Publications by Tags

Alex-GDA

Fundamental Benefit of Alternating Updates in Minimax Optimization

👥 Jaewook Lee*, Hanseul Cho*, and Chulhee Yun

📰 ICML 2024 (Short version at ICLR 2024 Workshop on Bridging the Gap Between Practice and Theory in Deep Learning (BGPT)) 🗓 May 02, 2024 🔗 [paper] [arxiv] [code]

🎉 Spotlight @ ICML 2024 (Top 3.5% among total submissions)

Algorithmic Tasks

Position Coupling: Improving Length Generalization of Arithmetic Transformers Using Task Structure

👥 Hanseul Cho*, Jaeyoung Cha*, Pranjal Awasthi, Srinadh Bhojanapalli, Anupam Gupta, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Long-Context Foundation Models (LCFM)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Arithmetic Tasks

Position Coupling: Improving Length Generalization of Arithmetic Transformers Using Task Structure

👥 Hanseul Cho*, Jaeyoung Cha*, Pranjal Awasthi, Srinadh Bhojanapalli, Anupam Gupta, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Long-Context Foundation Models (LCFM)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count

👥 Hanseul Cho*, Jaeyoung Cha*, Srinadh Bhojanapalli*, and Chulhee Yun

📰 Arxiv Preprint 🗓 October 22, 2024 🔗 [arxiv] [code]

Continual Learning

Convergence and Implicit Bias of Gradient Descent on Continual Linear Classification

👥 Hyunji Jung*, Hanseul Cho*, and Chulhee Yun

📰 Under Review 🗓 November 16, 2024

🎉 Best Paper Award at the 11th Joint Conference of Korean Artificial Intelligence Association (JKAIA 2024)

Convergence

Convergence and Implicit Bias of Gradient Descent on Continual Linear Classification

👥 Hyunji Jung*, Hanseul Cho*, and Chulhee Yun

📰 Under Review 🗓 November 16, 2024

🎉 Best Paper Award at the 11th Joint Conference of Korean Artificial Intelligence Association (JKAIA 2024)

DASH

DASH: Warm-Starting Neural Network Training in Stationary Settings without Loss of Plasticity

👥 Baekrok Shin*, Junsoo Oh*, Hanseul Cho, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Advancing Neural Network Training (WANT)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Direction-Aware SHrinking

DASH: Warm-Starting Neural Network Training in Stationary Settings without Loss of Plasticity

👥 Baekrok Shin*, Junsoo Oh*, Hanseul Cho, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Advancing Neural Network Training (WANT)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Fairness

Fair Streaming Principal Component Analysis: Statistical and Algorithmic Viewpoint

👥 Junghyun Lee*, Hanseul Cho*, Se-Young Yun, and Chulhee Yun

📰 NeurIPS 2023 🗓 September 22, 2023 🔗 [paper] [arxiv] [code]

Generalization

DASH: Warm-Starting Neural Network Training in Stationary Settings without Loss of Plasticity

👥 Baekrok Shin*, Junsoo Oh*, Hanseul Cho, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Advancing Neural Network Training (WANT)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Gradient Descent

Convergence and Implicit Bias of Gradient Descent on Continual Linear Classification

👥 Hyunji Jung*, Hanseul Cho*, and Chulhee Yun

📰 Under Review 🗓 November 16, 2024

🎉 Best Paper Award at the 11th Joint Conference of Korean Artificial Intelligence Association (JKAIA 2024)

Gradient Descent-Ascent (GDA)

Fundamental Benefit of Alternating Updates in Minimax Optimization

👥 Jaewook Lee*, Hanseul Cho*, and Chulhee Yun

📰 ICML 2024 (Short version at ICLR 2024 Workshop on Bridging the Gap Between Practice and Theory in Deep Learning (BGPT)) 🗓 May 02, 2024 🔗 [paper] [arxiv] [code]

🎉 Spotlight @ ICML 2024 (Top 3.5% among total submissions)

Implicit Bias

Convergence and Implicit Bias of Gradient Descent on Continual Linear Classification

👥 Hyunji Jung*, Hanseul Cho*, and Chulhee Yun

📰 Under Review 🗓 November 16, 2024

🎉 Best Paper Award at the 11th Joint Conference of Korean Artificial Intelligence Association (JKAIA 2024)

Incremental Learning

DASH: Warm-Starting Neural Network Training in Stationary Settings without Loss of Plasticity

👥 Baekrok Shin*, Junsoo Oh*, Hanseul Cho, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Advancing Neural Network Training (WANT)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Length Generalization

Position Coupling: Improving Length Generalization of Arithmetic Transformers Using Task Structure

👥 Hanseul Cho*, Jaeyoung Cha*, Pranjal Awasthi, Srinadh Bhojanapalli, Anupam Gupta, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Long-Context Foundation Models (LCFM)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count

👥 Hanseul Cho*, Jaeyoung Cha*, Srinadh Bhojanapalli*, and Chulhee Yun

📰 Arxiv Preprint 🗓 October 22, 2024 🔗 [arxiv] [code]

Linear Classification

Convergence and Implicit Bias of Gradient Descent on Continual Linear Classification

👥 Hyunji Jung*, Hanseul Cho*, and Chulhee Yun

📰 Under Review 🗓 November 16, 2024

🎉 Best Paper Award at the 11th Joint Conference of Korean Artificial Intelligence Association (JKAIA 2024)

Loss of Plasticity

DASH: Warm-Starting Neural Network Training in Stationary Settings without Loss of Plasticity

👥 Baekrok Shin*, Junsoo Oh*, Hanseul Cho, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Advancing Neural Network Training (WANT)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Out-of-distribution Generalization

Position Coupling: Improving Length Generalization of Arithmetic Transformers Using Task Structure

👥 Hanseul Cho*, Jaeyoung Cha*, Pranjal Awasthi, Srinadh Bhojanapalli, Anupam Gupta, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Long-Context Foundation Models (LCFM)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count

👥 Hanseul Cho*, Jaeyoung Cha*, Srinadh Bhojanapalli*, and Chulhee Yun

📰 Arxiv Preprint 🗓 October 22, 2024 🔗 [arxiv] [code]

Plasticity

PLASTIC: Improving Input and Label Plasticity for Sample Efficient Reinforcement Learning

👥 Hojoon Lee*, Hanseul Cho*, Hyunseung Kim*, Daehoon Gwak, Joonkee Kim, Jaegul Choo, Se-Young Yun, and Chulhee Yun

📰 NeurIPS 2023 🗓 September 22, 2023 🔗 [paper] [arxiv] [code]

Position Coupling

Position Coupling: Improving Length Generalization of Arithmetic Transformers Using Task Structure

👥 Hanseul Cho*, Jaeyoung Cha*, Pranjal Awasthi, Srinadh Bhojanapalli, Anupam Gupta, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Long-Context Foundation Models (LCFM)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count

👥 Hanseul Cho*, Jaeyoung Cha*, Srinadh Bhojanapalli*, and Chulhee Yun

📰 Arxiv Preprint 🗓 October 22, 2024 🔗 [arxiv] [code]

Positional Encoding

Position Coupling: Improving Length Generalization of Arithmetic Transformers Using Task Structure

👥 Hanseul Cho*, Jaeyoung Cha*, Pranjal Awasthi, Srinadh Bhojanapalli, Anupam Gupta, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Long-Context Foundation Models (LCFM)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count

👥 Hanseul Cho*, Jaeyoung Cha*, Srinadh Bhojanapalli*, and Chulhee Yun

📰 Arxiv Preprint 🗓 October 22, 2024 🔗 [arxiv] [code]

Principal Component Analysis (PCA)

Fair Streaming Principal Component Analysis: Statistical and Algorithmic Viewpoint

👥 Junghyun Lee*, Hanseul Cho*, Se-Young Yun, and Chulhee Yun

📰 NeurIPS 2023 🗓 September 22, 2023 🔗 [paper] [arxiv] [code]

Reinforcement Learning

PLASTIC: Improving Input and Label Plasticity for Sample Efficient Reinforcement Learning

👥 Hojoon Lee*, Hanseul Cho*, Hyunseung Kim*, Daehoon Gwak, Joonkee Kim, Jaegul Choo, Se-Young Yun, and Chulhee Yun

📰 NeurIPS 2023 🗓 September 22, 2023 🔗 [paper] [arxiv] [code]

Reset Mechanism

PLASTIC: Improving Input and Label Plasticity for Sample Efficient Reinforcement Learning

👥 Hojoon Lee*, Hanseul Cho*, Hyunseung Kim*, Daehoon Gwak, Joonkee Kim, Jaegul Choo, Se-Young Yun, and Chulhee Yun

📰 NeurIPS 2023 🗓 September 22, 2023 🔗 [paper] [arxiv] [code]

SGDA

SGDA with Shuffling: Faster Convergence For Nonconvex-PŁ Minimax Optimization

👥 Hanseul Cho and Chulhee Yun

📰 ICLR 2023 🗓 February 02, 2023 🔗 [paper] [arxiv]

🎉 NAVER Outstanding Theory Paper Award at the 7th Joint Conference of Korea Artificial Intelligence Association (JKAIA 2022)

Scratchpad

Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count

👥 Hanseul Cho*, Jaeyoung Cha*, Srinadh Bhojanapalli*, and Chulhee Yun

📰 Arxiv Preprint 🗓 October 22, 2024 🔗 [arxiv] [code]

Sequential Learning

Convergence and Implicit Bias of Gradient Descent on Continual Linear Classification

👥 Hyunji Jung*, Hanseul Cho*, and Chulhee Yun

📰 Under Review 🗓 November 16, 2024

🎉 Best Paper Award at the 11th Joint Conference of Korean Artificial Intelligence Association (JKAIA 2024)

Sharpness-aware Minimization

PLASTIC: Improving Input and Label Plasticity for Sample Efficient Reinforcement Learning

👥 Hojoon Lee*, Hanseul Cho*, Hyunseung Kim*, Daehoon Gwak, Joonkee Kim, Jaegul Choo, Se-Young Yun, and Chulhee Yun

📰 NeurIPS 2023 🗓 September 22, 2023 🔗 [paper] [arxiv] [code]

Streaming

Fair Streaming Principal Component Analysis: Statistical and Algorithmic Viewpoint

👥 Junghyun Lee*, Hanseul Cho*, Se-Young Yun, and Chulhee Yun

📰 NeurIPS 2023 🗓 September 22, 2023 🔗 [paper] [arxiv] [code]

Transformers

Position Coupling: Improving Length Generalization of Arithmetic Transformers Using Task Structure

👥 Hanseul Cho*, Jaeyoung Cha*, Pranjal Awasthi, Srinadh Bhojanapalli, Anupam Gupta, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Long-Context Foundation Models (LCFM)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count

👥 Hanseul Cho*, Jaeyoung Cha*, Srinadh Bhojanapalli*, and Chulhee Yun

📰 Arxiv Preprint 🗓 October 22, 2024 🔗 [arxiv] [code]

Unsupervised Learnability

Fair Streaming Principal Component Analysis: Statistical and Algorithmic Viewpoint

👥 Junghyun Lee*, Hanseul Cho*, Se-Young Yun, and Chulhee Yun

📰 NeurIPS 2023 🗓 September 22, 2023 🔗 [paper] [arxiv] [code]

Warm-Starting

DASH: Warm-Starting Neural Network Training in Stationary Settings without Loss of Plasticity

👥 Baekrok Shin*, Junsoo Oh*, Hanseul Cho, and Chulhee Yun

📰 NeurIPS 2024 (Short version at ICML 2024 Workshop on Advancing Neural Network Training (WANT)) 🗓 September 26, 2024 🔗 [paper] [arxiv] [code]

minimax optimization

SGDA with Shuffling: Faster Convergence For Nonconvex-PŁ Minimax Optimization

👥 Hanseul Cho and Chulhee Yun

📰 ICLR 2023 🗓 February 02, 2023 🔗 [paper] [arxiv]

🎉 NAVER Outstanding Theory Paper Award at the 7th Joint Conference of Korea Artificial Intelligence Association (JKAIA 2022)

Fundamental Benefit of Alternating Updates in Minimax Optimization

👥 Jaewook Lee*, Hanseul Cho*, and Chulhee Yun

📰 ICML 2024 (Short version at ICLR 2024 Workshop on Bridging the Gap Between Practice and Theory in Deep Learning (BGPT)) 🗓 May 02, 2024 🔗 [paper] [arxiv] [code]

🎉 Spotlight @ ICML 2024 (Top 3.5% among total submissions)

shuffling-based

SGDA with Shuffling: Faster Convergence For Nonconvex-PŁ Minimax Optimization

👥 Hanseul Cho and Chulhee Yun

📰 ICLR 2023 🗓 February 02, 2023 🔗 [paper] [arxiv]

🎉 NAVER Outstanding Theory Paper Award at the 7th Joint Conference of Korea Artificial Intelligence Association (JKAIA 2022)

without-replacement sampling

SGDA with Shuffling: Faster Convergence For Nonconvex-PŁ Minimax Optimization

👥 Hanseul Cho and Chulhee Yun

📰 ICLR 2023 🗓 February 02, 2023 🔗 [paper] [arxiv]

🎉 NAVER Outstanding Theory Paper Award at the 7th Joint Conference of Korea Artificial Intelligence Association (JKAIA 2022)