Shreyan Ganguly

Shreyan Ganguly

Auto-generated profile, based on publicly available information

Shreyan Ganguly

Indian Institute of Science India

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
8 August 2022 in Expert Systems with Applications
Verified
Journal Article
Closed access
SGShreyan Ganguly
Shreyan Ganguly
AGAditya Ganguly
Aditya Ganguly
SMSk Mohiuddin
Sk Mohiuddin
SMSamir Malakar
Samir Malakar
RSRam Sarkar
Ram Sarkar

27 June 2022 in Pattern Analysis and Applications
Verified
Journal Article
Closed access
PDF file
SGShreyan Ganguly
Shreyan Ganguly
SMSk Mohiuddin
Sk Mohiuddin
SMSamir Malakar
Samir Malakar
Erik CuevasErik Cuevas
Erik Cuevas
RSRam Sarkar
Ram Sarkar

26 April 2022
Verified
Book Chapter
Closed access
PDF file
SMSk Mohiuddin
Sk Mohiuddin
SGShreyan Ganguly
Shreyan Ganguly
SMSamir Malakar
Samir Malakar
Dmitrii KaplunDmitrii Kaplun
Dmitrii Kaplun
RSRam Sarkar
Ram Sarkar

11 May 2024 in arXiv
Verified
Preprint
Open access
PDF file
SGShreyan Ganguly
Shreyan Ganguly
RNRoshan Nayak
Roshan Nayak
RRRakshith Rao
Rakshith Rao
UDUjan Deb
Ujan Deb
PAPrathosh AP
Prathosh AP
Knowledge distillation, a widely used model compression technique, works on the basis of transferring knowledge from a cumbersome teacher model to a lightweight student model. The technique involves jointly optimizing the task specific and knowledge distillation losses with a weight assigned to them. Despite these weights playing a crucial role in the performance of the distillation process, current methods provide equal weight to both losses, leading to suboptimal performance. In this paper, we propose Adaptive Knowledge Distillation, a novel technique inspired by curriculum learning to adaptively weigh the losses at instance level. This technique goes by the notion that sample difficulty increases with teacher loss. Our method follows a plug-and-play paradigm that can be applied on top of any task-specific and distillation objectives. Experiments show that our method performs better than conventional knowledge distillation method and existing instance-level loss functions.