Research

 

Background and Interests

Being a technology lover I am naturally interested in AI, Machine Learning and cool ways in which we can make machines awesome. I finished my Bachelor’s in Computer Science from IIT Kharagpur in 2016. In my Bachelors I mostly worked on projects involving Machine Learning and Natural Language. At present, I am working for Google in the YouTube video classification team so we can serve better ads. And in my spare time, I try to be in touch with the recent developments in ML or AI in general.

You can find my detailed cv here.

Articles/Publications

Xinchen Yan, Mohi Khansari, Yunfei Bai, Jasmine Hsu, Arkanath Pathak, Abhinav Gupta, James Davidson, Honglak Lee
Learning grasping interaction with geometry-aware 3d representations
arXiv preprint arXiv:1708.07303, 2017

[BibTeX] [Download PDF]
@article{yan2017learning,
  title={Learning grasping interaction with geometry-aware 3d representations},
  author={Yan, Xinchen and Khansari, Mohi and Bai, Yunfei and Hsu, Jasmine and Pathak, Arkanath and Gupta, Abhinav and Davidson, James and Lee, Honglak},
  journal={arXiv preprint arXiv:1708.07303},
  year={2017}
}

Arkanath Pathak, Pawan Goyal, Plaban Bhowmick
A Two-Phase Approach Towards Identifying Argument Structure in Natural Language
arXiv preprint arXiv:1612.05420, 2016
[BibTeX] [Download PDF]
@article{pathak2016two,
  title={A Two-Phase Approach Towards Identifying Argument Structure in Natural Language},
  author={Pathak, Arkanath and Goyal, Pawan and Bhowmick, Plaban},
  journal={arXiv preprint arXiv:1612.05420},
  year={2016}
}

Arkanath Pathak, Nikhil R Pal
Clustering of mixed data by integrating fuzzy, probabilistic, and collaborative clustering framework
International Journal of Fuzzy Systems, 18(3): 339-348, 2016
[BibTeX] [Download PDF]
@article{pathak2016clustering,
  title={Clustering of mixed data by integrating fuzzy, probabilistic, and collaborative clustering framework},
  author={Pathak, Arkanath and Pal, Nikhil R},
  journal={International Journal of Fuzzy Systems},
  year={2016},
  number={3},
  pages={339--348},
  publisher={Springer Berlin Heidelberg}
}

Projects

Improving YouTube personalization using clustering of videos (Internship at Google, Bangalore)

People Involved: Sumit Sanghai (Internship Mentor), Vivek Sahasranaman, Ajai Tirumali

Duration: May – July, 2015

Description: We explored new ways to improve YouTube user profiles by trying to find ways of modelling interests that are not well represented by Knowledge Graph entities (e.g. ”70’s music”). Towards this goal, we used clusters of videos as users’ features. The input data for the clustering was derived from video correlations due to user co-watches. We tried k-means, HAC and LDA to generate video clusters. We built a simple video recommendation system using clusters as features. We had to deal with large amounts of input data, and thus, had to use compute clusters for distributing the tasks. Consequently, the project also involved heavy usage of distributed frameworks, like MapReduce. We also tried ways to generate cluster descriptors using n-gram language models and video entities.

Clustering of mixed data by integrating fuzzy, probabilistic and collaborative clustering framework (Research project carried at Indian Statistical Institute, Kolkata)

People Involved: Prof. Nikhil R. Pal (Project Guide)

Duration: May – June, 2014

Description: A new algorithm for clustering data with both numerical and categorical attributes was developed. Our work was published in International Journal of Fuzzy Systems.

A two-phase approach towards identifying argument structure in natural language (B.Tech. Project)

People Involved: Dr. Pawan Goyal and Dr. Plaban Bhowmick

Duration: February 2015 – April 2016

Description: The project is related to a relatively new research problem called Argumentation Mining. We present a two-stage strategy for extracting argument structure in a natural language text with an underlying argument. The task of automated identification of argument structure is difficult since it involves the problem of natural language inference. Furthermore, in the case of arguments the relationship is much more complex. Our work was presented as a full paper at NLPTEA 2016, held in conjunction with COLING 2016.