Jihwan Bang

I'm

About

Senior AI Researcher & Engineer

I'm an AI/ML researcher with a passion for efficiency. Currently, I hold the role of Senior AI/ML Researcher at Qualcomm AI Research, where I lead cutting-edge projects in the field of artificial intelligence and machine learning. My journey in AI began at NAVER, where I honed my skills and gained invaluable experience. My expertise lies in developing AI models tailored for end devices, with a primary focus on areas like face recognition, face spoofing detection, and portrait segmentation. I've dedicated my career to crafting AI solutions that deliver exceptional performance while keeping resource usage in check. I'm not just about creating models; I'm also adept at building robust data and model serving pipelines. I specialize in maintaining model performance even when faced with limited label budgets, employing Active Learning techniques to make the most out of available resources. My approach to training models is centered around efficiency, utilizing Continual Learning methods to ensure optimal results. I'm always eager to connect with fellow enthusiasts and professionals in the AI/ML community.

Download CV

Research Interest

• Retrieval Augmented Generation (RAG)
• Large Language Model (LLM)
• Vision Language Model
• Continual / Lifelong Learning
• Active Learning
• On-device Learning

Education

M.S., School of Electrical Engineering

Mar. 2017 - Feb. 2019

Korean Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea

B.S., School of Electrical Engineering & Mathematical Science

Mar. 2012 - Feb. 2017

Korean Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea

Professional Experience

Qualcomm Logo

2023 - Present

Senior AI Researcher

  • Research and developed the chatbot-related products that is deployed in snapdragon.

Naver Logo

2019 - 2023

AI Researcher

  • Research and developed Facesign products that is employed in the headquarters.
  • Research and developed the product that is efficient and automated AI pipeline for selecting data and finetuning model

Projects

Hawkeye

Open-Vocabulary/Zero-Shot Object Detector
Research on open-vocabulary or zero-shot object detectors
Design API to use deployed zero-shot object detectors

Feb. 2022 - Apr. 2023

Delphi

Efficient and Automated ML Pipeline for selecting data and finetuning model
Design Delphi structure and build up each component in the pipeline
Apply active learning to ML pipeline for selecting informative data from the incoming streams

Mar. 2019 - Apr. 2023

FaceSign

Identity Authentication System using Face without Any Cards
Design the pipeline to get the best performance in FaceSign
Design spoofing detection model to prevent recognizing fake faces

Jul. 2020 - Jan. 2022

Observer

Drone-based Object Tracking System
Design AI platform for fog computing and build up the cloud-based AI platform
Implement deep reinforcement learning algorithm for autonomous drone control

Apr. 2017 - Dec. 2017

Publications

* authors contributed equally. See also my Google Scholar .
      Vision(CVPR, ICCV, ECCV, WACV, BMVC)          NLP(ACL, EMNLP, NAACL, EACL)          ML(ICML, ICLR, NeurIPS, AAAI)          Etc.


Conferences & Journals

  •   PR 2024  
    Prompt-Guided Transformers for End-to-End Open-Vocabulary Object Detection
    Hwanjun Song*, Jihwan Bang*
  •   ACL 2024 MAIN  
    Crayon: Customized On-Device LLM via Instant Adapter Blending and Edge-Server Hybrid Inference
    Jihwan Bang*, Juntae Lee*, Kyuhong Shim, Seunghan Yang, Simyung Chang
  •   ICML 2024  
    One Size Fits All for Semantic Shifts: Adaptive Prompt Tuning for Continual Learning
    Doyoung Kim, Susik Yoon, Dongmin Park, Youngjun Lee, Hwanjun Song, Jihwan Bang, Jae-Gil Lee
  •   CVPR 2024  
    Active Prompt Learning in Vision Language Models
    Jihwan Bang, Sumyeong Ahn, Jae-Gil Lee
  •   AAAI 2024  
    Adaptive Shortcut Debiasing for Online Continual Learning
    Doyoung Kim, Dongmin Park, Yooju Shin, Jihwan Bang, Hwanjun Song, Jae-Gil Lee
  •   ICCV 2023 ORAL  
    Generating Instance-level Prompts for Rehearsal-free Continual Learning
    Dahuin Jung, Dongyoon Han, Jihwan Bang, Hwanjun Song
  •   ICLR 2023  
    Self-Supervised Set Representation Learning for Unsupervised Meta-Learning
    Dong Bok Lee, Seanie Lee, Kenji Kawaguchi, Yunji Kim, Jihwan Bang, Jung-Woo Ha, Sung Ju Hwang
  •   ICLR 2023  
    Online Boundary-Free Continual Learning by Scheduled Data Prior
    Hyunseo Koh, Minhyuk Seo, Jihwan Bang, Hwanjun Song, Deokki Hong, Seulki Park, Jung-Woo Ha, Jonghyun Choi
  •   NeurIPS 2022  
    Meta-Query-Net: Resolving Purity-Informativeness Dilemma in Open-set Active Learning
    Dongmin Park, Yooju Shin, Jihwan Bang, Youngjun Lee, Hwanjun Song, Jae-Gil Lee
  •   CVPR 2022  
    Online Continual Learning on a Contaminated Data Stream with Blurry Task Boundaries
    Jihwan Bang, Hyunseo Koh, Seulki Park, Hwanjun Song, Jung-Woo Ha, Jonghyun Choi
  •   CVPR 2021  
    Rainbow Memory: Continual Learning with a Memory of Diverse Samples
    Jihwan Bang*, Heesu Kim*, YoungJoon Yoo, Jung-Woo Ha, Jonghyun Choi
  •   WACV 2020  
    SINet: Extreme Lightweight Portrait Segmentation Networks with Spatial Squeeze Modules and Information Blocking Decoder
    Hyojin Park, Lars Lowe Sjösund, Nicolas Monet, YoungJoon Yoo, Jihwan Bang, Nojun Kwak
  •   MILCOM 2019  
    CH-MAC: A Cluster-based, Hybrid TDMA MAC Protocol over Wireless Ad-hoc Networks
    Hoyong Choi, Jihwan Bang, Namjo Ahn, Jinhwan Jung, Jungwook Choi, Soobum Park, Yung Yi
  •   Mobihoc 2018  
    Incentivizing Hosts via Multilateral Cooperation in User-Provided Networks: A Fluid Shapley Value Approach
    Hyojung Lee, Jihwan Bang, Yung Yi

Preprints (ArXiv)

  •   Arxiv 2020  
    Boosting Active Learning for Speech Recognition with Noisy Pseudo-labeled Samples
    Jihwan Bang*, Heesu Kim*, YoungJoon Yoo, Jung-Woo Ha
  •   Arxiv 2019  
    ExtremeC3Net: Extreme Lightweight Portrait Segmentation Networks using Advanced C3-modules
    Hyojin Park, Lars Lowe Sjösund, YoungJoon Yoo, Jihwan Bang, Nojun Kwak

Talks

[T3] From Data Processing To Moedl Deploy: Introduction and Research Trends in MLOps Pipeline
Chung-Ang University (CAU)

May. 2024

[T2] Online Continual Learning on a Contaminated Data Stream with Blurry Task Boundaries
Korean Artificial Intelligence Association (KAIA)

Nov. 2022

[T1] Low Price, High Quality: Improve Model Performance with Small Labelling Cost
NAVER DEVIEW

Nov. 2020

Reviews

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

2021 - Present

IEEE/CVF European Conference on Computer Vision (ECCV)

2022 - Present

IEEE/CVF International Conference on Computer Vision (ICCV)

2023 - Present

International Conference on Machine Learning (ICML)

2022 - Present

Conference on Neural Information Processing Systems (NeurIPS)

2022 - Present

Inernational Conference on Learning Representations (ICLR)

2024 - Present

Conference on Lifelong Learning Agents (CoLLAs)

2023

Patents

[P2] Face Recognition System and Method for Controlling the Same
kr 10-2021-0183218
[P1] Efficient Active Learning for Automatic Speech Recognition via Augmented Consistency Regularization
kr 10-2020-0111929

Awards

  • CLOVA WOW(Innovative) Project: FaceSign
    NAVER CLOVA

    2021

  • National Science & Technology Scholarship
    Korean Scholarship Foundation

    2012-2015