Matthew Bodenham

PhD Candidate

IDS Lab

Electrical Engineering & Computer Science

Daegu Gyeongbuk Institute of Science and Technology (DGIST)

333 Techno Jungang-daero

Dalseong-gun, Daegu, 42988

Republic of Korea

Email: mbodenham@dgist.ac.kr

I am a second year PhD candidate at IDS Lab, DGIST, South Korea. My research topics are in searching for efficient language models for edge devices, and vision based gaze tracking for augmented reality driving assistance. My research interests are in efficient AI, neural architecture search, and natural language processing on the edge.

I completed my MSc Data Science at the University of Bath, United Kingdom. My research project involved a novel approach in applying model compression methods to a state-of-the-art salient object detection models for the use in video compression. The project successfully applied filter pruning to Cascaded Partial Decoder for Fast and Accurate Salient Object Detection (Wu et al., CVPR 2019) with the result of running the compressed model on a mobile device.

Outside of academia, I am interested in Korean Language, Culture and Food. I spent nine months from 2018 to 2019 studying Korean at Seoul National University Language Institute. Currently, I am living and studying in Daegu, South Korea.

Conferences

  1. M. Bodenham, J. Kung, "On the Extensive Exploration of Transformer-based Language Models for Memory and Latency Bounded Edge Devices", Workshop on 2nd ROAD4NN, IEEE/ACM Design Automation Conference (DAC), December 2021.