March 2017

​

First journal paper accepted in IEEE Transactions on Fuzzy Systems (Impact factor 6.701)!

In another development, managed to improve the state-of-the-art on relation classification of i2b2 2010 data from 64% to 73% F-score, using a combination of attention, memory, and convolutions. Planning to evaluate the model on other datasets to prepare a paper for EMNLP 2017.

​

January 2017

​

Having received rave reviews on my TFS paper on the selection of fuzzy membership functions for multidimensional data, I am now devising and comparing techniques to make our proposed method robust to very high-dimensional data. The results are expected to be presented at IFSA 2017 and FUZZ-IEEE 2017. ​

​

October 2016

​

Working with Intel Egocentric Vision dataset for object detection from head-mounted camera images. Also created a smaller dataset from the GTEA data, which can be used to quickly train small models for recognizing common kitchen objects using limited resources. The dataset contains 3047 144x81 RGB images and 7 object classes. It can be found here.

​

September 2016

​

I will be working on using STNs for activity prediction from egocentric images. Was hoping to incorporate luminance invariance etc., into the project, but I am short on time and expertise on the subject.

​

August 2016

​

I have recently started working with Prof. Ashish Anand, Department of Computer Science and Engineering, IIT Guwahati, on relation extraction using convolutional neural networks and attention-based neural networks.

​

I am browsing through potential advisors and will start grad school applications for Fall 2017 in October.

​

July 2016

​

I presented my paper on "Visual Analysis and Representations of Type-2 Fuzzy Membership Functions" at the special session on "From Type-1 to Type-n Membership Functions" at the IEEE World Congress on Computational Intelligence, held at Vancouver, Canada. You can find the slides here.