Md Tanvirul Alam

Md Tanvirul Alam

PhD Student, RIT

Publications

Punctuation Restoration using Transformer Models for High-and Low-Resource Languages

Published in Proceedings of the 6th Workshop on Noisy User-generated Text (W-NUT 2020)@EMNLP, 2020

We used different Transformer models for the punctuation restoration task on a high (English) and low (resource) language. We also proposed a novel augmentation strategy tailored for improving performance on ASR transcriptions.

Recommended citation: T. Alam, A. Khan, and F. Alam, “Punctuation Restoration using Transformer Models for High-and Low-Resource Languages,” in Proceedings of the 6th Workshop on Noisy User-generated Text (W-NUT2020)@EMNLP. 2020. http://noisy-text.github.io/2020/pdf/2020.d200-1.18.pdf

Social Media Image Classification Benchmarks for Various Disaster Response Tasks

Published in ASONAM, 2020

This work consolidates disaster image classification datasets from various sources and across different tasks. Baseline results using different CNN models are also provided.

Recommended citation: F. Alam, F. Ofli, M. Imran,T. Alam, and U. Qazi, “Social Media Image Classification Benchmarks for Various Disaster Response Tasks,” in 2020 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM). IEEE, 2020

Lightweight CNN for Robust Voice Activity Detection

Published in International Conference on Speech and Computer (SPECOM), 2020

In this work, we designed a lightweight CNN architecture and improved its performance under noisy conditions using strong regularization (SpecAugment and DropBlock) and knowledge distillation techniques.

Recommended citation: T. Alam and A. Khan, “Lightweight CNN for Robust Voice Activity Detection,” in International Conference on Speech and Computer(SPECOM). Springer, 2020, pp. 1–12. https://link.springer.com/chapter/10.1007/978-3-030-60276-5_1