arXiv subjects

Qinshan Zhang

Publications and source records attributed to Qinshan Zhang.

4 recordsLinked to original sources

arXiv2026

Editable-DeepSC: Reliable Cross-Modal Semantic Communications for Facial Editing

Interactive computer vision (CV) plays a crucial role in various real-world applications, whose performance is highly dependent on communication networks. Nonetheless, the data-oriented characteristics of conventional communications often do not align with the special needs of interactive CV tasks. To alleviate this issue, the recently emerged semantic communications only transmit task-related semantic information and exhibit a promising landscape to address this problem. However, the communication challenges associated with Semantic Facial Editing, one of the most important interactive CV applications on social media, still remain largely unexplored. In this paper, we fill this gap by proposing Editable-DeepSC, a novel cross-modal semantic communication approach for facial editing. Firstly, we theoretically discuss different transmission schemes that separately handle communications and editings, and emphasize the necessity of Joint Editing-Channel Coding (JECC) via iterative attributes matching, which integrates editings into the communication chain to preserve more semantic mutual information. To compactly represent the high-dimensional data, we leverage inversion methods via pre-trained StyleGAN priors for semantic coding. To tackle the dynamic channel noise conditions, we propose SNR-aware channel coding via model fine-tuning. Extensive experiments indicate that Editable-DeepSC can achieve superior editings while significantly saving the transmission bandwidth, even under high-resolution and out-of-distribution (OOD) settings.

cs.IT↗

arXiv2026

Generalization Bounds for Transformer Channel Decoders

Transformer channel decoders, such as the Error Correction Code Transformer (ECCT), have shown strong empirical performance in channel decoding, yet their generalization behavior remains theoretically unclear. This paper studies the generalization performance of ECCT from a learning-theoretic perspective. By establishing a connection between multiplicative noise estimation errors and bit-error-rate (BER), we derive an upper bound on the generalization gap via bit-wise Rademacher complexity. The resulting bound characterizes the dependence on code length, model parameters, and training set size, and applies to both single-layer and multi-layer ECCTs. We further show that parity-check-based masked attention induces sparsity that reduces the covering number, leading to a tighter generalization bound. To the best of our knowledge, this work provides the first theoretical generalization guarantees for this class of decoders.

cs.IT↗

arXiv2025

Constructions of block MDS LDPC codes from punctured circulant matrices

Low density parity check (LDPC) codes, initially discovered by Gallager, exhibit excellent performance in iterative decoding, approaching the Shannon limit. MDS array codes, with favorable algebraic structures, are codes suitable for decoding large burst errors. The Blaum-Roth (BR) code, an MDS array code similar to the Reed-Solomon (RS) code but has a parity-check matrix prone to $4$-cycles. Fossorier proposed constructing quasi-cyclic LDPC codes from circulant permutation matrices but are not MDS array codes. This paper aims to construct codes that possess both the block MDS property and have no $4$-cycles in the Tanner graph of their parity-check matrices, namely the so-called block MDS LDPC codes. Non-binary block MDS QC codes were first constructed by [Tauz {\it et al. }IEEE ITW, 2025] using circulant shift matrices. We first generate a family of block MDS codes over $\F_2$ from punctured circulant permutation matrices. Second, we construct a family of block MDS LDPC codes from circulant matrices with column weight $> 1$ (CM$(t)$). Additionally, we present the Moore determinant formula for CM$(t)$s and a sufficient condition to avoid $4$-cycles in CM$(t)$-QC LDPC codes' Tanner graphs for $t> 1$. We also point out the non-existence of binary block MDS CPM-QC LDPC codes. Compared to the codes constructed in [Li {\it et al. }IEEE TIT, 2023] and [Xiao {\it et al. }IEEE TCOM, 2021], our block MDS LDPC codes show enhanced random-error-correction at a similar code length and rate. Meanwhile, these codes can effectively combat burst errors when considered as array codes. Both of our two types of constructions for block MDS LDPC codes are applicable to the scenario of the binary field.

cs.IT↗

arXiv2024

Editable-DeepSC: Cross-Modal Editable Semantic Communication Systems

Different from data-oriented communication systems that primarily focus on how to accurately transmit every bit of data, task-oriented semantic communication systems only transmit the specific semantic information required by downstream tasks, strive to minimize the communication overhead and maintain competitive tasks execution performance in the presence of channel noise. However, it is worth noting that in many scenarios, the transmitted semantic information needs to be dynamically modified according to the users' preferences in a conversational and interactive way, which few existing works take into consideration. In this paper, we propose a novel cross-modal editable semantic communication system, named Editable-DeepSC, to tackle this challenge. By utilizing inversion methods based on StyleGAN priors, Editable-DeepSC takes cross-modal text-image pairs as the inputs and transmits the edited information of images based on textual instructions. Extensive numerical results demonstrate that our proposed Editable-DeepSC can achieve remarkable editing effects and transmission efficiency under the perturbations of channel noise, outperforming existing data-oriented communication methods.

cs.IT↗