Papers by Dohyun Lee
Protecting Privacy Through Approximating Optimal Parameters for Sequence Unlearning in Language Models (2024.findings-acl)
Copied to clipboard
| Challenge: | Language models (LMs) demonstrate exceptional capabilities on tasks, but are vulnerable to extraction attacks. |
| Approach: | They propose Privacy Protection via Optimal Parameters (POP) which induces the model to forget about some of its training data. |
| Outcome: | The proposed method outperforms the state-of-the-art in retaining LM performance on 9 classification and 4 dialogue benchmarks. |
Exploring In-context Example Generation for Machine Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have demonstrated strong performance across various tasks with just a few examples. |
| Approach: | They propose a method that generates in-context example pairs without external resources. |
| Outcome: | The proposed method builds upon two prior criteria, relevance and diversity, which have been highlighted as key factors for in-context example selection. |
Opt-Out: Investigating Entity-Level Unlearning for Large Language Models via Optimal Transport (2025.acl-long)
Copied to clipboard
| Challenge: | Instruction-following large language models (LLMs) inadvertently disclose private, sensitive information to their users, underscoring the need for machine unlearning techniques to remove selective information from the models. |
| Approach: | They propose an optimal transport-based unlearning method that utilizes the Wasserstein distance from the model’s initial parameters to achieve more effective and fine-grained unlearning. |
| Outcome: | The proposed method surpasses existing methods and establishes a new standard for secure and adaptable LLMs that can accommodate user data removal requests without the need for full retraining. |
Revisiting LLMs as Zero-Shot Time Series Forecasters: Small Noise Can Break Large Models (2025.acl-short)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown remarkable performance across diverse tasks without domain-specific training, fueling interest in their potential for time series forecasting. |
| Approach: | They evaluate the effectiveness of LLMs as zero-shot forecasters compared to state-of-the-art domain-specific models by encoding sequences directly within prompts. |
| Outcome: | The proposed models perform well across multiple domains while reducing the need for domain-specific training. |
Handling Out-Of-Vocabulary Problem in Hangeul Word Embeddings (2021.eacl-main)
Copied to clipboard
| Challenge: | Word embedding is considered an essential factor in improving the performance of various Natural Language Processing (NLP) models. |
| Approach: | They propose a Hangeul word embedding model that infers original word embeds from typos while maintaining high performance. |
| Outcome: | The proposed model performs well against typos while maintaining high performance. |