arXiv subjects

Jaesung Bae

Publications and source records attributed to Jaesung Bae.

6 recordsLinked to original sources

arXiv2026

Something from Nothing: Data Augmentation for Robust Severity Level Estimation of Dysarthric Speech

Dysarthric speech quality assessment (DSQA) is critical for clinical diagnostics and inclusive speech technologies. However, subjective evaluation is costly and difficult to scale, and the scarcity of labeled data limits robust objective modeling. To address this, we propose a three-stage framework that leverages unlabeled dysarthric speech and large-scale typical speech datasets to scale training. A teacher model first generates pseudo-labels for unlabeled samples, followed by weakly supervised pretraining using a label-aware contrastive learning strategy that exposes the model to diverse speakers and acoustic conditions. The pretrained model is then fine-tuned for the downstream DSQA task. Experiments on five unseen datasets spanning multiple etiologies and languages demonstrate the robustness of our approach. Our Whisper-based baseline significantly outperforms SOTA DSQA predictors such as SpICE, and the full framework achieves an average SRCC of 0.761 across unseen test datasets.

eess.AS↗

arXiv2026

Semantics-Aware Generative Latent Data Augmentation for Learning in Low-Resource Domains

Despite strong performance in data-rich regimes, deep learning often underperforms in the data-scarce settings common in practice. While foundation models (FMs) trained on massive datasets demonstrate strong generalization by extracting general-purpose features, they can still suffer from scarce labeled data during downstream fine-tuning. To address this, we propose GeLDA, a semantics-aware generative latent data augmentation framework that leverages conditional diffusion models to synthesize samples in an FM-induced latent space. Because this space is low-dimensional and concentrates task-relevant information compared to the input space, GeLDA enables efficient, high-quality data generation. GeLDA conditions generation on auxiliary feature vectors that capture semantic relationships among classes or subdomains, facilitating data augmentation in low-resource domains. We validate GeLDA in two large-scale recognition tasks: (a) in zero-shot language-specific speech emotion recognition, GeLDA improves the Whisper-large baseline's unweighted average recall by 6.13%; and (b) in long-tailed image classification, it achieves 74.7% tail-class accuracy on ImageNet-LT, setting a new state-of-the-art result.

cs.LG↗

arXiv2025

That's Deprecated! Understanding, Detecting, and Steering Knowledge Conflicts in Language Models for Code Generation

This paper investigates how large language models (LLMs) behave when faced with discrepancies between their parametric knowledge and conflicting information contained in a prompt. Building on prior question-answering (QA) research, we extend the investigation of knowledge conflicts to the realm of code generation. We propose a domain-agnostic framework for constructing and interpreting such conflicts, along with a novel evaluation method and dataset tailored to code conflict scenarios. Our experiments indicate that sufficiently large LLMs encode the notion of a knowledge conflict in their parameters, enabling us to detect knowledge conflicts with up to \textbf{80.65\%} accuracy. Building on these insights, we show that activation-level steering can achieve up to a \textbf{12.6\%} improvement in steering success over a random baseline. However, effectiveness depends critically on balancing model size, task domain, and steering direction. The experiment code and data will be made publicly available after acceptance.

cs.CL↗

arXiv2025

Convex real projective structures on Coxeter orbifold $D^2(;n_1,n_2,n_3,n_4)\times \mathbb{R}$

The deformation space of real projective structures parametrizes the space of the convex real projective structures on an orbifold. The Coxeter orbifold can be obtained $D^2(;n_1,n_2,n_3,n_4)\times\mathbb{R}$ by embedding the Coxeter quadrilateral from $\mathbb{RP}^2$ to $\mathbb{RP}^3$, and extending the side edges to planes and perturbing these to obtain a convex polytope. This noncompact orbifold can be induced from two combinatorial polytopes. Using this fact, we determine the deformation space of real projective structures on this orbifold with parametrization, and see its very detailed properties.

math.GT↗

arXiv2020

Phase-Aware Speech Enhancement with a Recurrent Two Stage Net work

We propose a neural network-based speech enhancement (SE) method called the phase-aware recurrent two stage network (rTSN). The rTSN is an extension of our previously proposed two stage network (TSN) framework. This TSN framework was equipped with a boosting strategy (BS) that initially estimates the multiple base predictions (MBPs) from a prior neural network (pri-NN) and then the MBPs are aggregated by a posterior neural network (post-NN) to obtain the final prediction. The TSN outperformed various state-of-the-art methods; however, it adopted the simple deep neural network as pri-NN. We have found that the pri-NN affects the performance (in perceptual quality), more than post-NN; therefore we adopted the long short-term memory recurrent neural network (LSTM-RNN) as pri-NN to boost the context information usage within speech signals. Further, the TSN framework did not consider the phase reconstruction, though phase information affected the perceptual quality. Therefore, we proposed to adopt the phase reconstruction method based on the Griffin-Lim algorithm. Finally, we evaluated rTSN with baselines such as TSN in perceptual quality related metrics as well as the phone recognition error rate.

eess.AS↗

arXiv2019

End-Point Detection with State Transition Model based on Chunk-Wise Classification

A state transition model (STM) based on chunk-wise classification was proposed for end-point detection (EPD). In general, EPD is developed using frame-wise voice activity detection (VAD) with additional STM, in which the state transition is conducted based on VAD's frame-level decision (speech or non-speech). However, VAD errors frequently occur in noisy environments, even though we use state-of-the-art deep neural network based VAD, which causes the undesired state transition of STM. In this work, to build robust STM, a state transition is conducted based on chunk-wise classification as EPD does not need to be conducted in frame-level. The chunk consists of multiple frames and the classification of chunk between speech and non-speech is done by aggregating the decisions of VAD for multiple frames, so that some undesired VAD errors in a chunk can be smoothed by other correct VAD decisions. Finally, the model was evaluated in both qualitative and quantitative measures including phone error rate.

eess.AS↗