arXiv subjects

Yujin Jeong

Publications and source records attributed to Yujin Jeong.

13 recordsLinked to original sources

arXiv2026

Bridging the Gap Between Natural Language and Market Dynamics via High-Dimensional Representation Learning

Traditional multi-modal financial forecasting often relies on scalar sentiment scores, which fail to capture the nuances of financial news. To address this information loss, this paper explores high-dimensional representation learning by replacing discrete polarity ratings with dense FinBERT embeddings within a Transformer-based forecasting architecture. We benchmarked various embedding strategies on the FNSPID dataset, including raw embeddings, attention-weighted aggregation, and a custom Siamese network. While the attention-based mechanism struggled with the low signal-to-noise ratio typical of financial data, the integration of Siamese-optimized embeddings outperformed both the scalar baseline and raw embedding approaches, demonstrating that preserving high-dimensional narrative context yields improved predictive accuracy for short-term stock price movements.

cs.LG↗

arXiv2026

When Do Diffusion Models learn to Generate Multiple Objects?

Text-to-image diffusion models achieve impressive visual fidelity, yet they remain unreliable in multi-object generation. Despite extensive empirical evidence of these failures, the underlying causes remain unclear. We begin by asking how much of this limitation arises from the data itself. To disentangle data effects, we consider two regimes across different dataset sizes: (1) concept generalization, where each individual concept is observed during training under potentially imbalanced data distributions, and (2) compositional generalization, where specific combinations of concepts are systematically held out. To study these regimes, we introduce mosaic (Multi-Object Spatial relations, AttrIbution, Counting), a controlled framework for dataset generation. By training diffusion models on mosaic, we find that scene complexity plays a dominant role rather than concept imbalance, and that counting is uniquely difficult to learn in low-data regimes. Moreover, compositional generalization collapses as more concept combinations are held out during training. These findings highlight fundamental limitations of diffusion models and motivate stronger inductive biases and data design for robust multi-object compositional generation.

cs.CV↗

arXiv2026

Bayesian Nonparametric Modeling for Multivariate Conditional Copula Regression with Varying Coefficients

Multivariate mixed-type outcomes are difficult to model jointly, and additional complexity arises when both marginal effects and dependence structures vary with a covariate such as age or time. Existing approaches often impose restrictive dependence assumptions or lack sufficient flexibility to accommodate heterogeneous response types in a unified framework. To address this issue, we propose a Bayesian nonparametric framework for multivariate conditional copula regression with varying coefficients. The proposed model combines adaptive spline-based marginal regressions with an infinite mixture of Gaussian copulas whose weights vary with the covariate through a probit stick-breaking process. This construction provides flexible covariate-dependent dependence modeling while avoiding explicit global constraints on functional correlation matrices. We further establish approximation results for the proposed copula representation and develop a Markov chain Monte Carlo algorithm for posterior inference. Simulation studies show accurate recovery under correct specification and robust performance under copula misspecification. In an analysis of the BRFSS 2023 data, the proposed model reveals age-varying marginal effects and dependence patterns among multiple health outcomes, providing a coherent joint view of multimorbidity beyond separate marginal analyses.

stat.ME↗

arXiv2026

Tuning Just Enough: Lightweight Backdoor Attacks on Multi-Encoder Diffusion Models

As text-to-image diffusion models become increasingly deployed in real-world applications, concerns about backdoor attacks have gained significant attention. Prior work on text-based backdoor attacks has largely focused on diffusion models conditioned on a single lightweight text encoder. However, more recent diffusion models that incorporate multiple large-scale text encoders remain underexplored in this context. Given the substantially increased number of trainable parameters introduced by multiple text encoders, an important question is whether backdoor attacks can remain both efficient and effective in such settings. In this work, we study Stable Diffusion 3, which uses three distinct text encoders and has not yet been systematically analyzed for text-encoder-based backdoor vulnerabilities. To understand the role of text encoders in backdoor attacks, we define four categories of attack targets and identify the minimal sets of encoders required to achieve effective performance for each attack objective. Based on this, we further propose Multi-Encoder Lightweight aTtacks (MELT), which trains only low-rank adapters while keeping the pretrained text encoder weight frozen. We demonstrate that tuning fewer than 0.2% of the total encoder parameters is sufficient for successful backdoor attacks on Stable Diffusion 3, revealing previously underexplored vulnerabilities in practical attack scenarios in multi-encoder settings.

cs.LG↗

arXiv2025

Diffusion Classifiers Understand Compositionality, but Conditions Apply

Understanding visual scenes is fundamental to human intelligence. While discriminative models have significantly advanced computer vision, they often struggle with compositional understanding. In contrast, recent generative text-to-image diffusion models excel at synthesizing complex scenes, suggesting inherent compositional capabilities. Building on this, zero-shot diffusion classifiers have been proposed to repurpose diffusion models for discriminative tasks. While prior work offered promising results in discriminative compositional scenarios, these results remain preliminary due to a small number of benchmarks and a relatively shallow analysis of conditions under which the models succeed. To address this, we present a comprehensive study of the discriminative capabilities of diffusion classifiers on a wide range of compositional tasks. Specifically, our study covers three diffusion models (SD 1.5, 2.0, and, for the first time, 3-m) spanning 10 datasets and over 30 tasks. Further, we shed light on the role that target dataset domains play in respective performance; to isolate the domain effects, we introduce a new diagnostic benchmark \textsc{Self-Bench} comprised of images created by diffusion models themselves. Finally, we explore the importance of timestep weighting and uncover a relationship between domain gap and timestep sensitivity, particularly for SD3-m. To sum up, diffusion classifiers understand compositionality, but conditions apply! Code and dataset are available at https://github.com/eugene6923/Diffusion-Classifiers-Compositionality.

cs.CV↗

arXiv2025

Optimal Empirical Risk Minimization under Temporal Distribution Shifts

Temporal distribution shifts pose a key challenge for machine learning models trained and deployed in dynamically evolving environments. This paper introduces RIDER (RIsk minimization under Dynamically Evolving Regimes) which derives optimally-weighted empirical risk minimization procedures under temporal distribution shifts. Our approach is theoretically grounded in the random distribution shift model, where random shifts arise as a superposition of numerous unpredictable changes in the data-generating process. We show that common weighting schemes, such as pooling all data, exponentially weighting data, and using only the most recent data, emerge naturally as special cases in our framework. We demonstrate that RIDER consistently improves out-of-sample predictive performance when applied as a fine-tuning step on the Yearbook dataset, across a range of benchmark methods in Wild-Time. Moreover, we show that RIDER outperforms standard weighting strategies in two other real-world tasks: predicting stock market volatility and forecasting ride durations in NYC taxi data.

stat.ME↗

arXiv2025

Streaming Dense Voxel Representations for 3D Occupancy Prediction

In this paper, we explore dense voxel streaming for accurate and efficient 3D occupancy prediction. While dense voxel representations offer fine-grained spatial details and streaming paradigm enables efficient temporal processing, naively combining the two introduces key challenges: (i) warping-induced distortions caused by interpolation used for temporal alignment, and (ii) degraded dynamic object representations due to motion misalignment and detail loss in image-to-voxel projection. To address these, we propose StreamOcc, a novel framework that utilizes two aggregation strategies. Specifically, it first refines propagated voxel features to reduce warping artifacts before temporal accumulation, and then selectively injects instance-level query features encoding dynamic-object semantics into the corresponding occupied voxel regions, preserving temporally consistent modeling while strengthening dynamic object representations. Unlocking effective dense voxel streaming, StreamOcc achieves state-of-the-art performance on SurroundOcc-benchmark and Occ3D-nuScenes under real-time constraints, outperforming the prior best methods by +1.3/2.5 and +1.5/2.0 in (overall/dynamic object) mIoU, respectively, while running at 83.3 ms per frame with only 2.8 GB of memory. The project page is available at https://moonseokha.github.io/StreamOcc/.

cs.CV↗

arXiv2025

Out-of-distribution generalization under random, dense distributional shifts

Many existing approaches for estimating parameters in settings with distributional shifts operate under an invariance assumption. For example, under covariate shift, it is assumed that $p(y|x)$ remains invariant. We refer to such distribution shifts as sparse, since they may be substantial but affect only a part of the data generating system. In contrast, in various real-world settings, shifts might be dense. More specifically, these dense distributional shifts may arise through numerous small and random changes in the population and environment. First, we discuss empirical evidence for such random dense distributional shifts. Then, we develop tools to infer parameters and make predictions for partially observed, shifted distributions. Finally, we apply the framework to several real-world datasets and discuss diagnostics to evaluate the fit of the distributional uncertainty model.

stat.ME↗

arXiv2025

Calibrated inference: statistical inference that accounts for both sampling uncertainty and distributional uncertainty

How can we draw trustworthy scientific conclusions? One criterion is that a study can be replicated by independent teams. While replication is critically important, it is arguably insufficient. If a study is biased for some reason and other studies recapitulate the approach then findings might be consistently incorrect. It has been argued that trustworthy scientific conclusions require disparate sources of evidence. However, different methods might have shared biases, making it difficult to judge the trustworthiness of a result. We formalize this issue by introducing a "distributional uncertainty model", wherein dense distributional shifts emerge as the superposition of numerous small random changes. The distributional perturbation model arises under a symmetry assumption on distributional shifts and is strictly weaker than assuming that the data is i.i.d. from the target distribution. We show that a stability analysis on a single data set allows us to construct confidence intervals that account for both sampling uncertainty and distributional uncertainty.

stat.ME↗

arXiv2024

Read, Watch and Scream! Sound Generation from Text and Video

Despite the impressive progress of multimodal generative models, video-to-audio generation still suffers from limited performance and limits the flexibility to prioritize sound synthesis for specific objects within the scene. Conversely, text-to-audio generation methods generate high-quality audio but pose challenges in ensuring comprehensive scene depiction and time-varying control. To tackle these challenges, we propose a novel video-and-text-to-audio generation method, called \ours, where video serves as a conditional control for a text-to-audio generation model. Especially, our method estimates the structural information of sound (namely, energy) from the video while receiving key content cues from a user prompt. We employ a well-performing text-to-audio model to consolidate the video control, which is much more efficient for training multimodal diffusion models with massive triplet-paired (audio-video-text) data. In addition, by separating the generative components of audio, it becomes a more flexible system that allows users to freely adjust the energy, surrounding environment, and primary sound source according to their preferences. Experimental results demonstrate that our method shows superiority in terms of quality, controllability, and training efficiency. Code and demo are available at https://naver-ai.github.io/rewas.

cs.CV↗

arXiv2024

Identifying sparse treatment effects in high-dimensional outcome spaces

Based on technological advances in sensing modalities, randomized trials with primary outcomes represented as high-dimensional vectors have become increasingly prevalent. For example, these outcomes could be week-long time-series data from wearable devices or high-dimensional neuroimaging data, such as from functional magnetic resonance imaging. This paper focuses on randomized treatment studies with such high-dimensional outcomes characterized by sparse treatment effects, where interventions may influence a small number of dimensions, e.g., small temporal windows or specific brain regions. Conventional practices, such as using fixed, low-dimensional summaries of the outcomes, result in significantly reduced power for detecting treatment effects. To address this limitation, we propose a procedure that involves subset selection followed by inference. Specifically, given a potentially large set of outcome summaries, we identify the subset that captures treatment effects, which requires only one call to the Lasso, and subsequently conduct inference on the selected subset. Via theoretical analysis as well as simulations, we demonstrate that our method asymptotically selects the correct subset and increases statistical power.

stat.ME↗

arXiv2023

The Power of Sound (TPoS): Audio Reactive Video Generation with Stable Diffusion

In recent years, video generation has become a prominent generative tool and has drawn significant attention. However, there is little consideration in audio-to-video generation, though audio contains unique qualities like temporal semantics and magnitude. Hence, we propose The Power of Sound (TPoS) model to incorporate audio input that includes both changeable temporal semantics and magnitude. To generate video frames, TPoS utilizes a latent stable diffusion model with textual semantic information, which is then guided by the sequential audio embedding from our pretrained Audio Encoder. As a result, this method produces audio reactive video contents. We demonstrate the effectiveness of TPoS across various tasks and compare its results with current state-of-the-art techniques in the field of audio-to-video generation. More examples are available at https://ku-vai.github.io/TPoS/

cs.SD↗

arXiv2022

Zero-shot Visual Commonsense Immorality Prediction

Artificial intelligence is currently powering diverse real-world applications. These applications have shown promising performance, but raise complicated ethical issues, i.e. how to embed ethics to make AI applications behave morally. One way toward moral AI systems is by imitating human prosocial behavior and encouraging some form of good behavior in systems. However, learning such normative ethics (especially from images) is challenging mainly due to a lack of data and labeling complexity. Here, we propose a model that predicts visual commonsense immorality in a zero-shot manner. We train our model with an ETHICS dataset (a pair of text and morality annotation) via a CLIP-based image-text joint embedding. In a testing phase, the immorality of an unseen image is predicted. We evaluate our model with existing moral/immoral image datasets and show fair prediction performance consistent with human intuitions. Further, we create a visual commonsense immorality benchmark with more general and extensive immoral visual contents. Codes and dataset are available at https://github.com/ku-vai/Zero-shot-Visual-Commonsense-Immorality-Prediction. Note that this paper might contain images and descriptions that are offensive in nature.

cs.CV↗