SearcharxivSearch

arXiv subjects

Minkyu Lim

Publications and source records attributed to Minkyu Lim.

6 recordsLinked to original sources

Gradient Estimates Near the Natural Exponent for Very Weak Solutions to $A_p$-Weighted Quasilinear Elliptic Equations

We establish local Calder\'on--Zygmund estimates near the natural exponent for very weak solutions to matrix-weighted quasilinear equations\[ - \mathrm{div\,} A_{\mathbb{M}}(x,Du) = - \mathrm{div\,} A_{\mathbb{M}}(x,\mathbf{f}), \qquad A_{\mathbb{M}}(x,\xi)=\mathbb{M}(x)A(x,\mathbb{M}(x)\xi), \] where $A$ has $p$-growth and strong monotonicity, and $\mathbb{M}$ is a measurable positive-definite matrix field. We assume that $\mathbb{M}$ has bounded condition number and that \(\omega:=|\mathbb{M}|^p\in A_p\), without imposing uniform upper or lower bounds on $\mathbb{M}$. This extends the near-natural Calder\'on--Zygmund theory developed by Adimurthi--Phuc \cite{AP15} to the matrix-degenerate setting: There exists $\delta_0>0$ such that every very weak solution $u \in W^{p-\delta_{0}}_{\omega, \mathrm{loc}}$ satisfies \[ \mathbf{f}\in L^\gamma_{\omega,\mathrm{loc}} \Longrightarrow Du\in L^\gamma_{\omega,\mathrm{loc}} \] for $p-\delta_0\le\gamma\le p+\delta_0$. The proof requires handling the lack of energy estimates below the natural exponent and the use of Lipschitz truncation in the weighted setting. We achieve this through comparison estimates, higher integrability, and weighted analysis techniques.

math.AP

Existence of very weak solutions to nonlinear elliptic equation with nonstandard growth and global weighted gradient estimates

We study a general class of quasilinear elliptic equations with nonstandard growth to prove the existence of a very weak solution to such a problem. A key ingredient in the proof is a priori global weighted gradient estimate of a very weak solution, where the right hand side of the equation is the divergence of a vector-valued function with low degree of integrability. To obtain this estimate, we adopt a notion of reverse Hölder class of Muckenhoupt weights. Another crucial part of the proof is a generalized weighted div-curl lemma in the setting of Orlicz spaces.

math.AP

Improving Robustness of Neural Inverse Text Normalization via Data-Augmentation, Semi-Supervised Learning, and Post-Aligning Method

Inverse text normalization (ITN) is crucial for converting spoken-form into written-form, especially in the context of automatic speech recognition (ASR). While most downstream tasks of ASR rely on written-form, ASR systems often output spoken-form, highlighting the necessity for robust ITN in product-level ASR-based applications. Although neural ITN methods have shown promise, they still encounter performance challenges, particularly when dealing with ASR-generated spoken text. These challenges arise from the out-of-domain problem between training data and ASR-generated text. To address this, we propose a direct training approach that utilizes ASR-generated written or spoken text, with pairs augmented through ASR linguistic context emulation and a semi-supervised learning method enhanced by a large language model, respectively. Additionally, we introduce a post-aligning method to manage unpredictable errors, thereby enhancing the reliability of ITN. Our experiments show that our proposed methods remarkably improved ITN performance in various ASR scenarios.

cs.CL

Gradient estimates of very weak solutions to general quasilinear elliptic equations

We establish a gradient estimate for a very weak solution to a quasilinear elliptic equation with a nonstandard growth condition, which is a natural generalization of the $p$-Laplace equation. We investigate the maximum extent for the gradient estimate to hold without imposing any regularity assumption on the nonlinearity other than basic structure assumptions. Our results also include a higher integrability result of the gradient and the existence for the very weak solutions to such nonlinear problems.

math.AP

Integration of TensorFlow based Acoustic Model with Kaldi WFST Decoder

While the Kaldi framework provides state-of-the-art components for speech recognition like feature extraction, deep neural network (DNN)-based acoustic models, and a weighted finite state transducer (WFST)-based decoder, it is difficult to implement a new flexible DNN model. By contrast, a general-purpose deep learning framework, such as TensorFlow, can easily build various types of neural network architectures using a tensor-based computation method, but it is difficult to apply them to WFST-based speech recognition. In this study, a TensorFlow-based acoustic model is integrated with a WFST-based Kaldi decoder to combine the two frameworks. The features and alignments used in Kaldi are converted so they can be trained by the TensorFlow model, and the DNN-based acoustic model is then trained. In the integrated Kaldi decoder, the posterior probabilities are calculated by querying the trained TensorFlow model, and a beam search is performed to generate the lattice. The advantages of the proposed one-pass decoder include the application of various types of neural networks to WFST-based speech recognition and WFST-based online decoding using a TensorFlow-based acoustic model. The TensorFlow based acoustic models trained using the RM, WSJ, and LibriSpeech datasets show the same level of performance as the model trained using the Kaldi framework.

eess.AS

A Fast-Converged Acoustic Modeling for Korean Speech Recognition: A Preliminary Study on Time Delay Neural Network

In this paper, a time delay neural network (TDNN) based acoustic model is proposed to implement a fast-converged acoustic modeling for Korean speech recognition. The TDNN has an advantage in fast-convergence where the amount of training data is limited, due to subsampling which excludes duplicated weights. The TDNN showed an absolute improvement of 2.12% in terms of character error rate compared to feed forward neural network (FFNN) based modelling for Korean speech corpora. The proposed model converged 1.67 times faster than a FFNN-based model did.

cs.CL