arXiv subjects

Kai Li

Publications and source records attributed to Kai Li.

At least 127 records · Page 7Linked to original sources

arXiv2026

Learning to Trust Experience: A Monitor-Trust-Regulator Framework for Learning under Unobservable Feedback Reliability

Learning under unobservable feedback reliability poses a distinct challenge beyond optimization robustness: a system must decide whether to learn from an experience, not only how to learn stably. We study this setting as Epistemic Identifiability under Unobservable Reliability (EIUR), where each experience has a latent credibility, reliable and unreliable feedback can be locally indistinguishable, and data are generated in a closed loop by the learner's own evolving beliefs and actions. In EIUR, standard robust learning can converge stably yet form high-confidence, systematically wrong beliefs. We propose metacognitive regulation as a practical response: a second, introspective control loop that infers experience credibility from endogenous evidence in the learner's internal dynamics. We formalize this as a modular Monitor-Trust-Regulator (MTR) decomposition and instantiate it with self-diagnosis, which maintains a slowly varying experience-trust variable that softly modulates learning updates, without exogenous reliability labels or an explicit corruption model. Empirically, in the EIUR regimes studied here, self-diagnosis is associated with improved epistemic identifiability. In reinforcement learning, it enables calibrated skepticism and recovery under systematically corrupted rewards. In supervised learning, it exposes a critical dissociation: performance recovery does not imply epistemic recovery. Accuracy can rebound while internal belief dynamics remain locked-in by early misleading data, a failure detectable only through introspective diagnostics. Together, MTR and self-diagnosis provide an organizing abstraction and a concrete design template for intrinsic reliability assessment in autonomous learning under unobservable reliability.

cs.LG↗

arXiv2026

Training instability in deep learning follows low-dimensional dynamical principles

Deep learning systems achieve remarkable empirical performance, yet the stability of the training process itself remains poorly understood. Training unfolds as a high-dimensional dynamical system in which small perturbations to optimization, data, parameters, or learning signals can induce abrupt and irreversible collapse, undermining reproducibility and scalability. We propose a unified dynamical perspective that characterizes training stability as an intrinsic property of learning systems, organized along four interacting dimensions: optimization, environmental/data, parametric, and learning-signal stability. We operationalize this perspective through controlled perturbation auditing of training trajectories, probing how learning dynamics respond to structured disturbances without modifying learning algorithms. Across reinforcement learning and large language model training, we identify three recurring regularities: high final performance is frequently decoupled from training stability; controlled stochasticity consistently buffers learning dynamics across paradigms; and deviations in low-dimensional latent meta-states systematically precede observable performance collapse. Together, these findings establish training stability as a measurable and comparable dynamical property of learning systems, providing a descriptive foundation for studying learning dynamics beyond final performance outcomes.

cs.LG↗

arXiv2026

Toward Stable Semi-Supervised Remote Sensing Segmentation via Co-Guidance and Co-Fusion

Semi-supervised remote sensing (RS) image semantic segmentation offers a promising solution to alleviate the burden of exhaustive annotation, yet it fundamentally struggles with pseudo-label drift, a phenomenon where confirmation bias leads to the accumulation of errors during training. In this work, we propose Co2S, a stable semi-supervised RS segmentation framework that synergistically fuses priors from vision-language models and self-supervised models. Specifically, we construct a heterogeneous dual-student architecture comprising two distinct ViT-based vision foundation models initialized with pretrained CLIP and DINOv3 to mitigate error accumulation and pseudo-label drift. To effectively incorporate these distinct priors, an explicit-implicit semantic co-guidance mechanism is introduced that utilizes text embeddings and learnable queries to provide explicit and implicit class-level guidance, respectively, thereby jointly enhancing semantic consistency. Furthermore, a global-local feature collaborative fusion strategy is developed to effectively fuse the global contextual information captured by CLIP with the local details produced by DINOv3, enabling the model to generate highly precise segmentation results. Extensive experiments on six popular datasets demonstrate the superiority of the proposed method, which consistently achieves leading performance across various partition protocols and diverse scenarios. Project page is available at https://xavierjiezou.github.io/Co2S/.

cs.CV↗

arXiv2025

Detection of dark companions via the combination of eclipse timing variation, Hipparcos and/or Gaia astrometry: the cases of V Puppis and CY Ari

The third body is expected to shape the formation and evolution of close binary systems. In this work, we develop a method to detect and characterize the tertiary companion around eclipsing binaries through the combined analysis of eclipse timing variation, Hipparcos and/or Gaia astrometry. This method allows us to determine both the true mass and the inclination of the tertiary body that inferred from light-travel time effect. For the massive B-type binary V Pup, we do not confirm the previously reported 5.47-yr signal; instead, we identify a longer period of 14 yr. The orbital semi-major axis and mass of the outer body are revised to $a_C={17.88}_{-0.15}^{+0.15}$\,au and $M_C={7.73}_{-0.14}^{+0.14}\,M_\odot$, confirming it as a promising stellar-mass black-hole candidate for further follow-up study. For the tertiary of the contact binary CY Ari, we obtain $P_C=5.406_{-0.016}^{+0.017}$ yr, $e_C=0.526_{-0.027}^{+0.032}$, $I_C={85.6}_{-6.5}^{+7.8}$$^\circ$, and a true mass of $M_C=0.640_{-0.029}^{+0.029}\,M_\odot$, supporting the white dwarf hypothesis proposed in previous study. The orbits of both systems are nearly edge-on ($I=90^{\circ}$), implying that they may form in a coplanar environment. We highlight the advantages of our method for detecting dark companions in binary and triple systems.

astro-ph.SR↗

arXiv2025

Enhancing Supervised Composed Image Retrieval via Reasoning-Augmented Representation Engineering

Composed Image Retrieval (CIR) presents a significant challenge as it requires jointly understanding a reference image and a modified textual instruction to find relevant target images. Some existing methods attempt to use a two-stage approach to further refine retrieval results. However, this often requires additional training of a ranking model. Despite the success of Chain-of-Thought (CoT) techniques in reducing training costs for language models, their application in CIR tasks remains limited -- compressing visual information into text or relying on elaborate prompt designs. Besides, existing works only utilize it for zero-shot CIR, as it is challenging to achieve satisfactory results in supervised CIR with a well-trained model. In this work, we proposed a framework that includes the Pyramid Matching Model with Training-Free Refinement (PMTFR) to address these challenges. Through a simple but effective module called Pyramid Patcher, we enhanced the Pyramid Matching Model's understanding of visual information at different granularities. Inspired by representation engineering, we extracted representations from COT data and injected them into the LVLMs. This approach allowed us to obtain refined retrieval scores in the Training-Free Refinement paradigm without relying on explicit textual reasoning, further enhancing performance. Extensive experiments on CIR benchmarks demonstrate that PMTFR surpasses state-of-the-art methods in supervised CIR tasks. The code will be made public.

cs.CV↗

arXiv2025

FactorPortrait: Controllable Portrait Animation via Disentangled Expression, Pose, and Viewpoint

We introduce FactorPortrait, a video diffusion method for controllable portrait animation that enables lifelike synthesis from disentangled control signals of facial expressions, head movement, and camera viewpoints. Given a single portrait image, a driving video, and camera trajectories, our method animates the portrait by transferring facial expressions and head movements from the driving video while simultaneously enabling novel view synthesis from arbitrary viewpoints. We utilize a pre-trained image encoder to extract facial expression latents from the driving video as control signals for animation generation. Such latents implicitly capture nuanced facial expression dynamics with identity and pose information disentangled, and they are efficiently injected into the video diffusion transformer through our proposed expression controller. For camera and head pose control, we employ Plücker ray maps and normal maps rendered from 3D body mesh tracking. To train our model, we curate a large-scale synthetic dataset containing diverse combinations of camera viewpoints, head poses, and facial expression dynamics. Extensive experiments demonstrate that our method outperforms existing approaches in realism, expressiveness, control accuracy, and view consistency.

cs.CV↗

arXiv2025

Spectroscopic and Decade-long Photometric Observations of the Contact Binary V2790 Ori: Evidence for a Brown Dwarf Companion and a Solar-like Magnetic Activity Cycle

We present 22 sets of light curves and one radial velocity curve for the W UMa-type total eclipse contact binary system V2790 Ori, derived by combining all available public photometric data, the photometric data in previous studies, and our own spectroscopic and decade-long photometric observations. Our simultaneous analysis of the light curves and radial velocity curve shows that V2790 Ori is a W-subtype contact binary with a mass ratio of $q = 0.322(\pm0.001)$ and a shallow contact degree of $14.8(\pm0.6)\%$. The orbital period analysis based on 445 eclipsing minima reveals a secular decrease at a rate of $\dot P = -3.18 (\pm 0.75) \times 10^{-8}\mathrm{d~yr^{-1}}$, superimposed with a cyclic variation with an amplitude of $A = 8.98 (\pm 2.19) \times 10^{-4}~\mathrm{d}$ and a period of $ P_3 = 7.44 (\pm 0.52)~\mathrm{yr}$. The secular decrease is caused by AML via magnetic braking, while the cyclic period variation is explained by the light-travel time effect due to a third body, which is likely to be a brown dwarf. Furthermore, our analysis indicates a mass transfer from the more massive component to the less massive one at a rate of $1.22(\pm0.29) \times 10^{-8}~\mathrm{M_{\odot}~yr^{-1}}$. A model with a cool spot on each component was adopted to fit the O'Connell effect observed in the light curves. Since the O'Connell effect varies over time, we identified a solar-like magnetic activity cycle with a period of approximately 5.4 yr by analyzing the magnitude difference ($Δm$) at the two light maxima and the O'Connell effect ratio. In addition, evolutionary analysis suggests that V2790 Ori is a newly formed contact binary that evolved from a detached phase into the present contact configuration.

astro-ph.SR↗

arXiv2025

Adaptive Parameter Optimization for Robust Remote Photoplethysmography

Remote photoplethysmography (rPPG) enables contactless vital sign monitoring using standard RGB cameras. However, existing methods rely on fixed parameters optimized for particular lighting conditions and camera setups, limiting adaptability to diverse deployment environments. This paper introduces the Projection-based Robust Signal Mixing (PRISM) algorithm, a training-free method that jointly optimizes photometric detrending and color mixing through online parameter adaptation based on signal quality assessment. PRISM achieves state-of-the-art performance among unsupervised methods, with MAE of 0.77 bpm on PURE and 0.66 bpm on UBFC-rPPG, and accuracy of 97.3\% and 97.5\% respectively at a 5 bpm threshold. Statistical analysis confirms PRISM performs equivalently to leading supervised methods ($p > 0.2$), while maintaining real-time CPU performance without training. This validates that adaptive time series optimization significantly improves rPPG across diverse conditions.

cs.CV↗

arXiv2025

SCALER: SAM-Enhanced Collaborative Learning for Label-Deficient Concealed Object Segmentation

Existing methods for label-deficient concealed object segmentation (LDCOS) either rely on consistency constraints or Segment Anything Model (SAM)-based pseudo-labeling. However, their performance remains limited due to the intrinsic concealment of targets and the scarcity of annotations. This study investigates two key questions: (1) Can consistency constraints and SAM-based supervision be jointly integrated to better exploit complementary information and enhance the segmenter? and (2) beyond that, can the segmenter in turn guide SAM through reciprocal supervision, enabling mutual improvement? To answer these questions, we present SCALER, a unified collaborative framework toward LDCOS that jointly optimizes a mean-teacher segmenter and a learnable SAM. SCALER operates in two alternating phases. In \textbf{Phase \uppercase\expandafter{\romannumeral1}}, the segmenter is optimized under fixed SAM supervision using entropy-based image-level and uncertainty-based pixel-level weighting to select reliable pseudo-label regions and emphasize harder examples. In \textbf{Phase \uppercase\expandafter{\romannumeral2}}, SAM is updated via augmentation invariance and noise resistance losses, leveraging its inherent robustness to perturbations. Experiments demonstrate that SCALER yields consistent performance gains across eight semi- and weakly-supervised COS tasks. The results further suggest that SCALER can serve as a general training paradigm to enhance both lightweight segmenters and large foundation models under label-scarce conditions. Code will be released.

cs.CV↗

arXiv2025

FGNet: Leveraging Feature-Guided Attention to Refine SAM2 for 3D EM Neuron Segmentation

Accurate segmentation of neural structures in Electron Microscopy (EM) images is paramount for neuroscience. However, this task is challenged by intricate morphologies, low signal-to-noise ratios, and scarce annotations, limiting the accuracy and generalization of existing methods. To address these challenges, we seek to leverage the priors learned by visual foundation models on a vast amount of natural images to better tackle this task. Specifically, we propose a novel framework that can effectively transfer knowledge from Segment Anything 2 (SAM2), which is pre-trained on natural images, to the EM domain. We first use SAM2 to extract powerful, general-purpose features. To bridge the domain gap, we introduce a Feature-Guided Attention module that leverages semantic cues from SAM2 to guide a lightweight encoder, the Fine-Grained Encoder (FGE), in focusing on these challenging regions. Finally, a dual-affinity decoder generates both coarse and refined affinity maps. Experimental results demonstrate that our method achieves performance comparable to state-of-the-art (SOTA) approaches with the SAM2 weights frozen. Upon further fine-tuning on EM data, our method significantly outperforms existing SOTA methods. This study validates that transferring representations pre-trained on natural images, when combined with targeted domain-adaptive guidance, can effectively address the specific challenges in neuron segmentation.

cs.CV↗

arXiv2025

Feature-EndoGaussian: Feature Distilled Gaussian Splatting in Surgical Deformable Scene Reconstruction

Minimally invasive surgery (MIS) requires high-fidelity, real-time visual feedback of dynamic and low-texture surgical scenes. To address these requirements, we introduce FeatureEndo-4DGS (FE-4DGS), the first real time pipeline leveraging feature-distilled 4D Gaussian Splatting for simultaneous reconstruction and semantic segmentation of deformable surgical environments. Unlike prior feature-distilled methods restricted to static scenes, and existing 4D approaches that lack semantic integration, FE-4DGS seamlessly leverages pre-trained 2D semantic embeddings to produce a unified 4D representation-where semantics also deform with tissue motion. This unified approach enables the generation of real-time RGB and semantic outputs through a single, parallelized rasterization process. Despite the additional complexity from feature distillation, FE-4DGS sustains real-time rendering (61 FPS) with a compact footprint, achieves state-of-the-art rendering fidelity on EndoNeRF (39.1 PSNR) and SCARED (27.3 PSNR), and delivers competitive EndoVis18 segmentation, matching or exceeding strong 2D baselines for binary segmentation tasks (0.93 DSC) and remaining competitive for multi-label segmentation (0.77 DSC).

cs.CV↗

arXiv2025

Gravity induced entanglement of multiple massive particles with large spin

We investigate the generation rate of the quantum entanglement in a system composed of multiple massive particles with large spin, where the mass of a single particle can be split into multiple trajectories by a generalized Stern-Gerlach interferometer. Taking the coherent spin states (CSS) as the initial state and considering the gravitational interaction due to Newtonian potential, we compute the generation rate of the entanglement for different configurations of the setup. Explicitly, the optimal polar angles of the spin are found numerically for systems with three and four particles, respectively. We conclude that the amount of the entanglement increases with the number of particles as well as the spin, and the configuration of the prism with a particle at the center generates the best rate of the entanglement.

gr-qc↗

arXiv2025

Deep (Predictive) Discounted Counterfactual Regret Minimization

Counterfactual regret minimization (CFR) is a family of algorithms for effectively solving imperfect-information games. To enhance CFR's applicability in large games, researchers use neural networks to approximate its behavior. However, existing methods are mainly based on vanilla CFR and struggle to effectively integrate more advanced CFR variants. In this work, we propose an efficient model-free neural CFR algorithm, overcoming the limitations of existing methods in approximating advanced CFR variants. At each iteration, it collects variance-reduced sampled advantages based on a value network, fits cumulative advantages by bootstrapping, and applies discounting and clipping operations to simulate the update mechanisms of advanced CFR variants. Experimental results show that, compared with model-free neural algorithms, it exhibits faster convergence in typical imperfect-information games and demonstrates stronger adversarial performance in a large poker game.

cs.LG↗

arXiv2025

BridgeVoC: Revitalizing Neural Vocoder from a Restoration Perspective

This paper revisits the neural vocoder task through the lens of audio restoration and propose a novel diffusion vocoder called BridgeVoC. Specifically, by rank analysis, we compare the rank characteristics of Mel-spectrum with other common acoustic degradation factors, and cast the vocoder task as a specialized case of audio restoration, where the range-space spectral (RSS) surrogate of the target spectrum acts as the degraded input. Based on that, we introduce the Schrodinger bridge framework for diffusion modeling, which defines the RSS and target spectrum as dual endpoints of the stochastic generation trajectory. Further, to fully utilize the hierarchical prior of subbands in the time-frequency (T-F) domain, we elaborately devise a novel subband-aware convolutional diffusion network as the data predictor, where subbands are divided following an uneven strategy, and convolutional-style attention module is employed with large kernels for efficient T-F contextual modeling. To enable single-step inference, we propose an omnidirectional distillation loss to facilitate effective information transfer from the teacher model to the student model, and the performance is improved by combining target-related and bijective consistency losses. Comprehensive experiments are conducted on various benchmarks and out-of-distribution datasets. Quantitative and qualitative results show that while enjoying fewer parameters, lower computational cost, and competitive inference speed, the proposed BridgeVoC yields stateof-the-art performance over existing advanced GAN-, DDPMand flow-matching-based baselines with only 4 sampling steps. And consistent superiority is still achieved with single-step inference.

cs.SD↗

arXiv2025

DGTRSD & DGTRS-CLIP: A Dual-Granularity Remote Sensing Image-Text Dataset and Vision Language Foundation Model for Alignment

Vision Language Foundation Models based on CLIP architecture for remote sensing primarily rely on short text captions, which often result in incomplete semantic representations. Although longer captions convey richer information, existing models struggle to process them effectively because of limited text-encoding capacity, and there remains a shortage of resources that align remote sensing images with both short text and long text captions. To address this gap, we introduce DGTRSD, a dual-granularity remote sensing image-text dataset, where each image is paired with both a short text caption and a long text description, providing a solid foundation for dual-granularity semantic modeling. Based on this, we further propose DGTRS-CLIP, a dual-granularity curriculum learning framework that combines short text and long text supervision to achieve dual-granularity semantic alignment. Extensive experiments on four typical zero-shot tasks: long text cross-modal retrieval, short text cross-modal retrieval, image classification, and semantic localization demonstrate that DGTRS-CLIP consistently outperforms existing methods across all tasks. The code has been open-sourced and is available at https://github.com/MitsuiChen14/DGTRS.

cs.CV↗

arXiv2025

A New Hybrid Precoding Approach for Multi-user Massive MIMO over Fading Channels

Hybrid precoding is an indispensable technique to harness the full potential of a multi-user massive multiple-input, multiple-output (MU-MMIMO) system. In this paper, we propose a new hybrid precoding approach that combines digital and analog precoding to optimize data transmission over multiple antennas. This approach steers signals in specific directions, leading to maximizing sum-rate and suppressing side-lobe interference. When dealing with complex signals, changes in phase are naturally associated with changes in angle, and these variations are inherently correlated. The correlation between the angle and phase is essential for accurately determining the channel characteristics. An important aspect of this approach is that we model the angle and phase as correlated variables following a bivariate Gaussian distribution, and for the first time, we define a joint angle and phase entropy to measure the uncertainty of angle and phase variations in wireless channels. This entropy is crucial to adapt the proposed precoding method with variations. Simulation result validate the accuracy of our analytical findings, demonstrating 18.31% increase in sum-rate and an 11.47% improvement in robustness compared to other state-of-the-art methods.

cs.NI↗

arXiv2025

Strategic Task Offloading for Delay-Sensitive IoT Applications: A Game-Theory-Based Demand-Supply Mechanism with Participation Incentives

Delay-sensitive Internet of Things (IoT) applications have drawn significant attention. Running many of these applications on IoT devices is challenging due to the limited processing resources of these devices and the need for real-time responses. Task offloading can minimize latency by transferring computationally intensive tasks from IoT devices to resource-rich edge servers, ensuring delay and performance guarantees. In this paper, we develop a task-offloading approach for delay-sensitive IoT applications in edge computing environments. Unlike existing schemes, we model the task offloading problem as an economic demand and supply model to achieve market balance. The proposed model avoids under- and over-supply, ensuring the computational resources at edge servers (supply) are allocated in a manner that best meets the processing and computational needs of user devices (demand). Given the multi-agent nature of task offloading involving users and service providers with different preferences and objectives, we design a game-theoretic framework using a Vickrey-Clarke-Groves (VCG) auction. This framework analyzes agent interactions and decision-making processes. Additionally, we develop an incentive mechanism to encourage both parties to participate in the auction. The mechanism maximizes user task offloading to edge servers and motivates edge servers to share their computational resources, achieving profitability for both IoT users and edge servers. Simulations demonstrate our method maximizes social welfare, ensures truthfulness, maintains market balance, and provides latency guarantees for delay-sensitive IoT applications.

cs.NI↗

arXiv2025

Photometric and Spectroscopic Studies of Four Low Mass-ratio Contact Binaries with Period Longer than 0.7 days

Photometric and spectroscopic studies of four long-period low mass-ratio contact binaries, V0508 And, V0844 Aur, V0699 Cep, and NSVS 6259046, are performed. V0508 And, V0844 Aur, and V0699 Cep are found to be A-type low-mass-ratio medium-contact binaries, while NSVS 6259046 is found to be an A-type deep-contact binary. O - C analysis indicates no long-term variation in V0844 Aur. However, the orbital periods of the other three targets are increasing. We conclude that V0844 Aur, V0699 Cep and NSVS 6259046 are magnetically active, as evidenced by the presence and variable nature of the O'Connell effect in these systems. By analyzing the LAMOST spectroscopic data, we find out that NSVS 6259046 and V0508 And exhibit no chromospheric activity on the dates the LAMOST spectra were taken, while the low signal-to-noise ratio in LAMOST data for V0844 Aur prevents us from obtaining reliable results. We discover that V0699 Cep is an early-type contact binary with chromospheric activity. Their initial masses and ages are calculated. All four systems are determined to be currently stable. We collect 217 contact binaries with both spectroscopic and photometric observations, and compare the differences between short-period and long-period systems in terms of mass-luminosity relation and mass-radius relation, using 0.7 days as the period boundary.

astro-ph.SR↗