arXiv subjects

Liang Fan

Publications and source records attributed to Liang Fan.

4 recordsLinked to original sources

arXiv2026

GRACE: Grounded Reasoning via Adapter Composition and Evidence-Aware Calibration for Educational Visual Question Answering

Educational visual question answering, or VQA, requires models to solve curriculum-oriented multiple-choice questions using both language and visual evidence. Compared with conventional open-ended VQA, educational examples often include structured assessment metadata, diagrams or image contexts, and semantically close answer options, creating strong opportunities for question-option shortcuts. We develop and evaluate a parameter-efficient adaptation framework for a frozen multimodal large language model in this setting. We introduce GRACE, Grounded Reasoning via Adapter Composition and Evidence-Aware Calibration, a framework that uses the pedagogical state of each question to specialize lightweight language and vision adaptation. The state combines inference-visible subject, grouped skill, grade, visual-context, question-intent, and option-structure cues. GRACE uses factor-specific prompts and lightweight visual adapters, then applies evidence-aware option calibration to score all candidates under a shared multimodal context. On ScienceQA, GRACE improves a shared-adapter baseline from 90.5 percent to 93.1 percent overall accuracy and from 88.7 percent to 91.2 percent on image-context questions. Removing pedagogical composition, option calibration, or the visual adapter reduces overall accuracy by 1.4, 1.0, and 1.5 points, respectively. These controlled results show that structured educational state is an effective routing signal for parameter-efficient multimodal adaptation.

cs.MM↗

arXiv2026

Forecasting with Guidance: Representation-Level Supervision for Time Series Forecasting

Nowadays, time series forecasting is predominantly approached through the end-to-end training of deep learning architectures using error-based objectives. While this is effective at minimizing average loss, it encourages the encoder to discard informative yet extreme patterns. This results in smooth predictions and temporal representations that poorly capture salient dynamics. To address this issue, we propose ReGuider, a plug-in method that can be seamlessly integrated into any forecasting architecture. ReGuider leverages pretrained time series foundation models as semantic teachers. During training, the input sequence is processed together by the target forecasting model and the pretrained model. Rather than using the pretrained model's outputs directly, we extract its intermediate embeddings, which are rich in temporal and semantic information, and align them with the target model's encoder embeddings through representation-level supervision. This alignment process enables the encoder to learn more expressive temporal representations, thereby improving the accuracy of downstream forecasting. Extensive experimentation across diverse datasets and architectures demonstrates that our ReGuider consistently improves forecasting performance, confirming its effectiveness and versatility.

cs.LG↗

arXiv2025

From Disorder to Function: Entropy-Engineered Broadband Photonics with Ion-Transport-Stabilized Spectral Fidelity

The high-entropy halide-perovskite field has expanded rapidly, yet a key gap remains: configurational entropy is not yet a reliable, designable lever to co-deliver expanded photonic functionality and operational robustness with a composition-transferable mechanistic basis. Here we develop entropy-engineered rare-earth halide double-perovskite single crystals, Cs2Na(Sb, RE)Cl6 (RE3+ = Sc3+, Er3+, Yb3+, Tm3+), that simultaneously expand near-infrared (NIR) functionality and establish a mechanistic stability rule. Near-equiatomic B(III)-site alloying yields a single-phase high-entropy solid solution (Delta_Sconfig about 1.6R). Sb3+ serves as a sensitizer that unifies excitation and cooperatively activates multiple lanthanide channels, transforming the parent single-mode response into a broadband NIR output (~850-1600 nm) with three spectrally orthogonal fingerprint bands at 996, 1220, and 1540 nm. This tri-peak, self-referenced output enables redundancy-based ratiometric solvent identification and quantitative mixture sensing with reduced susceptibility to intensity drift. Accelerated aging under humidity and oxygen shows improved phase and emission stability versus single-component analogues. DFT and molecular dynamics attribute the robustness to strongly suppressed RE$^{3+}$/Cl$^-$ self-diffusion despite comparable H$_2$O/O$_2$ adsorption, kinetically impeding ion-migration-assisted reconstruction and degradation. Integration into a phosphor-converted LED delivers spectrally stable, broadband NIR illumination, establishing entropy engineering as a practical handle to couple expanded photonic functionality with mechanistically accountable durability in metal-halide photonics.

physics.optics↗

arXiv2025

ClearGCD: Mitigating Shortcut Learning For Robust Generalized Category Discovery

In open-world scenarios, Generalized Category Discovery (GCD) requires identifying both known and novel categories within unlabeled data. However, existing methods often suffer from prototype confusion caused by shortcut learning, which undermines generalization and leads to forgetting of known classes. We propose ClearGCD, a framework designed to mitigate reliance on non-semantic cues through two complementary mechanisms. First, Semantic View Alignment (SVA) generates strong augmentations via cross-class patch replacement and enforces semantic consistency using weak augmentations. Second, Shortcut Suppression Regularization (SSR) maintains an adaptive prototype bank that aligns known classes while encouraging separation of potential novel ones. ClearGCD can be seamlessly integrated into parametric GCD approaches and consistently outperforms state-of-the-art methods across multiple benchmarks.

cs.CV↗