Mainak Pal | মৈনাক পাল


I am a pre-final year undergraduate at Jadavpur University, majoring in Electronics and Telecommunication Engineering.

My research interest lies in Machine Learning and its crossroads in Computer Vision, Computational Neuroscience and Visual Cognition. I am a part of the Artificial Intelligence Laboratory, Jadavpur University, where I am advised by Dr. Amit Konar.

Presently, I am working at Serre Lab, Brown University on behavioral analysis of animal locomotion under Dr. Thomas Serre. I have been fortunate to work on Zero-shot Learning at VIP Lab, IIT Bombay under the kind guidance of Dr. Biplab Banerjee. Earlier, I have worked on Natural Language Processing at NLP Lab, JU. Here I was advised by Dr. Sudip Kumar Naskar.

CV  /  Google Scholar  /  LinkedIn  /  Github

profile photo

  • [May 2020]  our work has been accepted for publication as a book chapter at Interpretable Artificial Intelligence - A Perspective of Granular Computing, Springer-Verlag
  • [May 2020]  preprint of our work on transductive Zero-shot Learning is now available
  • [Apr 2020]  I'll be interning at Serre Lab, Brown University over Summer 2020
  • [Aug 2019]  awarded the RUSA 2.0 Travel Grant to present at ICCCMLA-2019, Goa, India
  • [Jul 2019]  paper on vector quantization clustering accepted at ICCCMLA-2019
  • [Mar 2019]  I'll be interning at VIP Lab, IIT Bombay over Summer 2019
  • [Jan 2019]  paper on sentiment analysis accepted at SemEval-2019

   Research Experience

Summer Research InternMay. 2020 - Present
Serre Lab, Brown University, USA
Supervisor: Dr. Thomas Serre


Remote InternJan. 2020 - Present
Xu Lab, CMU, USA
Supervisor: Dr. Min Xu


Undergraduate Research AssistantApr. 2019 - Present
Artificial Intelligence Laboratory, Jadavpur University, India
Supervisor: Dr. Amit Konar


Summer Research AssistantMay. 2019 - Jul. 2019
VIP Lab, IIT Bombay, India
Supervisor: Dr. Biplab Banerjee


Undergraduate Research AssistantAug. 2018 - May. 2019
NLP Lab, Jadavpur University, India
Supervisor: Dr. Sudip Kumar Naskar

*indicates equal contribution
A Generative Model Based Approach for Zero-shot Breast Cancer Segmentation Explaining Pixels’ Contribution to the Model’s Prediction
Preeti Mukherjee* Mainak Pal*, Lidia Ghosh, Amit Konar

Abstract accepted for publication as a book chapter at Interpretable Artificial Intelligence - A Perspective of Granular Computing, Springer-Verlag
Generative Model-Driven Structure Aligning Transductive Zero-shot Learning
Omkar Gune, Mainak Pal*, Preeti Mukherjee*, Biplab Banerjee, Subhasis Chaudhuri

Under review at Journal of Visual Communication and Image Representation, ELSEVIER
project / preprint / abstract / bibtex

Zero-shot Learning (ZSL) is a transfer learning technique which aims at transferring knowledge from seen classes to unseen classes. This knowledge transfer is possible because of underlying semantic space which is common to seen and unseen classes. Most existing approaches learn a projection function using labelled seen class data which maps visual data to semantic data. In this work, we propose a shallow but effective neural network-based model for learning such a projection function which aligns the visual and semantic data in the latent space while simultaneously making the latent space embeddings discriminative. As the above projection function is learned using the seen class data, the so-called projection domain shift exists. We propose a transductive approach to reduce the effect of domain shift, where we utilize unlabeled visual data from unseen classes to generate corresponding semantic features for unseen class visual samples. While these semantic features are initially generated using a conditional variational auto-encoder, they are used along with the seen class data to improve the projection function. We experiment on both inductive and transductive setting of ZSL and generalized ZSL and show superior performance on standard benchmark datasets AWA1, AWA2, CUB, SUN, FLO, and APY. We also show the efficacy of our model in the case of extremely less labelled data regime on different datasets in the context of ZSL.

title={Generative Model-driven Structure Aligning Discriminative Embeddings for Transductive Zero-shot Learning},
author={Omkar Gune and Mainak Pal and Preeti Mukherjee and Biplab Banerjee and Subhasis Chaudhuri},
Multi-resolution Hierarchical Clustering by Vector Quantization
Mainak Pal*, Preeti Mukherjee*, Amit Konar

Advances in Cybernetics, Cognition, and Machine Learning for Communication Technologies
Springer, Singapore
ICCCMLA'2019 Goa, India
project / code / abstract / bibtex / presentation

Clustering aims at grouping of objects or data-points based on a certain measure of similarity. Existing clustering algorithms estimate the measure of similarity of expected data-points to fall in a cluster with respect to presumed or computed cluster centroids. Such approach of distance measure between cluster centroid and possible data points to lie in cluster often results in misclustering, particularly for points equidistant to multiple cluster centroids. This paper offers an interesting solution to this problem by quantization of the attributes of the preferred cluster centroids and then checking the existence of the respective attributes of data-points within the quantized intervals for possible inclusion of data-point in the cluster. This approach, referred to as vector quantization offers additional merits of clustering at different user- defined resolutions of data-points of varying local density. Experiments undertaken confirm the superior performance of the proposed clustering over the state-of-art algorithms with respect to Jaccard coefficient on breast cancer dataset.

author="Pal, Mainak and Mukherjee, Preeti and Konar, Amit",
editor="Gunjan, Vinit Kumar and Senatore, Sabrina and Kumar, Amit and Gao, Xiao-Zhi and Merugu, Suresh",
title="Multi-resolution Hierarchical Clustering by Vector Quantization",
bookTitle="Advances in Cybernetics, Cognition, and Machine Learning for Communication Technologies", year="2020",
publisher="Springer Singapore",
Efficient Machine Learning and Neural Network Approaches for Identifying and Categorizing Offensive Language in Tweets
Preeti Mukherjee*, Mainak Pal*, Somnath Banerjee Sudip Kumar Naskar

Minneapolis, Minnesota, USA
project / code / abstract / bibtex

This paper describes our system submissions as part of our participation (team name: JU_ETCE_17_21) in the SemEval 2019 shared task 6: “OffensEval: Identifying and Catego- rizing Offensive Language in Social Media”. We participated in all the three sub-tasks: i) Sub-task A: offensive language identification, ii) Sub-task B: automatic categorization of of- fense types, and iii) Sub-task C: offense target identification. We employed machine learn- ing as well as deep learning approaches for the sub-tasks. We employed Convolutional Neural Network (CNN) and Recursive Neu- ral Network (RNN) Long Short-Term Memory (LSTM) with pre-trained word embeddings. We used both word2vec and Glove pre-trained word embeddings. We obtained the best F1- score using CNN based model for sub-task A, LSTM based model for sub-task B and Lo- gistic Regression based model for sub-task C. Our best submissions achieved 0.7844, 0.5459 and 0.48 F1-scores for sub-task A, sub-task B and sub-task C respectively.

title = "{JU}{\_}{ETCE}{\_}17{\_}21 at {S}em{E}val-2019 Task 6: Efficient Machine Learning and Neural Network Approaches for Identifying and Categorizing Offensive Language in Tweets",
author = "Mukherjee, Preeti and Pal, Mainak and Banerjee, Somnath and Naskar, Sudip Kumar",
booktitle = "Proceedings of the 13th International Workshop on Semantic Evaluation",
month = jun,
year = "2019",
address = "Minneapolis, Minnesota, USA",
publisher = "Association for Computational Linguistics",
url = "",
doi = "10.18653/v1/S19-2118",
pages = "662--667",
Manuscript in preparation
Structure Prediction in a Time-Series Using Vector Quantization Based Clustering
Preeti Mukherjee* Mainak Pal*, Amit Konar

Automatic behavioral analysis of C.Elegans locomotion
May'20 - Present

Working on various recurrent neural models to automate behavioral analysis of C.Elegans locomotion.

Supervisor: Dr. Thomas Serre

Generative adversarial approach for unsupervised domain adaptation
June'20 - Present

Extending our previous work on ZSL in unsupervised domain adaptation. Working on various generative models to achieve better latent layer representation of multimodal visual feature space.

Supervisor: Dr. Biplab Banerjee

Computational Vision in Cryo-electron Tomography
Jan'20 - Present

Exploring various methods to extract information from tomographic data.

Supervisor: Dr. Min Xu

Zero-shot breast cancer segmentation
Nov'19 - Mar'20

Trained the BiGan model on healthy data so that the trained model can construct nearest healthy samples from unhealthy data. Based on RISE model, we proposed a novel architecture for automatic segmentation of the tumor region from our previous understandings. Our approach is capable of segmenting tumors without using any unhealthy samples while training.

Supervisor: Dr. Amit Konar

Trunsductive Zero-shot learning
May'19 - July'19

Explored zero-shot learning application on various domains. Implemented various autoencoders on latent space and semantic space. Extended the Structure Aligning Discriminative Latent Embedding for Zero-Shot Learning in transductive settings.

Supervisor: Dr. Biplab Banerjee

Vector Quantization Clustering
Mar'19 - Feb'20

In existing clustering algorithms, larger attributes have more contribution in the distance measure in comparison to the attributes of small values. Thus, attributes of smaller values even if differ by larger magnitude are not encountered in the clustering algorithms - causing false clustering. To overcome this problem, we have proposed an novel clustering algorithm based on quantization at each attribte level. Our approach performs better than state-of-arts and also computationally less expensive. We have worked on several gene-micro array datasets, breast cancer dataset. Proposed method is also helpful in time-series modeling.
[publication] [manuscript in preparation]

Supervisor: Dr. Amit Konar

Identifying and Categorizing Offensive Language in Social Media
Sep'18 - Feb'19

Worked on sentiment analysis of tweets. Explored multiprocessing. Explored different techniques of machine learning ( like Logistic Regression, Linear SVC, LinearSVC with L1-based feature selection, Multinomial NB, Bernoulli NB etc. ). Implemented several Deep Learning networks like CNN-word2vec, attention based Bi - RNN with LSTM. Our team participated at task 6 on SemEval-2019. Tasks were three fold. First, Offensive language identification. Second, if the tweet is offensive, then automatic categorization of whether that is targeted or not. Third, if the tweet is targeted, then target of offence identification, like: is it targeting individual, group or others? Our methods performed well and task description paper published at SemEval.

Supervisor: Dr. Sudip Kumar Naskar

Some exploratory voyages...

Perro Gato : An Image Classifier [code]
Real-time 2D plot of Azimuth Plane using Ultrasonic Wave Sensor (HC SR04) [code] [demo]
Approximate-Pi [code] [demo]
Cloud : A nodejs app deployed aiming to help beginners finding Open Source projects [code] [demo]
Mosom : A weather forecast app built in PyQt [code]
Shoot The Ball : A game developed using Processing [code] [demo]

IEEE Computer Society Student Branch Chapter, Jadavpur University

Member [Mar'20 - Present]
Founder and Chairperson [Mar'19 - Mar'20]
IEEE Student Branch, Jadavpur University

Member [Mar'20 - Present]
Webmaster, Tech lead [Mar'19 - Mar'20]

You surely have seen this website template before :P. Thanks, Jon Barron!