14 papers
ReState: Noise-Level-Aware State Correction for Training-Free Object Removal in Few-Step Diffusion Models
Donghwan Lee, Sanghyun Jo, Sihyun Kim +2
Training-free object removal commonly redirects self-attention away from the target region, but on few-step distilled diffusion models this often leaves object-shaped residuals. We…
Value-Informed Schedule Trimming for Accelerated Sequential Monte Carlo Guidance in Discrete Diffusion
Jaehyeon Kim, Ziseok Lee, Siyoung Choi +1
Reward-guided discrete diffusion enables test-time alignment with downstream objectives, but applying Sequential Monte Carlo (SMC) guidance at every denoising step is computational…
Risk-Calibrated Proposal Transport for Finite-Particle Diffusion Steering
Ziseok Lee, Jaehyeon Kim, Seungwon Kim +6
Inference-time steering combines pretrained diffusion experts or rewards without retraining by changing the dynamics that transport noise to data. Feynman-Kac correction compensate…
SurfSpec: Enhancing Off-Target-Agnostic Specificity by Bounding Pocket-Ligand Geometric Mismatch
Minyeong Hwang, Yoorim Gang, Ziseok Lee +5
Lead optimization in structure-based drug design aims to improve target binding while avoiding unintended interactions with off-target pockets. However, existing affinity-driven me…
PatchGate: Narrowing the Verbalization Gap with Intrinsic Object Inventories in Frozen Vision-Language Models
Jihyung Ko, Eunji Jung, Hyeongsub Kim +4
Reliable image captioning in Vision-Language Models (VLMs) requires captions to be both precise and complete, avoiding unsupported object mentions while covering visible objects. E…
On the Collapse of Generative Paths: A Criterion and Correction for Diffusion Steering
Ziseok Lee, Minyeong Hwang, Wooyeol Lee +6
Inference-time steering adapts pretrained diffusion and flow models to new tasks without retraining, often utilizing ratio-of-densities constructions that reweight time-indexed mar…
ISAC: Training-Free Instance-to-Semantic Attention Control for Multi-Instance Generation
Sanghyun Jo, Wooyeol Lee, Ziseok Lee +3
Recent open-weight text-to-image (T2I) diffusion models still struggle with multi-instance prompts, often omitting or merging instances and mixing semantics among similar objects.…
Early Timestep Zero-Shot Candidate Selection for Instruction-Guided Image Editing
Joowon Kim, Ziseok Lee, Donghyeon Cho +4
Despite recent advances in diffusion models, achieving reliable image generation and editing remains challenging due to the inherent diversity induced by stochastic noise in the sa…
COIN: Confidence Score-Guided Distillation for Annotation-Free Cell Segmentation
Sanghyun Jo, Seo Jin Lee, Seungwoo Lee +3
Cell instance segmentation (CIS) is crucial for identifying individual cell morphologies in histopathological images, providing valuable insights for biological and medical researc…
TRACE: Your Diffusion Model is Secretly an Instance Edge Detector
Sanghyun Jo, Ziseok Lee, Wooyeol Lee +3
High-quality instance and panoptic segmentation has traditionally relied on dense instance-level annotations such as masks, boxes, or points, which are costly, inconsistent, and di…
HybridLinker: Topology-Guided Posterior Sampling for Enhanced Diversity and Validity in 3D Molecular Linker Generation
Minyeong Hwang, Ziseok Lee, Kwang-Soo Kim +2
Linker generation is critical in drug discovery applications such as lead optimization and PROTAC design, where molecular fragments are assembled into diverse drug candidates via m…
Token-Supervised Value Models for Enhancing Mathematical Problem-Solving Capabilities of Large Language Models
Jung Hyun Lee, June Yong Yang, Byeongho Heo +4
With the rapid advancement of test-time compute search strategies to improve the mathematical problem-solving capabilities of large language models (LLMs), the need for building ro…
TTD: Text-Tag Self-Distillation Enhancing Image-Text Alignment in CLIP to Alleviate Single Tag Bias
Sanghyun Jo, Soohyun Ryu, Sungyub Kim +2
We identify a critical bias in contemporary CLIP-based models, which we denote as single tag bias. This bias manifests as a disproportionate focus on a singular tag (word) while ne…
Data-Efficient Unsupervised Interpolation Without Any Intermediate Frame for 4D Medical Images
JungEun Kim, Hangyul Yoon, Geondo Park +2
4D medical images, which represent 3D images with temporal information, are crucial in clinical practice for capturing dynamic changes and monitoring long-term disease progression.…