arXiv subjects

Yu Tian

Publications and source records attributed to Yu Tian.

At least 37 records · Page 2Linked to original sources

arXiv2026

Decoupling Knowledge and Privacy: Post-Task Self-Distillation Replay for LLM Continual Learning

Privacy-preserving continual learning (PPCL) must reduce the reproduction of sensitive content while retaining useful knowledge across sequential tasks. Formal privacy guarantees characterize randomized mechanisms, whereas operational output control concerns whether a trained model selectively reduces the likelihood of sensitive content in its outputs. In this work, we investigate the latter together with continual-learning utility under realistic task evolution. Retention and privacy correction operate at different granularities: task acquisition requires broad preservation of current- and old-task behavior, whereas privacy correction targets sparse annotated positions. Joint optimization leaves the current-task preservation target continually changing. We propose SPARK, a retention-correction decomposition that first freezes the learned post-task distribution and then applies selective correction around this stable reference. Self-Distillation Replay learns the current task while distilling behavior from previous tasks, and Post-Task Privacy Correction reduces annotated-PII likelihood while anchoring current- and old-task non-PII behavior to the resulting checkpoint. Extensive evaluations demonstrate that SPARK achieves effective selective PII suppression while preserving strong continual-learning utility and knowledge retention across diverse settings. Code and data will be released upon publication.

cs.LG↗

arXiv2026

Beyond Dense States: Sparse Transcoders as Causally Testable Operators for LLM Latent Reasoning

Latent reasoning reduces the token-generation cost of chain-of-thought reasoning by replacing explicit intermediate tokens with continuous latent transitions. However, existing latent reasoning methods usually rely on dense and entangled transitions, making their reasoning trajectories difficult to inspect or intervene on. We introduce LSTR (Latent Sparse Transcoder Reasoning), a framework that turns sparse transcoders from post-hoc diagnostic tools into in-loop, intervenable transition components for latent reasoning. At each latent step, a Latent Transition Transcoder (LTT) combines a linear skip path with a Top-k sparse innovation path, exposing a small set of active sparse features. Under matched compression settings, LSTR offers a mechanistically inspectable alternative to dense latent reasoning. On GSM8K-Aug, ablating only a few top-active sparse features reduces accuracy by up to 16.5%, whereas analogous interventions have much smaller effects in dense latent baselines. These results indicate that the active sparse features are causally involved in the latent transition process, rather than merely post-hoc descriptors. Additional experiments on mathematical benchmarks and StrategyQA suggest that sparse latent transitions can preserve the compression benefits of latent reasoning while making the resulting trajectories more inspectable and intervenable.

cs.AI↗

arXiv2026

Efficient Waveform Capture without Absorption with Synthesis of Complex Frequencies

A optical waveform can be synthesized by complex-frequency waves as well as by real-frequency harmonic waves. While single complex-frequency wave with exponentially rising waveform can be perfectly absorbed in lossless structures. Here, we propose that diverse optical waveforms can be captured without any absorption through the synthesis of complex frequencies in a lossless system. The scattering matrix zeros of the system correspond to a set of complex frequencies with exponentially rising waveforms, each of which can be virtually and perfectly absorbed. Thus, an input waveform, decomposed into these complex frequencies automatically, can be captured without any absorption. Then, in a well-designed coupled cavity system, various waveforms such as exponentially decaying, Gaussian, rectangular, triangular, and randomly generated piecewise-linear profiles, are captured with high efficiency. The proposed mechanism has potential applications in enhancing light-matter interactions and optical energy storage.

physics.optics↗

arXiv2026

Agentic-SQL Revisited: Autonomy-Based Taxonomy and Empirical Benchmark Analysis for LLM Text-to-SQL

LLM-based Text-to-SQL progress is reported across heterogeneous benchmarks, backbones, and inference protocols, making cross-system comparison fragile. We reframe the field as a leaderboard aggregation: we collect the metrics authors themselves report and organize them along an inference-autonomy axis spanning constrained, in-context, iterative, agentic, and reasoning-internalized generation, with traceable provenance for every cell. To anchor the aggregation empirically, we run a focused case study on Spider, comparing 8B open-source backbones with and without chain-of-thought (CoT) supervision against few-shot DeepSeek~V3 and GLM-4 baselines. Four patterns emerge: Spider gains transfer unevenly to BIRD and Spider~2.0; autonomy buys robustness at non-trivial cost; reasoning internalization sits between answer-only decoding and externally orchestrated agents; and CoT gains concentrate on Hard and Extra-Hard queries. We release a Python harness mirroring the autonomy axis so that future methods can be added directly to the leaderboard.

cs.AI↗

arXiv2026

When Entropy Is Not Enough: Reclaiming Lost Semantics in LLM Output Length Prediction

Efficient LLM serving is often bottlenecked by the need to pad sequences to a fixed maximum length, and this wastes compute and degrades throughput. Predicting output lengths in advance makes it possible to adopt length-aware scheduling, and this reduces the overhead. This advantage is especially pronounced in long-context reasoning and reinforcement learning applications. Existing approaches, such as entropy-guided token pooling, use token-wise entropy as their primary signal, but they tend to ignore differences in semantic content across tokens. So, important tokens are often underweighted, and tokens carrying little information receive disproportionate emphasis. This hurts the reliability of length prediction. We introduce ESTP (Entropy-and-Semantic Token Pooling), a lightweight framework that addresses this issue by combining entropy with attention-based importance scores. These scores are derived directly from the self-attention weights computed during the LLM prefill phase, and this allows ESTP to capture both uncertainty and semantic importance with minimal additional computation. Since the framework reuses prefill activations, it adds almost no extra memory overhead and introduces only minimal latency. On the ForeLen benchmark, ESTP outperforms baseline methods, achieves better prediction accuracy and lower error rates in most scenarios. When integrated with a length-aware scheduler in end-to-end system tests, it further helps improve overall throughput and reduce the padding ratio. Our results offer a practical and effective building block for length-aware LLM serving systems.

cs.AI↗

arXiv2026

Extremal non-rotating black holes have no fermionic Love

The static tidal Love numbers (TLNs) of $4$-dimensional black holes vanish for bosonic perturbations but are generically nonzero for fermions, with rare exceptions. In this paper, we show that for static, spherically symmetric black holes, fermionic TLNs vanish if and only if the black hole is extremal, in the sense that its horizon is degenerate. This follows from a closed formula for the static fermionic TLN of any asymptotically flat black hole, obtained by solving the static massless Dirac equation exactly on an arbitrary such spacetime and imposing regularity at the horizon. As applications, we analyze the Culetu--Simpson--Visser regular black hole and the loop-quantum-gravity remnant black holes, whose extremal configurations lead to vanishing fermionic yet nonvanishing bosonic TLNs.

gr-qc↗

arXiv2026

Formation of holographic vortex in a rotating shell-shaped superfluid

We investigate the holographic superfluid dynamics subjected to external rotation on a spherical geometry. Through a linear perturbation analysis, we identify several dynamically unstable phases in the phase diagram, each characterized by distinct unstable modes. Employing fully nonlinear numerical simulations, we further demonstrate that these unstable modes generically drive the system into vortex-antivortex configurations with definite winding numbers, determined by the symmetry of the corresponding unstable modes.

hep-th↗

arXiv2026

AWM-VLA: AlignedWorld Modeling for Efficient and Explainable Vision-Language-Action Policies

Vision-language-action (VLA) models have become a powerful paradigm for generalist robotic manipulation, yet they are often reactive: the policy maps the current observation directly to an action chunk without reasoning about the long-term consequences of its decisions. Prior attempts to endow policies with world models either reconstruct future frames in pixel space---expensive and dominated by task-irrelevant detail---or decouple the world model from the policy, weakening control. We present AWM-VLA, a unified framework that embeds aligned world modeling directly inside a diffusion-transformer policy. Following the Future Latent REpresentation Alignment (FLARE) principle, we add learnable future tokens whose intermediate activations are aligned with vision-language embeddings of future observations, enabling the policy to anticipate long-term consequences while generating actions. We extend this paradigm in two ways. First, we introduce an object-centric decoupled alignment objective that predicts future object-level semantics alongside the global future embedding, improving both interpretability and multi-instruction generalization. Second, we balance the global and object-centric alignment terms against the action flow-matching loss through a principled weighting, yielding a controllable accuracy--interpretability trade-off. On RoboCasa and humanoid tabletop manipulation benchmarks, AWM-VLA outperforms prior VLA and world-model baselines by up to 21% in success rate, improves generalization to novel objects and instructions, and produces object-centric rationales that are preferred by human raters in 83 of cases. Our approach adds only a few learnable tokens to the policy and is compatible with any diffusion or flow-matching policy, making aligned world modeling an inexpensive, broadly applicable component of generalist manipulation.

cs.CV↗

arXiv2026

Ollivier's Ricci Curvature on Complex-weighted Graphs

Understanding the geometry of complex networks is critical for effective modeling and analysis across domains. While discrete notions of Ricci curvature have emerged as powerful tools for characterizing both local and global network structure, existing formulations are largely confined to undirected networks with real-valued weights. This limits the use of curvature-based analysis of directional and complex-weighted relations that arise naturally in many applications, from social and biological systems to quantum and signal-processing networks. In this work, we introduce a principled extension of Ollivier's Ricci curvature to complex-weighted graphs, which encompasses directed graphs as a special case. We establish fundamental theoretical properties of this new notion, including relations to the magnetic Laplacian and combinatorial upper and lower bounds that relate curvature to cycle structure in local neighborhoods. We further develop computational methods for curvature estimation and demonstrate their utility in community detection on directed networks.

cs.SI↗

arXiv2026

UniJEPA: A Unified Joint-Embedding Predictive Architecture for Task-Agnostic Visual World Modeling

Joint-Embedding Predictive Architectures (JEPAs) have emerged as a principled framework for self-supervised learning of world models in compact latent spaces, yet existing methods are fragmented: some predict masked parts of a single image in latent space (I-JEPA), others learn to predict global photometric transformations (Image World Models), while video-scale JEPAs predict future temporal states and are post-trained for action-conditioned planning (V-JEPA~2, DINO-World, DINO-WM). These objectives are treated as distinct recipes with separate encoders, predictors, and anti-collapse regularizers, hindering a single model from unifying image-level and video-level world modeling. We present UniJEPA, a unified JEPA that jointly learns photometric prediction (image-level transformations) and temporal prediction (video-level next-state dynamics) in one shared latent space. A single end-to-end objective, composed of a next-embedding prediction loss and a Gaussian regularizer, yields a provably anti-collapse encoder-predictor pair trainable from raw pixels without EMA, stop-gradient, or pre-trained encoders. We show that the same latent space supports controllable abstraction: photometric prediction learns invariant structure while temporal prediction learns equivariant dynamics. After action-conditioned post-training on offline trajectories, UniJEPA enables zero-shot planning by treating goal features as prediction targets. On image, video, and control benchmarks, UniJEPA matches or surpasses task-specific JEPAs while requiring a single loss hyperparameter, and plans up to tens of times faster than generative world models at comparable accuracy.

cs.CV↗

arXiv2026

Longitudinal 3D Foundation Modeling for Neoadjuvant Breast Cancer Response Prediction from Serial DCE-MRI

Pathologic complete response (pCR) is an important endpoint in neoadjuvant chemotherapy (NAC) for breast cancer, and predicting pCR from imaging during treatment could support treatment response assessment. Many existing imaging-based approaches rely on a single static timepoint, which fails to capture changes that occur during treatment. In this work, we present a longitudinal framework that combines a frozen 3D foundation encoder (Pillar-0) with our Temporal Dynamics Network (TDN) to predict treatment response from serial Dynamic Contrast-Enhanced (DCE) MRI acquired across four clinical timepoints from pre-treatment to pre-surgery. The TDN combines time-aware volumetric embeddings with clinical and treatment data to predict pCR. Evaluated on 982 patients from the combined I-SPY2 and ACRIN-6698 cohort, the proposed model achieves strong performance across all reported metrics when longitudinal 3D imaging is fused with clinical data (test AUROC: 73.6%, balanced accuracy: 69.1%). While clinical variables provide the strongest individual predictive signal, longitudinal 3D imaging contributes complementary information when fused with clinical data, improving pCR prediction. Our source code is available at: https://github.com/omarftt/longitudinal_temporal_pillar.

eess.IV↗

arXiv2026

Enhancing Trustworthy Clinical Diagnosis Decision-Making in Large Language Models via Etiology-Aware Attention Supervision

Objective: Large Language Models (LLMs) have demonstrated strong capabilities in medical text understanding and generation. However, their trustworthiness in diagnosis-oriented medical tasks remains constrained by the lack of structured guidance on how clinically relevant diagnostic evidence is internally attended to and utilized during model learning. Method: We propose an Etiology-Aware Attention Supervision framework that introduces structured etiological information as an external supervisory signal for training large language models. Specifically, we construct Clinical Etiology Schema (CES) derived from authoritative clinical guidelines for three acute abdominal conditions: acute appendicitis, acute pancreatitis, and acute cholecystitis. Based on CES annotations, we develop an Etiology-Aware Head Identification strategy to identify attention heads that consistently align with etiological evidence. Building on this analysis, we design a structure-guided parameter-efficient fine-tuning approach that steers attention distributions toward clinically relevant evidence through an additional supervision loss, without modifying the base model architecture. Result: Experiments conducted on a Consistent Diagnosis Cohort demonstrate that the proposed framework improves average diagnostic accuracy by 15.65% compared with baseline models. Attention-based metrics, including Inference Focus Score and Inference Attention Frequency, show more concentrated attention on etiologically relevant evidence. External evaluation on a Discrepant Diagnosis Cohort further confirms the robustness of diagnostic performance improvements under real-world clinical inconsistencies.

cs.CL↗

arXiv2026

Emergence of critical phenomena from the black hole interior

The emergence of the $r=0$ singularity inside a spherically symmetric charged black hole, is studied numerically within the Einstein-Maxwell-real scalar model. When the scalar field reaches a critical strength, the $r=0$ singularity emerges inside the black hole at the tip of the causal diamond. By varying the parameter $p$ of the initial profile for the scalar field towards the critical value ${p_*}$, we observe the areal radius at the tip follows a power-law scaling, ${r_S } \propto {| {p - {p_*}}|^γ}$, with a universal critical exponent $γ\approx 0.5$. This remarkable discovery, analogous to Choptuik's critical phenomena for the black hole formation, provides the first evidence of the universality and scaling for the emergence of the $r=0$ singularity inside spherically symmetric charged black holes, offering new insights into the nonlinear dynamics of strong gravitational field.

gr-qc↗

arXiv2026

DiffuseAgent-MI: Distributionally-Grounded,Tool-Integrated Self-Evolving Agents for Faithful Visual Reasoning

Tool-integrated vision-language agents have made remarkable progress on compositional and multi-step visual reasoning. Yet their outputs frequently exhibit unfaithfulness: the stated reasoning path diverges from the computation that actually produced the answer, undermining reliability in safety-critical applications. We present DiffuseAgent-MI, a self-evolving agent whose perceptual grounding is governed by a KL-minimal energy model over feature units, providing a distributional view of visual mechanistic interpretability. The agent learns an energy landscape that softly constrains generated samples to lie near the native prior conditioned on the chosen interpretable unit, closing the gap between the explanation and the internal representation. A verifier then supplies trajectory-level faithfulness rewards, and a repair branch re-conditions the energy when the verifier flags an unfaithful step. On GeoQA, SciVis, VQA-v2 and an in-house multimodal reasoning set, DiffuseAgent-MI improves accuracy by up to 5.1 points over prior self-evolving agents while more than doubling mutual-information faithfulness and human-interpretability agreement. Our analysis shows the energy term and the verifier are complementary: the former guarantees distributional faithfulness, the latter trajectory-level faithfulness, and only their combination closes both gaps.

cs.CV↗

arXiv2026

CREST: Curvature-Regulated Event-Centric Sampling for Efficient Long-Video Understanding

Selecting informative frames from long videos is a combinatorial problem that existing methods address either through efficient heuristics without explicit modeling of query-conditioned temporal structure, or through multi stage retrieval pipelines with substantial preprocessing cost. We propose \textbf{CREST}, a training-free frame selection method grounded in the temporal geometry of query--frame relevance. CREST is based on the observation that relevance over time exhibits structured local variation: sharp curvature around salient events and flatter regions in redundant segments. By using local curvature to guide selection, CREST allocates a fixed frame budget more effectively across brief decisive events and slowly evolving evidence. Under a fixed backbone and frame budget, CREST achieves higher accuracy than AKS, a lightweight relevance--coverage baseline, on LongVideoBench and VideoMME, while retaining 93--95\% of the accuracy of MIRA, a stronger multi-stage retrieval pipeline, at only 3--4\% of its preprocessing cost.\footnote{Code and implementation details are included in the supplementary material and will be released publicly upon acceptance.} On TempRel, our diagnostic benchmark for temporal frame selection, CREST achieves a 6.88\% relative improvement over AKS. Pairwise LLM-as-a-judge evaluation further shows that CREST-selected frames yield more coherent frame-conditioned descriptions, with win rates of 60.58\% and 54.50\% on the two benchmarks. These results show that local temporal geometry provides a simple and efficient basis for long-video frame selection.

cs.CV↗

arXiv2026

Physics-Informed Neural Networks for 2D Plane Wave Scattering in Arbitrary Dielectric Structures

In this paper, we introduce a meshless physics-informed neural network based computational framework for solving two-dimensional electromagnetic wave scattering in inhomogeneous media. The framework embeds frequency-domain Maxwell equations and radiation boundary conditions directly into the neural network loss function, enabling accurate prediction of scattered fields for both transverse magnetic (TM) and transverse electric (TE) polarizations across various dielectric configurations. Application of the method to single-cylinder, concentric multilayer cylindrical shells, three arbitrarily arranged cylinders, and composite irregular structures demonstrates that for the TM polarization, all relative $L^{2}$ errors mostly remain at particularly low levels of $\le0.1$. For the TE polarization, sharp variations of the dielectric properties of scatterers lead to singularities in the governing equations, which result in decreased accuracy of the method. This challenge is overcome by introducing at dielectric boundaries a hyperbolic-tangent smoothing function. This procedure significantly improves the accuracy of the method, with the corresponding results closely matching the predictions of the finite-difference time-domain method. This framework exhibits stable convergence behavior across all of the investigated configurations, thus confirming its robustness and scalability to complex electromagnetic scattering problems.

physics.optics↗

arXiv2026

Media Meets Communication in 6G: Fundamentals, Key Technologies, and Applications

The rapid advancement of sixth-generation (6G) networks is accelerating the convergence of media intelligence and communication intelligence, driving media communication beyond conventional bit-level delivery toward intelligent, semantic-aware, and generative paradigms. Emerging media services require not only high data rates and low latency, but also semantic awareness, perceptual quality assurance, adaptive resource orchestration, trustworthy content processing, and personalized media generation. Meanwhile, media technologies are evolving from handcrafted signal processing and conventional coding toward artificial intelligence (AI)-driven representation learning, content understanding, and generative reconstruction. Motivated by these trends, this paper presents a systematic survey of media communication technologies for 6G vision communication by revisiting the evolution of communication and media technologies and clarifying the intrinsic relationship between media content processing and wireless transmission. We introduce a unified framework consisting of four key dimensions: AI-driven media technologies, media-aware wireless transmission, large model-enabled media communication, and intelligent network infrastructures. Specifically, AI-driven media technologies encompass media coding, content understanding, quality assessment, security and compliance detection, and AIGC-enabled media generation, while media-aware wireless transmission is examined from three complementary perspectives: semantic joint source-channel optimization, which jointly encodes task-relevant semantic information; source-aware transmission optimization, which leverages media characteristics for channel adaptation, prediction, and compensation; and channel-aware source optimization, which adapts media coding and reconstruction based on real-time channel conditions.

cs.IT↗

arXiv2026

Quantum Mpemba effect in holography

We investigate the quantum Mpemba effect in a holographic superfluid, in which states with stronger initial symmetry breaking relax faster toward the symmetry-restored equilibrium. We demonstrate its emergence by identifying the shifted free energy computed from the energy flux into the black hole horizon as monotonic distance measure. By decomposing the nonlinear bulk dynamics based on quasinormal modes, we reveal that the anomalous relaxation is governed by a dynamical competition in which the slowest-decaying mode is suppressed while the second mode is amplified. These findings provide a holographic perspective on the quantum Mpemba effect in nonequilibrium relaxation involving strongly coupled degrees of freedom.

hep-th↗