arXiv subjects

Yong Suk Choi

Publications and source records attributed to Yong Suk Choi.

11 recordsLinked to original sources

arXiv2026

Two Clocks in Diffusion MLLMs: When Answers Stabilize Before Rationales Unfold

An answer candidate in a masked diffusion MLLM can stabilize while its rationale is still unfolding. We distinguish retrospective stabilization of the logged candidate from token commitment, and examine these two clocks relative to rationale generation. Analyzing our results across three visual question-answering benchmarks, we find that 89.4-98.1% of the rationale-side canvas remains unwritten at stabilization in single-block, EOS-suppressed LaViDa runs. On V*Bench, reducing block length from 128 to 8 changes this fraction from 89.4% to 1.7%, together with answer coverage and the eligible observation window. Under EOS-enabled prompting, direct instructions improve Nemotron's overall accuracy by 15.0 and 19.5 percentage points on M3CoT and ScienceQA, but reduce LaViDa/V*Bench accuracy by 11.0 points. A symmetric decomposition associates the larger absolute component of each change with coverage rather than conditional accuracy. Matched-canvas image ablations measure visual sensitivity alongside answer stabilization, separating the two temporal readouts. Together, these measurements distinguish answer stabilization, rationale unfolding, and visual sensitivity, and identify coverage as the larger component of the prompting differences.

cs.CV↗

arXiv2026

CWM: Controllable White-Box Meta-Prompting for Adaptive Retrieval-Augmented Generation and Reasoning Ability

Recently, Large Language Models (LLMs) have gained significant attention due to their strong language understanding and generation capabilities, demonstrating impressive reasoning abilities as well as effective utilization of external knowledge. Many studies have proposed methods that specialize in improving performance for individual tasks. However, ironically, only a limited number of attempts have explored general-purpose, task-agnostic methods. In this work, we present a unified framework integrating reasoning and Retrieval-Augmented Generation (RAG) tasks. We further propose Controllable White-Box Meta-Prompting (CWM), a low-cost white-box method for adaptive RAG tasks previously dominated by black-box approaches, without requiring external decision modules or multi-sampling. CWM achieves state-of-the-art performance on three adaptive RAG benchmarks across recent LLMs, including GPT-oss-20b, Qwen3-14b, and Llama3.1-8b, while also demonstrating strong generality by extending to reasoning tasks. In addition, CWM provides controllability by enabling retrieval decisions to be regulated through the manipulation of internal model signals. Our code is available at https://github.com/JeongEunhye00/CWM.

cs.AI↗

arXiv2026

ST-Veto: Spatio-Temporal Token Veto for Diffusion MLLMs via Taylor Prediction and Visual Grounding

Vision Language Models (VLMs) achieve strong reasoning with Chain-of-Thought (CoT) prompting but incur high sequential-generation cost, error accumulation, and limited self-correction. Diffusion Multimodal Large Language Models (dMLLMs) unmask tokens in an order-agnostic process, improving efficiency and enabling iterative refinement, yet their reasoning and how to enhance it remain underexplored. We propose a training-free method, Spatio-Temporal Token Veto (ST-Veto), which leverages the ability to observe all token positions at each diffusion step. Rather than relying only on current-step confidence, ST-Veto vetoes temporally unstable tokens via second-order Taylor prediction of confidence dynamics and filters weakly grounded tokens using image-attention mass, swapping them with safer candidates. Across multiple dMLLMs and multimodal reasoning benchmarks, ST-Veto consistently outperforms standard decoding policies and prior VLM reasoning methods, improving accuracy by up to 9% with no additional training or generation cost. Analyses show that ST-Veto steers generation toward higher-confidence, better-grounded paths.

cs.AI↗

arXiv2026

Thinking Diffusion: Penalize and Guide Visual-Grounded Reasoning in Diffusion Multimodal Language Models

Diffusion large language models (dLLMs) are emerging as promising alternatives to autoregressive (AR) LLMs. Recently, this paradigm has been extended to multimodal tasks, leading to the development of diffusion multimodal large language models (dMLLMs). These models are expected to retain the reasoning capabilities of LLMs while enabling faster inference through parallel generation. However, when combined with Chain-of-Thought (CoT) reasoning, dMLLMs exhibit two critical issues. First, we observe that dMLLMs often generate the final answer token at a very early timestep. This trend indicates that the model determines the answer before sufficient reasoning, leading to degraded reasoning performance. Second, during the initial timesteps, dMLLMs show minimal dependency on visual prompts, exhibiting a fundamentally different pattern of visual information utilization compared to AR vision-language models. In summary, these findings indicate that dMLLMs tend to generate premature final answers without sufficiently grounding on visual inputs. To address these limitations, we propose Position and Step Penalty (PSP) and Visual Reasoning Guidance (VRG). PSP penalizes tokens in later positions during early timesteps, delaying premature answer generation and encouraging progressive reasoning across timesteps. VRG, inspired by classifier-free guidance, amplifies visual grounding signals to enhance the model's alignment with visual evidence. Extensive experiments across various dMLLMs demonstrate that our method achieves up to 7.5% higher accuracy while delivering more than 3x speedup compared to reasoning with four times more diffusion steps.

cs.AI↗

arXiv2026

LogoDiffuser: Training-Free Multilingual Logo Generation and Stylization via Letter-Aware Attention Control

Recent advances in text-to-image generation have been remarkable, but generating multilingual design logos that harmoniously integrate visual and textual elements remains a challenging task. Existing methods often distort character geometry when applying creative styles and struggle to support multilingual text generation without additional training. To address these challenges, we propose LogoDiffuser, a training-free method that synthesizes multilingual logo designs using the multimodal diffusion transformer. Instead of using textual prompts, we input the target characters as images, enabling robust character structure control regardless of language. We first analyze the joint attention mechanism to identify core tokens, which are tokens that strongly respond to textual structures. With this observation, our method integrates character structure and visual design by injecting the most informative attention maps. Furthermore, we perform layer-wise aggregation of attention maps to mitigate attention shifts across layers and obtain consistent core tokens. Extensive experiments and user studies demonstrate that our method achieves state-of-the-art performance in multilingual logo generation.

cs.CV↗

arXiv2025

Editable Noise Map Inversion: Encoding Target-image into Noise For High-Fidelity Image Manipulation

Text-to-image diffusion models have achieved remarkable success in generating high-quality and diverse images. Building on these advancements, diffusion models have also demonstrated exceptional performance in text-guided image editing. A key strategy for effective image editing involves inverting the source image into editable noise maps associated with the target image. However, previous inversion methods face challenges in adhering closely to the target text prompt. The limitation arises because inverted noise maps, while enabling faithful reconstruction of the source image, restrict the flexibility needed for desired edits. To overcome this issue, we propose Editable Noise Map Inversion (ENM Inversion), a novel inversion technique that searches for optimal noise maps to ensure both content preservation and editability. We analyze the properties of noise maps for enhanced editability. Based on this analysis, our method introduces an editable noise refinement that aligns with the desired edits by minimizing the difference between the reconstructed and edited noise maps. Extensive experiments demonstrate that ENM Inversion outperforms existing approaches across a wide range of image editing tasks in both preservation and edit fidelity with target prompts. Our approach can also be easily applied to video editing, enabling temporal consistency and content manipulation across frames.

cs.CV↗

arXiv2025

CLAWS:Creativity detection for LLM-generated solutions using Attention Window of Sections

Recent advances in enhancing the reasoning ability of large language models (LLMs) have been remarkably successful. LLMs trained with reinforcement learning (RL) for reasoning demonstrate strong performance in challenging tasks such as mathematics and coding, even with relatively small model sizes. However, despite these improvements in task accuracy, the assessment of creativity in LLM generations has been largely overlooked in reasoning tasks, in contrast to writing tasks. The lack of research on creativity assessment in reasoning primarily stems from two challenges: (1) the difficulty of defining the range of creativity, and (2) the necessity of human evaluation in the assessment process. To address these challenges, we propose CLAWS, a method that defines and classifies mathematical solutions into typical, creative, and hallucinated categories without human evaluation, by leveraging attention weights across prompt sections and output. CLAWS outperforms five existing white-box detection methods (Perplexity, Logit Entropy, Window Entropy, Hidden Score, and Attention Score) on five 7-8B math RL models (DeepSeek, Qwen, Mathstral, OpenMath2, and Oreal). We validate CLAWS on 4545 math problems collected from 181 math contests (AJHSME, AMC, AIME).

cs.CL↗

arXiv2023

Balancing Lexical and Semantic Quality in Abstractive Summarization

An important problem of the sequence-to-sequence neural models widely used in abstractive summarization is exposure bias. To alleviate this problem, re-ranking systems have been applied in recent years. Despite some performance improvements, this approach remains underexplored. Previous works have mostly specified the rank through the ROUGE score and aligned candidate summaries, but there can be quite a large gap between the lexical overlap metric and semantic similarity. In this paper, we propose a novel training method in which a re-ranker balances the lexical and semantic quality. We further newly define false positives in ranking and present a strategy to reduce their influence. Experiments on the CNN/DailyMail and XSum datasets show that our method can estimate the meaning of summaries without seriously degrading the lexical aspect. More specifically, it achieves an 89.67 BERTScore on the CNN/DailyMail dataset, reaching new state-of-the-art performance. Our code is publicly available at https://github.com/jeewoo1025/BalSum.

cs.CL↗

arXiv2021

Graph Based Network with Contextualized Representations of Turns in Dialogue

Dialogue-based relation extraction (RE) aims to extract relation(s) between two arguments that appear in a dialogue. Because dialogues have the characteristics of high personal pronoun occurrences and low information density, and since most relational facts in dialogues are not supported by any single sentence, dialogue-based relation extraction requires a comprehensive understanding of dialogue. In this paper, we propose the TUrn COntext awaRE Graph Convolutional Network (TUCORE-GCN) modeled by paying attention to the way people understand dialogues. In addition, we propose a novel approach which treats the task of emotion recognition in conversations (ERC) as a dialogue-based RE. Experiments on a dialogue-based RE dataset and three ERC datasets demonstrate that our model is very effective in various dialogue-based natural language understanding tasks. In these experiments, TUCORE-GCN outperforms the state-of-the-art models on most of the benchmark datasets. Our code is available at https://github.com/BlackNoodle/TUCORE-GCN.

cs.CL↗

arXiv2019

A Tool for Spatio-Temporal Analysis of Social Anxiety with Twitter Data

In this paper, we present a tool for analyzing spatio-temporal distribution of social anxiety. Twitter, one of the most popular social network services, has been chosen as data source for analysis of social anxiety. Tweets (posted on the Twitter) contain various emotions and thus these individual emotions reflect social atmosphere and public opinion, which are often dependent on spatial and temporal factors. The reason why we choose anxiety among various emotions is that anxiety is very important emotion that is useful for observing and understanding social events of communities. We develop a machine learning based tool to analyze the changes of social atmosphere spatially and temporally. Our tool classifies whether each Tweet contains anxious content or not, and also estimates degree of Tweet anxiety. Furthermore, it also visualizes spatio-temporal distribution of anxiety as a form of web application, which is incorporated with physical map, word cloud, search engine and chart viewer. Our tool is applied to a big tweet data in South Korea to illustrate its usefulness for exploring social atmosphere and public opinion spatio-temporally.

cs.CL↗

arXiv2019

Semantic Relation Classification via Bidirectional LSTM Networks with Entity-aware Attention using Latent Entity Typing

Classifying semantic relations between entity pairs in sentences is an important task in Natural Language Processing (NLP). Most previous models for relation classification rely on the high-level lexical and syntactic features obtained by NLP tools such as WordNet, dependency parser, part-of-speech (POS) tagger, and named entity recognizers (NER). In addition, state-of-the-art neural models based on attention mechanisms do not fully utilize information of entity that may be the most crucial features for relation classification. To address these issues, we propose a novel end-to-end recurrent neural model which incorporates an entity-aware attention mechanism with a latent entity typing (LET) method. Our model not only utilizes entities and their latent types as features effectively but also is more interpretable by visualizing attention mechanisms applied to our model and results of LET. Experimental results on the SemEval-2010 Task 8, one of the most popular relation classification task, demonstrate that our model outperforms existing state-of-the-art models without any high-level features.

cs.CL↗