Searcharxiv⌕ Search

arXiv subjects

Yen-Cheng Chang

Publications and source records attributed to Yen-Cheng Chang.

6 recordsLinked to original sources
arXiv2026

Genuine Multi-Entropy in Abelian Chern--Simons Theory: Exact Key-Ring Collapse and Its Breakdown for Generic Link States

We study genuine multi-entropy in Abelian $U(1)_k$ Chern-Simons theory. For key-ring link states, where only the linking numbers between one distinguished component $K$ and the remaining $\mathtt{q}-1$ components are nonzero, we derive an exact closed-form expression for the $\mathtt{q}$-partite R\'enyi multi-entropy for general $\mathtt{q}$, level $k$, and R\'enyi index $n$. For $\mathtt{q}=4$, this shows that the genuine multi-entropy $\mathrm{GM}^{(4)}_n$ collapses exactly onto the tripartite information $I_{3,n}$ for all $n$, while for $\mathtt{q}=5$ it is likewise completely determined, for all $n$, by a linear combination of tripartite and bipartite R\'enyi multi-entropies. We then go beyond the key-ring class and study general four-component link states with arbitrary pairwise linking numbers. A numerical scan over Chern--Simons levels $2\leq k\leq24$ shows that the all-$n$ collapse found analytically for key-ring states does not survive for generic link states. Remarkably, the collapse remains exact at $n=2$ for every level examined. At $n=3$, violations occur, within the scanned range, only when $3\mid k$, with a further dependence on the $3$-adic valuation of $k$. At $n=4$ and $n=5$, violations occur for every level examined, with rates that vary strongly with $k$. These results show that the breakdown is not controlled simply by the zero-divisor structure of composite $\mathbb Z_k$, but instead exhibits a nontrivial joint dependence on the R\'enyi index and the arithmetic structure of the Chern--Simons level.

hep-th↗
arXiv2026

Efficient Streaming Algorithms for Two-Dimensional Congruence Testing and Congruence Hashing

Geometric congruence asks whether two point multisets are identical up to translation and rotation, while congruence hashing seeks compact encodings that support efficient congruence queries. We study these problems in the streaming model under finite-precision rational inputs, where each coordinate is $p/q$ with $|p|,|q|\le U$. Our main results are two randomized polylogarithmic-space algorithms for 2D congruence identification (CI), which additionally requires outputting a valid transformation when congruent. With probability at least $1-\delta$, our 3-pass product-anchor algorithm uses $O((\log n+\log U+\log \frac{1}{\delta})\log\log n\log \frac{1}{\delta})$ space in the turnstile model, while our 3-pass complex-moment algorithm uses $O(\log n(\log n+\log U+\log \frac{1}{\delta}))$ space in the insertion-only model. Using CI as a building block, we obtain a 4-pass insertion-only congruence hashing algorithm over $m$ query sets using $O(m(\log n+\log U+\log m+\log \frac{1}{\delta}))$ space and producing signatures of length $O(\log \frac{1}{\delta}+\log U+\log m)$. Both algorithms presample primes for modular hashing to handle precision. The former algorithm uses finite-field embeddings and number-theoretic guarantees to recover rotations, and the latter hinges on a new non-vanishing complex moment criterion, thus avoiding additional number-theoretic conditions and the classical vanishing-moment obstacle. For the hardness results, we prove that any $p$-pass randomized streaming algorithm for 2D CI with error at most $\delta$ requires $\Omega(\frac{1}{p}(\log n+\log U+\log \frac{1}{\delta}))$ space, matching the turnstile upper bound up to a $\log\log n$ factor. Moreover, approximate CI requires $\text{poly}(n)$ space even with $\text{poly}(n)$ passes. This sharply contrasts with the RAM model, where both exact and approximate versions are solvable in polynomial time.

cs.DS↗
arXiv2022

g2pW: A Conditional Weighted Softmax BERT for Polyphone Disambiguation in Mandarin

Polyphone disambiguation is the most crucial task in Mandarin grapheme-to-phoneme (g2p) conversion. Previous studies have approached this problem using pre-trained language models, restricted output, and extra information from Part-Of-Speech (POS) tagging. Inspired by these strategies, we propose a novel approach, called g2pW, which adapts learnable softmax-weights to condition the outputs of BERT with the polyphonic character of interest and its POS tagging. Rather than using the hard mask as in previous works, our experiments show that learning a soft-weighting function for the candidate phonemes benefits performance. In addition, our proposed g2pW does not require extra pre-trained POS tagging models while using POS tags as auxiliary features since we train the POS tagging model simultaneously with the unified encoder. Experimental results show that our g2pW outperforms existing methods on the public CPP dataset. All codes, model weights, and a user-friendly package are publicly available.

cs.CL↗
arXiv2022

Traditional Chinese Synthetic Datasets Verified with Labeled Data for Scene Text Recognition

Scene text recognition (STR) has been widely studied in academia and industry. Training a text recognition model often requires a large amount of labeled data, but data labeling can be difficult, expensive, or time-consuming, especially for Traditional Chinese text recognition. To the best of our knowledge, public datasets for Traditional Chinese text recognition are lacking. This paper presents a framework for a Traditional Chinese synthetic data engine which aims to improve text recognition model performance. We generated over 20 million synthetic data and collected over 7,000 manually labeled data TC-STR 7k-word as the benchmark. Experimental results show that a text recognition model can achieve much better accuracy either by training from scratch with our generated synthetic data or by further fine-tuning with TC-STR 7k-word.

cs.CV↗
arXiv2022

SMILE: Sequence-to-Sequence Domain Adaption with Minimizing Latent Entropy for Text Image Recognition

Training recognition models with synthetic images have achieved remarkable results in text recognition. However, recognizing text from real-world images still faces challenges due to the domain shift between synthetic and real-world text images. One of the strategies to eliminate the domain difference without manual annotation is unsupervised domain adaptation (UDA). Due to the characteristic of sequential labeling tasks, most popular UDA methods cannot be directly applied to text recognition. To tackle this problem, we proposed a UDA method with minimizing latent entropy on sequence-to-sequence attention-based models with classbalanced self-paced learning. Our experiments show that our proposed framework achieves better recognition results than the existing methods on most UDA text recognition benchmarks. All codes are publicly available.

cs.CV↗