Tin Nguyen-Thanh

I am a PhD student at Auburn University, Alabama, US where I am working on XAI in multimodalities. I am also passionate about Embodied AI where an AI can have the ability to understand not only language but also other different modalities such as vision, sound.

At Auburn University, I've been working under the supervision of Professor Anh Nguyen.

ngthanhtinqn@gmail.com  /  CV  /  Blog  /  Google Scholar  /  ResearchGate  /  Short Science  /  Notion  /  Twitter  /  Github /  Coding / 

Category: Latest New  /  Research Interest  /  Experience  /  Publications /  Projects /  Talks /  Academic Service

2024

March 13, 2024, One paper got accepted @ NAACL 2024 Findings.

2023

Jul 4, 2023, One paper got accepted @ Knowledge-based Systems (IF: 8.8).

2022

Nov 1, 2022, One paper got accepted @ Expert Systems with Applications (IF: 10.35).

• Interpretable AI: Part-based, Concept-based XAI.
• Domain Generalization.

I did my Master at Sejong University, Seoul, Korea on Aug 2022. In this period, I was working on Deep Reinforcement Learning for Instruction Following Navigation with Professor Yong-Guk Kim.

I did my Bachelors at the Vietnam National University, The University of Science, Ho Chi Minh City on Aug 2019. My thesis was about Robot Localization and Object Tracking with Professor Ly Quoc Ngoc.

We proposed a part-based bird classifier that makes predictions based on part-wise descriptions. Our method directly utilizes human-provided descriptions (in this work, from GPT4). It outperforms CLIP and M&V by 10 points in CUB and 28 points in NABirds.

We are the first to explore integrating habitat information, one of the four major cues for identifying birds by ornithologists, into modern bird classifiers.

This study proposes an attention-based module for analyzing the sentiment and emotion of memes.

This study proposes a coarse-to-fine fusion module between vision and language. This will help an agent learn a joint representation while navigating in a virtual environment.

Introducing Fruit-CoV, a two-stage vision framework, which is capable of detecting SARS-CoV-2 infections through recorded cough sounds. In this challenge, we won 100mil VND (~ $4275) for the 1st place.

Applying Yolo3 to Particle Filter to enhance its speed and accuracy, furthermore, an end-to-end localization framework using a stereo camera and IMU in the unknown environment.

Achieved 1st on the public leaderboard, applying SAN, multihop, CNNRoBerta as multimodalities, and Only Text and Image as Single modalities.

[Most interesting discussed idea award] Achieved 3rd on the private leaderboard, applying Swin Transformer as the Encoder (and other types), and LSTM Attention as the Decoder.

Choosen to be in the Special Issue of VNU Journal of Science (JCSCE)

Implement by C++. Project using Background Subtraction Method, Kalman Filter, and Hungary Algorithm.

Multi-Agent (2 robots) for Instruction Following Navigation task.

Here, I list down the papers that I have reviewed/presented:

• [Oct 3, 2023] CLIP-Event: Connecting Text and Images with Event Structures
• [Apr 25, 2023] Zero-Shot Classification by Logical Reasoning on Natural Language Explanations
• [Mar 21, 2023] Open-Vocabulary Semantic Segmentation With Mask-Adapted CLIP
• [Feb 14, 2023] STAIR: Learning Sparse Text and Image Representation in Grounded Tokens
• [Nov 22, 2022] Re-labeling ImageNet - from Single to Multi-Labels, from Global to Localized Labels
• [Oct 18, 2022] Class Activation Latent Mapping - Keep CALM and Improve Visual Feature Attribution
• [Sep 20, 2022] Mask2Former: Masked-attention Mask Transformer for Universal Image Segmentation
  

• ACL (Top Conference in NLP)
• Knowledge-based Systems (Q1 Journal)

• [Aug, 2022 - May, 2023] Teacher at K-6 AI Club: AI Club for children around the elementary-school-age (K-6) to learn math, coding, robotics, and artificial intelligence. This is a completely FREE, voluntary, educational event. Supported by Auburn University and an NSF CAREER award.

Nguyen Thanh Tin