Kyungsu Kim

Sanghyun Jo, Seo Jin Lee, Seohyung Hong et al.

Cell instance segmentation models trained on cell-specific datasets suffer severe performance drops on out-of-distribution cell types, while interactive foundation models overcome this through per-instance prompting at a cost that is prohibitively expensive for histopathology images containing hundreds to thousands of densely packed instances. We introduce Group Prompting, a new paradigm that shifts interactive segmentation from per-instance $O(N)$ to per-type $O(T)$, where a single click per cell type suffices to segment all instances of that type. Our key observation is that the frozen image encoder of the Segment Anything Model (SAM) already clusters same-type cells in its feature space before any prompt is given. Exploiting this property, we propose Chain-of-Prompts (CoP), a training-free framework that recursively expands a single user click by (1) identifying reliable same-type locations through non-parametric gating of multi-scale encoder features, and (2) selecting the most spatially distant reliable point as the next prompt to maximize coverage. On three cell-type-annotated benchmarks, CoP with one click per type retains over 90% of per-instance performance and surpasses fully-supervised methods without any additional training. On four morphologically homogeneous benchmarks, a single click retains over 99%. Project Page: https://shjo-april.github.io/Chain-of-Prompts/

Kyungsu Kim, Minju Park, Haesun Joung et al.

As digital music production has become mainstream, the selection of appropriate virtual instruments plays a crucial role in determining the quality of music. To search the musical instrument samples or virtual instruments that make one's desired sound, music producers use their ears to listen and compare each instrument sample in their collection, which is time-consuming and inefficient. In this paper, we call this task as Musical Instrument Retrieval and propose a method for retrieving desired musical instruments using reference music mixture as a query. The proposed model consists of the Single-Instrument Encoder and the Multi-Instrument Encoder, both based on convolutional neural networks. The Single-Instrument Encoder is trained to classify the instruments used in single-track audio, and we take its penultimate layer's activation as the instrument embedding. The Multi-Instrument Encoder is trained to estimate multiple instrument embeddings using the instrument embeddings computed by the Single-Instrument Encoder as a set of target embeddings. For more generalized training and realistic evaluation, we also propose a new dataset called Nlakh. Experimental results showed that the Single-Instrument Encoder was able to learn the mapping from the audio signal of unseen instruments to the instrument embedding space and the Multi-Instrument Encoder was able to extract multiple embeddings from the mixture of music and retrieve the desired instruments successfully. The code used for the experiment and audio samples are available at: https://github.com/minju0821/musical_instrument_retrieval

Sanghyun Jo, In-Jae Yu, Kyungsu Kim

Although weakly supervised semantic segmentation using only image-level labels (WSSS-IL) is potentially useful, its low performance and implementation complexity still limit its application. The main causes are (a) non-detection and (b) false-detection phenomena: (a) The class activation maps refined from existing WSSS-IL methods still only represent partial regions for large-scale objects, and (b) for small-scale objects, over-activation causes them to deviate from the object edges. We propose RecurSeed, which alternately reduces non- and false detections through recursive iterations, thereby implicitly finding an optimal junction that minimizes both errors. We also propose a novel data augmentation (DA) approach called EdgePredictMix, which further expresses an object's edge by utilizing the probability difference information between adjacent pixels in combining the segmentation results, thereby compensating for the shortcomings when applying the existing DA methods to WSSS. We achieved new state-of-the-art performances on both the PASCAL VOC 2012 and MS COCO 2014 benchmarks (VOC val: 74.4%, COCO val: 46.4%). The code is available at https://github.com/shjo-april/RecurSeed_and_EdgePredictMix.

SungYub Kim, Sihwan Park, Kyungsu Kim et al.

Explaining generalizations and preventing over-confident predictions are central goals of studies on the loss landscape of neural networks. Flatness, defined as loss invariability on perturbations of a pre-trained solution, is widely accepted as a predictor of generalization in this context. However, the problem that flatness and generalization bounds can be changed arbitrarily according to the scale of a parameter was pointed out, and previous studies partially solved the problem with restrictions: Counter-intuitively, their generalization bounds were still variant for the function-preserving parameter scaling transformation or limited only to an impractical network structure. As a more fundamental solution, we propose new prior and posterior distributions invariant to scaling transformations by \textit{decomposing} the scale and connectivity of parameters, thereby allowing the resulting generalization bound to describe the generalizability of a broad class of networks with the more practical class of transformations such as weight decay with batch normalization. We also show that the above issue adversely affects the uncertainty calibration of Laplace approximation and propose a solution using our invariant posterior. We empirically demonstrate our posterior provides effective flatness and calibration measures with low complexity in such a practical parameter transformation case, supporting its practical effectiveness in line with our rationale.

Jung Hyun Lee, June Yong Yang, Byeongho Heo et al.

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 robust verifiers has become increasingly important. However, all these inference strategies rely on existing verifiers originally designed for Best-of-N search, which makes them sub-optimal for tree search techniques at test time. During tree search, existing verifiers can only offer indirect and implicit assessments of partial solutions or under-value prospective intermediate steps, thus resulting in the premature pruning of promising intermediate steps. To overcome these limitations, we propose token-supervised value models (TVMs) - a new class of verifiers that assign each token a probability that reflects the likelihood of reaching the correct final answer. This new token-level supervision enables TVMs to directly and explicitly evaluate partial solutions, effectively distinguishing between promising and incorrect intermediate steps during tree search at test time. Experimental results demonstrate that combining tree-search-based inference strategies with TVMs significantly improves the accuracy of LLMs in mathematical problem-solving tasks, surpassing the performance of existing verifiers.

Kyungsu Kim, Chae Yeon Lim, Joong Bo Shin et al.

The cone-beam computed tomography (CBCT) provides 3D volumetric imaging of a target with low radiation dose and cost compared with conventional computed tomography, and it is widely used in the detection of paranasal sinus disease. However, it lacks the sensitivity to detect soft tissue lesions owing to reconstruction constraints. Consequently, only physicians with expertise in CBCT reading can distinguish between inherent artifacts or noise and diseases, restricting the use of this imaging modality. The development of artificial intelligence (AI)-based computer-aided diagnosis methods for CBCT to overcome the shortage of experienced physicians has attracted substantial attention. However, advanced AI-based diagnosis addressing intrinsic noise in CBCT has not been devised, discouraging the practical use of AI solutions for CBCT. To address this issue, we propose an AI-based computer-aided diagnosis method using CBCT with a denoising module. This module is implemented before diagnosis to reconstruct the internal ground-truth full-dose scan corresponding to an input CBCT image and thereby improve the diagnostic performance. The external validation results for the unified diagnosis of sinus fungal ball, chronic rhinosinusitis, and normal cases show that the proposed method improves the micro-, macro-average AUC, and accuracy by 7.4, 5.6, and 9.6% (from 86.2, 87.0, and 73.4 to 93.6, 92.6, and 83.0%), respectively, compared with a baseline while improving human diagnosis accuracy by 11% (from 71.7 to 83.0%), demonstrating technical differentiation and clinical effectiveness. This pioneering study on AI-based diagnosis using CBCT indicates denoising can improve diagnostic performance and reader interpretability in images from the sinonasal area, thereby providing a new approach and direction to radiographic image reconstruction regarding the development of AI-based diagnostic solutions.

Sanghyun Jo, In-Jae Yu, Kyungsu Kim

Weakly-supervised semantic segmentation aims to reduce labeling costs by training semantic segmentation models using weak supervision, such as image-level class labels. However, most approaches struggle to produce accurate localization maps and suffer from false predictions in class-related backgrounds (i.e., biased objects), such as detecting a railroad with the train class. Recent methods that remove biased objects require additional supervision for manually identifying biased objects for each problematic class and collecting their datasets by reviewing predictions, limiting their applicability to the real-world dataset with multiple labels and complex relationships for biasing. Following the first observation that biased features can be separated and eliminated by matching biased objects with backgrounds in the same dataset, we propose a fully-automatic/model-agnostic biased removal framework called MARS (Model-Agnostic biased object Removal without additional Supervision), which utilizes semantically consistent features of an unsupervised technique to eliminate biased objects in pseudo labels. Surprisingly, we show that MARS achieves new state-of-the-art results on two popular benchmarks, PASCAL VOC 2012 (val: 77.7%, test: 77.2%) and MS COCO 2014 (val: 49.4%), by consistently improving the performance of various WSSS models by at least 30% without additional supervision.

Eungbeom Kim, Jinhee Kim, Yoori Oh et al.

In this paper, we aim to unveil the impact of data augmentation in audio-language multi-modal learning, which has not been explored despite its importance. We explore various augmentation methods at not only train-time but also test-time and find out that proper data augmentation can lead to substantial improvements. Specifically, applying our proposed audio-language paired augmentation PairMix, which is the first multi-modal audio-language augmentation method, outperforms the baselines for both automated audio captioning and audio-text retrieval tasks. To fully take advantage of data augmentation, we also present multi-level test-time augmentation (Multi-TTA) for the test-time. We successfully incorporate the two proposed methods and uni-modal augmentations and achieve 47.5 SPIDEr on audio captioning, which is an 18.2% relative increase over the baseline. In audio-text retrieval, the proposed methods also show an improvement in performance as well.

Kyungsu Kim, Seong Je Oh, Chae Yeon Lim et al.

Image translation based on a generative adversarial network (GAN-IT) is a promising method for the precise localization of abnormal regions in chest X-ray images (AL-CXR) even without the pixel-level annotation. However, heterogeneous unpaired datasets undermine existing methods to extract key features and distinguish normal from abnormal cases, resulting in inaccurate and unstable AL-CXR. To address this problem, we propose an improved two-stage GAN-IT involving registration and data augmentation. For the first stage, we introduce an advanced deep-learning-based registration technique that virtually and reasonably converts unpaired data into paired data for learning registration maps, by sequentially utilizing linear-based global and uniform coordinate transformation and AI-based non-linear coordinate fine-tuning. This approach enables independent and complex coordinate transformation of each detailed location of the lung while recognizing the entire lung structure, thereby achieving higher registration performance with resolving inherent artifacts caused by unpaired conditions. For the second stage, we apply data augmentation to diversify anomaly locations by swapping the left and right lung regions on the uniform registered frames, further improving the performance by alleviating imbalance in data distribution showing left and right lung lesions. The proposed method is model agnostic and shows consistent AL-CXR performance improvement in representative AI models. Therefore, we believe GAN-IT for AL-CXR can be clinically implemented by using our basis framework, even if learning data are scarce or difficult for the pixel-level disease annotation.

JungEun Kim, Hangyul Yoon, Geondo Park et al.

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. However, acquiring 4D medical images poses challenges due to factors such as radiation exposure and imaging duration, necessitating a balance between achieving high temporal resolution and minimizing adverse effects. Given these circumstances, not only is data acquisition challenging, but increasing the frame rate for each dataset also proves difficult. To address this challenge, this paper proposes a simple yet effective Unsupervised Volumetric Interpolation framework, UVI-Net. This framework facilitates temporal interpolation without the need for any intermediate frames, distinguishing it from the majority of other existing unsupervised methods. Experiments on benchmark datasets demonstrate significant improvements across diverse evaluation metrics compared to unsupervised and supervised baselines. Remarkably, our approach achieves this superior performance even when trained with a dataset as small as one, highlighting its exceptional robustness and efficiency in scenarios with sparse supervision. This positions UVI-Net as a compelling alternative for 4D medical imaging, particularly in settings where data availability is limited. The source code is available at https://github.com/jungeun122333/UVI-Net.

Sanghyun Jo, Fei Pan, In-Jae Yu et al.

Weakly-supervised semantic segmentation (WSS) ensures high-quality segmentation with limited data and excels when employed as input seed masks for large-scale vision models such as Segment Anything. However, WSS faces challenges related to minor classes since those are overlooked in images with adjacent multiple classes, a limitation originating from the overfitting of traditional expansion methods like Random Walk. We first address this by employing unsupervised and weakly-supervised feature maps instead of conventional methodologies, allowing for hierarchical mask enhancement. This method distinctly categorizes higher-level classes and subsequently separates their associated lower-level classes, ensuring all classes are correctly restored in the mask without losing minor ones. Our approach, validated through extensive experimentation, significantly improves WSS across five benchmarks (VOC: 79.8\%, COCO: 53.9\%, Context: 49.0\%, ADE: 32.9\%, Stuff: 37.4\%), reducing the gap with fully supervised methods by over 84\% on the VOC validation set. Code is available at https://github.com/shjo-april/DHR.

Sanghyun Jo, Soohyun Ryu, Sungyub Kim et al.

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 neglecting other pertinent tags, stemming from CLIP's text embeddings that prioritize one specific tag in image-text relationships. When deconstructing text into individual tags, only one tag tends to have high relevancy with CLIP's image embedding, leading to biased tag relevancy. In this paper, we introduce a novel two-step fine-tuning approach, Text-Tag Self-Distillation (TTD), to address this challenge. TTD first extracts image-relevant tags from text based on their similarity to the nearest pixels then employs a self-distillation strategy to align combined masks with the text-derived mask. This approach ensures the unbiased image-text alignment of the CLIP-based models using only image-text pairs without necessitating additional supervision. Our technique demonstrates model-agnostic improvements in multi-tag classification and segmentation tasks, surpassing competing methods that rely on external resources. The code is available at https://github.com/shjo-april/TTD.

Sanghyun Jo, Ziseok Lee, Wooyeol Lee et al.

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 difficult to scale. Unsupervised and weakly-supervised approaches reduce this burden but remain constrained by semantic backbone constraints and human bias, often producing merged or fragmented outputs. We present TRACE (TRAnsforming diffusion Cues to instance Edges), showing that text-to-image diffusion models secretly function as instance edge annotators. TRACE identifies the Instance Emergence Point (IEP) where object boundaries first appear in self-attention maps, extracts boundaries through Attention Boundary Divergence (ABDiv), and distills them into a lightweight one-step edge decoder. This design removes the need for per-image diffusion inversion, achieving 81x faster inference while producing sharper and more connected boundaries. On the COCO benchmark, TRACE improves unsupervised instance segmentation by +5.1 AP, and in tag-supervised panoptic segmentation it outperforms point-supervised baselines by +1.7 PQ without using any instance-level labels. These results reveal that diffusion models encode hidden instance boundary priors, and that decoding these signals offers a practical and scalable alternative to costly manual annotation. Code is available at https://github.com/shjo-april/DiffEGG.

Kyungsu Kim, Junghyun Koo, Sungho Lee et al.

Recent advancements in neural audio codecs have enabled the use of tokenized audio representations in various audio generation tasks, such as text-to-speech, text-to-audio, and text-to-music generation. Leveraging this approach, we propose TokenSynth, a novel neural synthesizer that utilizes a decoder-only transformer to generate desired audio tokens from MIDI tokens and CLAP (Contrastive Language-Audio Pretraining) embedding, which has timbre-related information. Our model is capable of performing instrument cloning, text-to-instrument synthesis, and text-guided timbre manipulation without any fine-tuning. This flexibility enables diverse sound design and intuitive timbre control. We evaluated the quality of the synthesized audio, the timbral similarity between synthesized and target audio/text, and synthesis accuracy (i.e., how accurately it follows the input MIDI) using objective measures. TokenSynth demonstrates the potential of leveraging advanced neural audio codecs and transformers to create powerful and versatile neural synthesizers. The source code, model weights, and audio demos are available at: https://github.com/KyungsuKim42/tokensynth

Sanghyun Jo, Donghwan Lee, Eunji Jung et al.

Object removal differs from common inpainting, since it must prevent the masked target from reappearing and reconstruct the occluded background with structural and contextual fidelity, rather than merely filling a hole plausibly. Recent dataset-free approaches that redirect self-attention inside the mask fail in two ways: non-target foregrounds are often misinterpreted as background, which regenerates unwanted objects, and direct attention manipulation disrupts fine details and hinders coherent integration of background cues. We propose EraseLoRA, a novel dataset-free framework that replaces attention surgery with background-aware reasoning and test-time adaptation. First, Background-aware Foreground Exclusion (BFE), uses a multimodal large-language models to separate target foreground, non-target foregrounds, and clean background from a single image-mask pair without paired supervision, producing reliable background cues while excluding distractors. Second, Background-aware Reconstruction with Subtype Aggregation (BRSA), performs test-time optimization that treats inferred background subtypes as complementary pieces and enforces their consistent integration through reconstruction and alignment objectives, preserving local detail and global structure without explicit attention intervention. We validate EraseLoRA as a plug-in to pretrained diffusion models and across benchmarks for object removal, demonstrating consistent improvements over dataset-free baselines and competitive results against dataset-driven methods. The code will be made available upon publication.

Sanghyun Jo, Seo Jin Lee, Seungwoo Lee et al.

Cell instance segmentation (CIS) is crucial for identifying individual cell morphologies in histopathological images, providing valuable insights for biological and medical research. While unsupervised CIS (UCIS) models aim to reduce the heavy reliance on labor-intensive image annotations, they fail to accurately capture cell boundaries, causing missed detections and poor performance. Recognizing the absence of error-free instances as a key limitation, we present COIN (COnfidence score-guided INstance distillation), a novel annotation-free framework with three key steps: (1) Increasing the sensitivity for the presence of error-free instances via unsupervised semantic segmentation with optimal transport, leveraging its ability to discriminate spatially minor instances, (2) Instance-level confidence scoring to measure the consistency between model prediction and refined mask and identify highly confident instances, offering an alternative to ground truth annotations, and (3) Progressive expansion of confidence with recursive self-distillation. Extensive experiments across six datasets show COIN outperforming existing UCIS methods, even surpassing semi- and weakly-supervised approaches across all metrics on the MoNuSeg and TNBC datasets. The code is available at https://github.com/shjo-april/COIN.

Ziseok Lee, Minyeong Hwang, Sanghyun Jo et al.

Inference-time steering enables pretrained diffusion/flow models to be adapted to new tasks without retraining. A widely used approach is the ratio-of-densities method, which defines a time-indexed target path by reweighting probability-density trajectories from multiple models with positive, or in some cases, negative exponents. This construction, however, harbors a critical and previously unformalized failure mode: Marginal Path Collapse, where intermediate densities become non-normalizable even though endpoints remain valid. Collapse arises systematically when composing heterogeneous models trained on different noise schedules or datasets, including a common setting in molecular design where de-novo, conformer, and pocket-conditioned models must be combined for tasks such as flexible-pose scaffold decoration. We provide a novel and complete solution for the problem. First, we derive a simple path existence criterion that predicts exactly when collapse occurs from noise schedules and exponents alone. Second, we introduce Adaptive path Correction with Exponents (ACE), which extends Feynman-Kac steering to time-varying exponents and guarantees a valid probability path. On a synthetic 2D benchmark and on flexible-pose scaffold decoration, ACE eliminates collapse and enables high-guidance compositional generation, improving distributional and docking metrics over constant-exponent baselines and even specialized task-specific scaffold decoration models. Our work turns ratio-of-densities steering with heterogeneous experts from an unstable heuristic into a reliable tool for controllable generation.

Minyeong Hwang, Ziseok Lee, Kwang-Soo Kim et al.

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 molecular linker. Existing methods fall into point cloud-free and point cloud-aware categories based on their use of fragments' 3D poses alongside their topologies in sampling the linker's topology. Point cloud-free models prioritize sample diversity but suffer from lower validity due to overlooking fragments' spatial constraints, while point cloud-aware models ensure higher validity but restrict diversity by enforcing strict spatial constraints. To overcome these trade-offs without additional training, we propose HybridLinker, a framework that enhances point cloud-aware inference by providing diverse bonding topologies from a pretrained point cloud-free model as guidance. At its core, we propose LinkerDPS, the first diffusion posterior sampling (DPS) method operating across point cloud-free and point cloud-aware spaces, bridging molecular topology with 3D point clouds via an energy-inspired function. By transferring the diverse sampling distribution of point cloud-free models into the point cloud-aware distribution, HybridLinker significantly surpasses baselines, improving both validity and diversity in foundational molecular design and applied drug optimization tasks, establishing a new DPS framework in the molecular domains beyond imaging.

Kyungsu Kim

This thesis studies the effectiveness of Long Short Term Memory model in forecasting future Job Openings and Labor Turnover Survey data in the United States. Drawing on multiple economic indicators from various sources, the data are fed directly into LSTM model to predict JOLT job openings in subsequent periods. The performance of the LSTM model is compared with conventional autoregressive approaches, including ARIMA, SARIMA, and Holt-Winters. Findings suggest that the LSTM model outperforms these traditional models in predicting JOLT job openings, as it not only captures the dependent variables trends but also harmonized with key economic factors. These results highlight the potential of deep learning techniques in capturing complex temporal dependencies in economic data, offering valuable insights for policymakers and stakeholders in developing data-driven labor market strategies

Kyungsu Kim, Junhyun Park, Saul Langarica et al.

This study demonstrates the first in-hospital adaptation of a cloud-based AI, similar to ChatGPT, into a secure model for analyzing radiology reports, prioritizing patient data privacy. By employing a unique sentence-level knowledge distillation method through contrastive learning, we achieve over 95% accuracy in detecting anomalies. The model also accurately flags uncertainties in its predictions, enhancing its reliability and interpretability for physicians with certainty indicators. These advancements represent significant progress in developing secure and efficient AI tools for healthcare, suggesting a promising future for in-hospital AI applications with minimal supervision.

Sanghyun Jo, Wooyeol Lee, Ziseok Lee et al.

Text-to-image diffusion models excel at generating single-instance scenes but struggle with multi-instance scenarios, often merging or omitting objects. Unlike previous training-free approaches that rely solely on semantic-level guidance without addressing instance individuation, our training-free method, Instance-to-Semantic Attention Control (ISAC), explicitly resolves incomplete instance formation and semantic entanglement through an instance-first modeling approach. This enables ISAC to effectively leverage a hierarchical, tree-structured prompt mechanism, disentangling multiple object instances and individually aligning them with their corresponding semantic labels. Without employing any external models, ISAC achieves up to 52% average multi-class accuracy and 83% average multi-instance accuracy by effectively forming disentangled instances. The code will be made available upon publication.

Kyungsu Kim

In this paper, I explored how a range of regression and machine learning techniques can be applied to monthly U.S. unemployment data to produce timely forecasts. I compared seven models: Linear Regression, SGDRegressor, Random Forest, XGBoost, CatBoost, Support Vector Regression, and an LSTM network, training each on a historical span of data and then evaluating on a later hold-out period. Input features include macro indicators (GDP growth, CPI), labor market measures (job openings, initial claims), financial variables (interest rates, equity indices), and consumer sentiment. I tuned model hyperparameters via cross-validation and assessed performance with standard error metrics and the ability to predict the correct unemployment direction. Across the board, tree-based ensembles (and CatBoost in particular) deliver noticeably better forecasts than simple linear approaches, while the LSTM captures underlying temporal patterns more effectively than other nonlinear methods. SVR and SGDRegressor yield modest gains over standard regression but don't match the consistency of the ensemble and deep-learning models. Interpretability tools ,feature importance rankings and SHAP values, point to job openings and consumer sentiment as the most influential predictors across all methods. By directly comparing linear, ensemble, and deep-learning approaches on the same dataset, our study shows how modern machine-learning techniques can enhance real-time unemployment forecasting, offering economists and policymakers richer insights into labor market trends. In the comparative evaluation of the models, I employed a dataset comprising thirty distinct features over the period from January 2020 through December 2024.

Joowon Kim, Ziseok Lee, Donghyeon Cho et al.

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 sampling process. Instruction-guided image editing with diffusion models offers user-friendly capabilities, yet editing failures, such as background distortion, frequently occur. Users often resort to trial and error, adjusting seeds or prompts to achieve satisfactory results, which is inefficient. While seed selection methods exist for Text-to-Image (T2I) generation, they depend on external verifiers, limiting applicability, and evaluating multiple seeds increases computational complexity. To address this, we first establish a multiple-seed-based image editing baseline using background consistency scores, achieving Best-of-N performance without supervision. Building on this, we introduce ELECT (Early-timestep Latent Evaluation for Candidate Selection), a zero-shot framework that selects reliable seeds by estimating background mismatches at early diffusion timesteps, identifying the seed that retains the background while modifying only the foreground. ELECT ranks seed candidates by a background inconsistency score, filtering unsuitable samples early based on background consistency while preserving editability. Beyond standalone seed selection, ELECT integrates into instruction-guided editing pipelines and extends to Multimodal Large-Language Models (MLLMs) for joint seed and prompt selection, further improving results when seed selection alone is insufficient. Experiments show that ELECT reduces computational costs (by 41 percent on average and up to 61 percent) while improving background consistency and instruction adherence, achieving around 40 percent success rates in previously failed cases - without any external supervision or training.

Junhyun Park, Chanyu Moon, Donghwan Lee et al.

Contrastive Language-Image Pre-Training (CLIP) has enabled zero-shot classification in radiology, reducing reliance on manual annotations. However, conventional contrastive learning struggles with normal case detection due to its strict intra-sample alignment, which disrupts normal sample clustering and leads to high false positives (FPs) and false negatives (FNs). To address these issues, we propose OFF-CLIP, a contrastive learning refinement that improves normal detection by introducing an off-diagonal term loss to enhance normal sample clustering and applying sentence-level text filtering to mitigate FNs by removing misaligned normal statements from abnormal reports. OFF-CLIP can be applied to radiology CLIP models without requiring any architectural modifications. Experimental results show that OFF-CLIP significantly improves normal classification, achieving a 0.61 Area under the curve (AUC) increase on VinDr-CXR over CARZero, the state-of-the-art zero-shot classification baseline, while maintaining or improving abnormal classification performance. Additionally, OFF-CLIP enhances zero-shot grounding by improving pointing game accuracy, confirming better anomaly localization. These results demonstrate OFF-CLIP's effectiveness as a robust and efficient enhancement for medical vision-language models.

Byungju Kim, Jaeyoung Lee, Kyungsu Kim et al.

Although well-trained deep neural networks have shown remarkable performance on numerous tasks, they rapidly forget what they have learned as soon as they begin to learn with additional data with the previous data stop being provided. In this paper, we introduce a novel algorithm, Incremental Class Learning with Attribute Sharing (ICLAS), for incremental class learning with deep neural networks. As one of its component, we also introduce a generative model, incGAN, which can generate images with increased variety compared with the training data. Under challenging environment of data deficiency, ICLAS incrementally trains classification and the generation networks. Since ICLAS trains both networks, our algorithm can perform multiple times of incremental class learning. The experiments on MNIST dataset demonstrate the advantages of our algorithm.

Junyeong Kim, Minuk Ma, Trung Pham et al.

This paper considers a network referred to as Modality Shifting Attention Network (MSAN) for Multimodal Video Question Answering (MVQA) task. MSAN decomposes the task into two sub-tasks: (1) localization of temporal moment relevant to the question, and (2) accurate prediction of the answer based on the localized moment. The modality required for temporal localization may be different from that for answer prediction, and this ability to shift modality is essential for performing the task. To this end, MSAN is based on (1) the moment proposal network (MPN) that attempts to locate the most appropriate temporal moment from each of the modalities, and also on (2) the heterogeneous reasoning network (HRN) that predicts the answer using an attention mechanism on both modalities. MSAN is able to place importance weight on the two modalities for each sub-task using a component referred to as Modality Importance Modulation (MIM). Experimental results show that MSAN outperforms previous state-of-the-art by achieving 71.13\% test accuracy on TVQA benchmark dataset. Extensive ablation studies and qualitative analysis are conducted to validate various components of the network.

Jinseok Kim, Kyungsu Kim, Jae-Joon Kim

For the gradient computation across the time domain in Spiking Neural Networks (SNNs) training, two different approaches have been independently studied. The first is to compute the gradients with respect to the change in spike activation (activation-based methods), and the second is to compute the gradients with respect to the change in spike timing (timing-based methods). In this work, we present a comparative study of the two methods and propose a new supervised learning method that combines them. The proposed method utilizes each individual spike more effectively by shifting spike timings as in the timing-based methods as well as generating and removing spikes as in the activation-based methods. Experimental results showed that the proposed method achieves higher performance in terms of both accuracy and efficiency than the previous approaches.

Hyungjun Kim, Kyungsu Kim, Jinseok Kim et al.

Binary Neural Networks (BNNs) have been garnering interest thanks to their compute cost reduction and memory savings. However, BNNs suffer from performance degradation mainly due to the gradient mismatch caused by binarizing activations. Previous works tried to address the gradient mismatch problem by reducing the discrepancy between activation functions used at forward pass and its differentiable approximation used at backward pass, which is an indirect measure. In this work, we use the gradient of smoothed loss function to better estimate the gradient mismatch in quantized neural network. Analysis using the gradient mismatch estimator indicates that using higher precision for activation is more effective than modifying the differentiable approximation of activation function. Based on the observation, we propose a new training scheme for binary activation networks called BinaryDuo in which two binary activations are coupled into a ternary activation during training. Experimental results show that BinaryDuo outperforms state-of-the-art BNNs on various benchmarks with the same amount of parameters and computing cost.

Yunjae Jung, Dahun Kim, Sanghyun Woo et al.

Visual storytelling is a task of creating a short story based on photo streams. Unlike existing visual captioning, storytelling aims to contain not only factual descriptions, but also human-like narration and semantics. However, the VIST dataset consists only of a small, fixed number of photos per story. Therefore, the main challenge of visual storytelling is to fill in the visual gap between photos with narrative and imaginative story. In this paper, we propose to explicitly learn to imagine a storyline that bridges the visual gap. During training, one or more photos is randomly omitted from the input stack, and we train the network to produce a full plausible story even with missing photo(s). Furthermore, we propose for visual storytelling a hide-and-tell model, which is designed to learn non-local relations across the photo streams and to refine and improve conventional RNN-based models. In experiments, we show that our scheme of hide-and-tell, and the network design are indeed effective at storytelling, and that our model outperforms previous state-of-the-art methods in automatic metrics. Finally, we qualitatively show the learned ability to interpolate storyline over visual gaps.

Junyeong Kim, Minuk Ma, Kyungsu Kim et al.

This paper proposes a method to gain extra supervision via multi-task learning for multi-modal video question answering. Multi-modal video question answering is an important task that aims at the joint understanding of vision and language. However, establishing large scale dataset for multi-modal video question answering is expensive and the existing benchmarks are relatively small to provide sufficient supervision. To overcome this challenge, this paper proposes a multi-task learning method which is composed of three main components: (1) multi-modal video question answering network that answers the question based on the both video and subtitle feature, (2) temporal retrieval network that predicts the time in the video clip where the question was generated from and (3) modality alignment network that solves metric learning problem to find correct association of video and subtitle modalities. By simultaneously solving related auxiliary tasks with hierarchically shared intermediate layers, the extra synergistic supervisions are provided. Motivated by curriculum learning, multi task ratio scheduling is proposed to learn easier task earlier to set inductive bias at the beginning of the training. The experiments on publicly available dataset TVQA shows state-of-the-art results, and ablation studies are conducted to prove the statistical validity.

Junyeong Kim, Minuk Ma, Kyungsu Kim et al.

This paper proposes the progressive attention memory network (PAMN) for movie story question answering (QA). Movie story QA is challenging compared to VQA in two aspects: (1) pinpointing the temporal parts relevant to answer the question is difficult as the movies are typically longer than an hour, (2) it has both video and subtitle where different questions require different modality to infer the answer. To overcome these challenges, PAMN involves three main features: (1) progressive attention mechanism that utilizes cues from both question and answer to progressively prune out irrelevant temporal parts in memory, (2) dynamic modality fusion that adaptively determines the contribution of each modality for answering the current question, and (3) belief correction answering scheme that successively corrects the prediction score on each candidate answer. Experiments on publicly available benchmark datasets, MovieQA and TVQA, demonstrate that each feature contributes to our movie story QA architecture, PAMN, and improves performance to achieve the state-of-the-art result. Qualitative analysis by visualizing the inference mechanism of PAMN is also provided.

Byungju Kim, Hyunwoo Kim, Kyungsu Kim et al.

We propose a novel regularization algorithm to train deep neural networks, in which data at training time is severely biased. Since a neural network efficiently learns data distribution, a network is likely to learn the bias information to categorize input data. It leads to poor performance at test time, if the bias is, in fact, irrelevant to the categorization. In this paper, we formulate a regularization loss based on mutual information between feature embedding and bias. Based on the idea of minimizing this mutual information, we propose an iterative algorithm to unlearn the bias information. We employ an additional network to predict the bias distribution and train the network adversarially against the feature embedding network. At the end of learning, the bias prediction network is not able to predict the bias not because it is poorly trained, but because the feature embedding network successfully unlearns the bias information. We also demonstrate quantitative and qualitative experimental results which show that our algorithm effectively removes the bias information from feature embedding.