arXiv subjects

Che-Wei Chang

Publications and source records attributed to Che-Wei Chang.

4 recordsLinked to original sources

arXiv2026

PRISM-Net: Dynamic Routing Network with VLM-Guided Disentangled Attribute Distillation for Explainable Fine-Grained Product Retrieval

Fine-grained visual product retrieval in large-scale e-commerce multimedia systems demands discriminative representations capable of simultaneously resolving macro-level category semantics and micro-level visual attributes such as structural contour (line) and chromatic appearance (color). However, conventional backbones including convolutional neural networks (CNNs) and vision Transformers (ViTs) inherently suffer from feature entanglement and shortcut learning. To learn the micro-level visual attributes without costly manual attribute annotations, we propose phase-routed interpretable semantic mixture-of-experts network (PRISM-Net), an explainable multi-expert architecture that disentangles holistic visual features into disentangled line and color expert pathways governed by a conditional meta-router. In phase 1, we introduce an external VLM-Guided sigmoid semantic distillation mechanism with gradient-detached routing gates, utilizing zero-shot text prompts from SigLIP to supervise specialized structural and chromatic experts. In phase 2, the disentangled expert embeddings are adaptively fused with macro base features via a learned meta-router gate to calibrate category boundaries. Extensive 5-fold cross-validation across 275,454 e-commerce images (18 categories) and full-scale retrieval benchmarks (55,091 queries against a 220,363-image gallery) across SwinV2, ConvNeXt-V2, and DINOv3 demonstrate substantial gains. On SwinV2, PRISM-Net boosts full-fold product mAP@10 from 80.98% to 83.90% (+2.92%), line mAP@10 by +5.74%, and color mAP@10 by +9.12%. Furthermore, unsupervised distribution analysis of the meta-router gate reveals that both ViT and CNN backbones emergently allocate up to 6.8x higher structural line weights to rigid appliances and hardware while dedicating >96% chromatic weight to clothing.

cs.CV↗

arXiv2025

Improving Table Retrieval with Question Generation from Partial Tables

Recent advances in open-domain question answering over tables have widely adopted large language models (LLMs) under the Retriever-Reader architecture. Prior works have effectively leveraged LLMs to tackle the complex reasoning demands of the Reader component, such as text-to-text, text-to-SQL, and multi hop reasoning. In contrast, the Retriever component has primarily focused on optimizing the query representation-training retrievers to retrieve relevant tables based on questions, or to select keywords from questions for matching table segments. However, little attention has been given to enhancing how tables themselves are represented in embedding space to better align with questions. To address this, we propose QGpT (Question Generation from Partial Tables), a simple yet effective method that uses an LLM to generate synthetic questions based on small portions of a table. These questions are generated to simulate how a user might query the content of the table currently under consideration. The generated questions are then jointly embedded with the partial table segments used for generation, enhancing semantic alignment with user queries. Without the need to embed entire tables, our method significantly improves retrieval performance across multiple benchmarks for both dense and late-interaction retrievers.

cs.IR↗

arXiv2021

Single-shot quantitative polarization imaging of complex birefringent structure dynamics

Polarization light microscopes are powerful tools for probing molecular order and orientation in birefringent materials. While a multitude of polarization light microscopy techniques are often used to access steady-state properties of birefringent samples, quantitative measurements of the molecular orientation dynamics on the millisecond time scale have remained a challenge. We propose polarized shearing interference microscopy (PSIM), a single-shot quantitative polarization imaging method, for extracting the retardance and orientation angle of the laser beam transmitting through optically anisotropic specimens with complex structures. The measurement accuracy and imaging performances of PSIM are validated by imaging a rotating wave plate and a bovine tendon specimen. We demonstrate that PSIM can quantify the dynamics of a flowing lyotropic chromonic liquid crystal in a microfluidic channel at an imaging speed of 506 frames per second (only limited by the camera frame rate), with a field-of-view of up to $350\times350 μm^2$ and a diffraction-limit spatial resolution of $\sim 2μm$. We envision that PSIM will find a broad range of applications in quantitative material characterization under dynamical conditions.

physics.optics↗

arXiv2018

Controlled spatial separation of spins and coherent dynamics in spin-orbit-coupled nanostructures

The spatial separation of electron spins followed by the control of their individual spin dynamics has recently emerged as an essential ingredient in many proposals for spin-based technologies because it would enable both of the two spin species to be simultaneously utilized, distinct from most of the current spintronic studies and technologies wherein only one spin species could be handled at a time. Here we demonstrate that the spatial spin splitting of a coherent beam of electrons can be achieved and controlled using the interplay between an external magnetic field and Rashba spin-orbit interaction in semiconductor nanostructures. The technique of transverse magnetic focusing is used to detect this spin separation. More notably, our ability to engineer the spin-orbit interactions enables us to simultaneously manipulate and probe the coherent spin dynamics of both spin species and hence their correlation, which could open a route towards spintronics and spin-based quantum information processing.

cond-mat.mes-hall↗