection from a three-dimensional model using deep learning. We started with raw 3D geometric mesh data and then applied harmonic mapping to convert the meshes to images. We finally detected key-points from the mapped image using a convolutional neural network (CNN).
Bangla Handwritten Character Recognition: Recognition of Bangla handwritten characters is a difficult but important task for various emerging applications. For better recognition performance, good feature representation of the character images is a primary requirement. In this study, we investigate a recently proposed machine learning approach called deep learning for Bangla hand written character recognition, with a focus on automatic learning of good representations. This approach differs from the traditional methods of preprocessing the characters for constructing the handcrafted features such as loops and strokes. Among different deep learning structures, we employ the deep belief network (DBN) that takes the raw character images as input and learning proceeds in two steps - an unsupervised feature learning followed by a supervised fine tuning of the network parameters. Unlike traditional neural networks, the DBN is a probabilistic generative model, i.e., we can generate samples from the model and it can fit both the semi-supervised and supervised learning settings. We demonstrate the advantages of unsupervised feature learning through the experimental studies carried on the Bangla basic characters and numerals data set collected from the Indian Statistical Institute. In my undergraduate thesis, I worked on this domain using deep learning and to the best of our knowledge that was the first attempt to apply deep neural networks for Bangla language. Later we extended our study and employed convolutional neural network (CNN) and achieved better performance.
Pet-Projects:
In my free time I usually work some fun projects. Some of the fun projects are in the following:
Sports Detection: I work on the recognition of different sports and gestures from a particular game (for example, referee's call in soccer). Many of those problems are unsolved but culturally and commercially valuable. However, having a good data set is a major obstacle, that is why transfer learning can shed light.
Bangla Musical Genre Classification: Bangla musics are very rich and diverse. There is no tool available to classify genre especially from Bangla music. Traditional way for music is to use hand-crafted features which are not feasible in this big data arena. Our focus is on representation learning to solve this problem.