arXiv subjects

Dan Li

Publications and source records attributed to Dan Li.

At least 55 records · Page 3Linked to original sources

arXiv2026

Mamoda2.5: Enhancing Unified Multimodal Model with DiT-MoE

We present Mamoda2.5, a unified AR-Diffusion framework that seamlessly integrates multimodal understanding and generation within a single architecture. To efficiently enhance the model's generation capability, we equip the Diffusion Transformer backbone with a fine-grained Mixture-of-Experts (MoE) design (128 experts, Top-8 routing), yielding a 25B-parameter model that activates only 3B parameters, significantly reducing training costs while scaling up the model capacity. Mamoda2.5 achieves top-tier generation performance on VBench 2.0 and sets a new record in video editing quality, surpassing evaluated open-source models and matching the performance of current top-tier proprietary models, including the Kling O1 on OpenVE-Bench. Furthermore, we introduce a joint few-step distillation and reinforcement learning framework that compresses the 30-step editing model into a 4-step model and greatly accelerates model inference. Compared to open-source baselines, Mamoda2.5 achieves up to $95.9\times$ faster video editing inference. In real-world applications, Mamoda2.5 has been successfully deployed for content moderation and creative restoration tasks in advertising scenarios, achieving a 98% success rate in internal advertising video editing scenario.

cs.CV↗

arXiv2026

Human-Less LLM Serving: Quantifying the Human Tax on Throughput

Every major LLM serving system is designed to meet TTFT and TPOT SLOs. These metrics capture latency as a human user perceives it, and the mechanisms built to satisfy them are now standard infrastructure. We observe that long-horizon AI tasks call LLMs programmatically in tight loops where no human observes TTFT or TPOT. We ask: how much throughput do serving systems sacrifice to meet TTFT and TPOT SLAs that these workloads never need? We conduct a systematic measurement study across chunk sizes, SLO settings, context lengths, and concurrency levels. We find that the human tax on throughput grows substantially with context length and lands in the 60-93% range. At 64K token contexts, tightening the TTFT SLO to production-typical settings costs a large fraction of throughput versus the human-less baseline. The human tax is larger at higher concurrency and is qualitatively similar across SGLang and Sarathi-Serve. We term the unconstrained optimum human-less serving and provide a prototype demonstrating that it is practical on real workloads. Our findings argue that serving systems should expose workload-class-aware SLA configurations rather than silently applying the human tax uniformly to all traffic.

cs.NI↗

arXiv2026

WaveMoE: A Wavelet-Enhanced Mixture-of-Experts Foundation Model for Time Series Forecasting

Time series foundation models (TSFMs) have recently achieved remarkable success in universal forecasting by leveraging large-scale pretraining on diverse time series data. Complementing this progress, incorporating frequency-domain information yields promising performance in enhancing the modeling of complex temporal patterns, such as periodicity and localized high-frequency dynamics, which are prevalent in real-world time series. To advance this direction, we propose a new perspective that integrates explicit frequency-domain representations into scalable foundation models, and introduce WaveMoE, a wavelet-enhanced mixture-of-experts foundation model for time series forecasting. WaveMoE adopts a dual-path architecture that jointly processes time series tokens and wavelet tokens aligned along a unified temporal axis, and coordinates them through a shared expert routing mechanism that enables consistent expert specialization while efficiently scaling model capacity. Preliminary experimental results on 16 diverse benchmark datasets indicate that WaveMoE has the potential to further improve forecasting performance by incorporating wavelet-domain corpora.

cs.LG↗

arXiv2026

Tensor-Efficient High-Dimensional Q-learning

High-dimensional reinforcement learning(RL) faces challenges with complex calculations and low sample efficiency in large state-action spaces. Q-learning algorithms struggle particularly with the curse of dimensionality, where the number of state-action pairs grows exponentially with problem size. While neural network-based approaches like Deep Q-Networks have shown success, they do not explicitly exploit problem structure. Many high-dimensional control tasks exhibit low-rank structure in their value functions, and tensor-based methods using low-rank decomposition offer parameter-efficient representations. However, existing tensor-based Q-learning methods focus on representation fidelity without leveraging this structure for exploration. We propose Tensor-Efficient Q-Learning (TEQL), which represents the Q-function as a low-rank CP tensor over discretized state-action spaces and exploits the tensor structure for uncertainty-aware exploration. TEQL incorporates Error-Uncertainty Guided Exploration (EUGE), which combines tensor approximation error with visit counts to guide action selection, along with frequency-aware regularization to stabilize updates. Under matched parameter budgets, experiments on classic control tasks demonstrate that TEQL outperforms both matrix-based low-rank methods and deep RL baselines in sample efficiency, making it suitable for resource-constrained applications where sampling costs are high.

cs.LG↗

arXiv2026

Democratizing Federated Learning with Blockchain and Multi-Task Peer Prediction

The synergy between Federated Learning and blockchain has been considered promising; however, the computationally intensive nature of contribution measurement conflicts with the strict computation and storage limits of blockchain systems. We propose a novel concept to decentralize the AI training process using blockchain technology and Multi-task Peer Prediction. By leveraging smart contracts and cryptocurrencies to incentivize contributions to the training process, we aim to harness the mutual benefits of AI and blockchain. We discuss the advantages and limitations of our design.

cs.CR↗

arXiv2026

MsFormer: Enabling Robust Predictive Maintenance Services for Industrial Devices

Providing reliable predictive maintenance is a critical industrial AI service essential for ensuring the high availability of manufacturing devices. Existing deep-learning methods present competitive results on such tasks but lack a general service-oriented framework to capture complex dependencies in industrial IoT sensor data. While Transformer-based models show strong sequence modeling capabilities, their direct deployment as robust AI services faces significant bottlenecks. Specifically, streaming sensor data collected in real-world service environments often exhibits multi-scale temporal correlations driven by machine working principles. Besides, the datasets available for training time-to-failure predictive services are typically limited in size. These issues pose significant challenges for directly applying existing models as robust predictive services. To address these challenges, we propose MsFormer, a lightweight Multi-scale Transformer designed as a unified AI service model for reliable industrial predictive maintenance. MsFormer incorporates a Multi-scale Sampling (MS) module and a tailored position encoding mechanism to capture sequential correlations across multi-streaming service data. Additionally, to accommodate data-scarce service environments, MsFormer adopts a lightweight attention mechanism with straightforward pooling operations instead of self-attention. Extensive experiments on real-world datasets demonstrate that the proposed framework achieves significant performance improvements over state-of-the-art methods. Furthermore, MsFormer outperforms across industrial devices and operating conditions, demonstrating strong generalizability while maintaining a highly reliable Quality of Service (QoS).

cs.LG↗

arXiv2026

Rating Quality of Diverse Time Series Data by Meta-learning from LLM Judgment

High-quality time series (TS) data are essential for ensuring TS model performance, rendering research on rating TS data quality indispensable. Existing methods have shown promising rating accuracy within individual domains, primarily by extending data quality rating techniques such as influence functions and Shapley values to account for temporal characteristics. However, they neglect the fact that real-world TS data can span vastly different domains and exhibit distinct properties, hampering the accurate and efficient rating of diverse TS data. In this paper, we propose TSRating, a novel and unified framework for rating the quality of time series data crawled from diverse domains. TSRating leverages LLMs' inherent ample knowledge, acquired during their extensive pretraining, to comprehend and discern quality differences in diverse TS data. We verify this by devising a series of prompts to elicit quality comparisons from LLMs for pairs of TS samples. We then fit a dedicated rating model, termed TSRater, to convert the LLMs' judgments into efficient quality predictions by inferring future TS samples through TSRater's inference. To ensure cross-domain adaptability, we develop a meta-learning scheme to train TSRater on quality comparisons collected from nine distinct domains. To improve training efficiency, we employ signSGD for inner-loop updates, thus circumventing the demanding computation of hypergradients. Extensive experimental results on eleven benchmark datasets across three time series tasks, each using both conventional TS models and TS foundation models, demonstrate that TSRating outperforms baselines in terms of estimation accuracy, efficiency, and domain adaptability.

cs.LG↗

arXiv2026

Lightweight Time Series Data Valuation on Time Series Foundation Models via In-Context Finetuning

Time series foundation models (TSFMs) have demonstrated increasing capabilities due to their extensive pretraining on large volumes of diverse time series data. Consequently, the quality of time series data is crucial to TSFM performance, rendering an accurate and efficient data valuation of time series for TSFMs indispensable. However, traditional data valuation methods, such as influence functions, face severe computational bottlenecks due to their poor scalability with growing TSFM model sizes and often fail to preserve temporal dependencies. In this paper, we propose LTSV, a Lightweight Time Series Valuation on TSFMS via in-context finetuning. Grounded in the theoretical evidence that in-context finetuning approximates the influence function, LTSV estimates a sample's contribution by measuring the change in context loss after in-context finetuning, leveraging the strong generalization capabilities of TSFMs to produce robust and transferable data valuations. To capture temporal dependencies, we introduce temporal block aggregation, which integrates per-block influence scores across overlapping time windows. Experiments across multiple time series datasets and models demonstrate that LTSV consistently provides reliable and strong valuation performance, while maintaining manageable computational requirements. Our results suggest that in-context finetuning on time series foundation models provides a practical and effective bridge between data attribution and model generalization in time series learning.

cs.LG↗

arXiv2026

BACE-RUL: A Bi-directional Adversarial Network with Covariate Encoding for Machine Remaining Useful Life Prediction

Prognostic and Health Management (PHM) are crucial ways to avoid unnecessary maintenance for Cyber-Physical Systems (CPS) and improve system reliability. Predicting the Remaining Useful Life (RUL) is one of the most challenging tasks for PHM. Existing methods require prior knowledge about the system, contrived assumptions, or temporal mining to model the life cycles of machine equipment/devices, resulting in diminished accuracy and limited applicability in real-world scenarios. This paper proposes a Bi-directional Adversarial network with Covariate Encoding for machine Remaining Useful Life (BACE-RUL) prediction, which only adopts sensor measurements from the current life cycle to predict RUL rather than relying on previous consecutive cycle recordings. The current sensor measurements of mechanical devices are encoded to a conditional space to better understand the implicit inner mechanical status. The predictor is trained as a conditional generative network with the encoded sensor measurements as its conditions. Various experiments on several real-world datasets, including the turbofan aircraft engine dataset and the dataset collected from degradation experiments of Li-Ion battery cells, show that the proposed model is a general framework and outperforms state-of-the-art methods.

cs.LG↗

arXiv2026

The Steiner Tree Problem: Novel QUBO Formulation and Quantum Annealing Implementation

The Steiner Tree Problem (STP) is a well-known NP-hard combinatorial optimization problem, which has wide applications in network design, integrated circuit layout, bioinformatics, and other fields. However, traditional algorithms often struggle to balance efficiency and solution quality when dealing with large-scale STP instances. In this paper, we propose a new quantum annealing-based algorithm for solving the STP: we first model the STP into a quadratic unconstrained binary optimization (QUBO) form suitable for quantum annealing, then design a corresponding encoding strategy, and finally verify the algorithm through experimental tests. The results show that our quantum annealing-based method can obtain high-quality solutions with relatively low computational overhead for moderate-scale STP instances, providing a new feasible path for handling this intractable combinatorial optimization problem.

quant-ph↗

arXiv2026

Supercharging Packet-level Network Simulation of Large Model Training via Memoization and Fast-Forwarding

Packet-level discrete-event simulation (PLDES) is a prevalent tool for evaluating detailed performance of large model training. Although PLDES offers high fidelity and generality, its slow performance has plagued networking practitioners. Existing optimization techniques either simplify the network model, resulting in large errors; or execute it in parallel using multiple processors, with an upper bound on speedup. This paper explores an alternative optimization direction that reduces the computational loads of PLDES while maintaining high fidelity. Our key insight is that, in distributed LLM training, packet-level traffic behaviors often exhibit repetitive contention patterns and steady-states where flow rates stabilize, ignoring these redundant discrete events speeds up the simulation considerably and the error is negligible. We realize this idea by proposing Wormhole, a user-transparent PLDES kernel capable of automatically memoization for unsteady-states and skipping for steady-states. Wormhole adopts network partitioning, state memoization and reuse, and rate-based steady-state identification to accurately determine the periods of each flow's steady-state, while maintaining simulation consistency after fast-forwarding. Experiments demonstrate that Wormhole can achieve a 744x speedup over the original ns-3 (510x for MoE workload), with a bounded error of <1%. Applying current multithreading parallel techniques and Wormhole together allows a 1012x speedup, reducing the simulation time for one GPT-13B training under 128 GPUs from 9 hours to 5 minutes.

cs.NI↗

arXiv2026

EntWorld: A Holistic Environment and Benchmark for Verifiable Enterprise GUI Agents

Recent advances in Multimodal Large Language Models (MLLMs) have enabled agents to operate in open-ended web and operating system environments. However, existing benchmarks predominantly target consumer-oriented scenarios (e.g., e-commerce and travel booking), failing to capture the complexity and rigor of professional enterprise workflows. Enterprise systems pose distinct challenges, including high-density user interfaces, strict business logic constraints, and a strong reliance on precise, state-consistent information retrieval-settings in which current generalist agents often struggle. To address this gap, we introduce EntWorld, a large-scale benchmark consisting of 1,756 tasks across six representative enterprise domains, including customer relationship management (CRM), information technology infrastructure library (ITIL), and enterprise resource planning (ERP) systems. Unlike previous datasets that depend on fragile execution traces or extensive manual annotation, EntWorld adopts a schema-grounded task generation framework that directly reverse-engineers business logic from underlying database schemas, enabling the synthesis of realistic, long-horizon workflows. Moreover, we propose a SQL-based deterministic verification mechanism in building datasets that replaces ambiguous visual matching with rigorous state-transition validation. Experimental results demonstrate that state-of-the-art models (e.g., GPT-4.1) achieve 47.61% success rate on EntWorld, substantially lower than the human performance, highlighting a pronounced enterprise gap in current agentic capabilities and the necessity of developing domain-specific agents. We release EntWorld as a rigorous testbed to facilitate the development and evaluation of the next generation of enterprise-ready digital agents.

cs.AI↗

arXiv2026

A Large-Scale IPv6-Based Measurement of the Starlink Network

Low Earth Orbit (LEO) satellite networks have attracted considerable attention for their ability to deliver global, low-latency broadband Internet services. In this paper, we present a large-scale measurement study of the Starlink network, the largest LEO satellite constellation to date. We first propose an efficient method for discovering active Starlink user routers, identifying approximately 5.98 million IPv6 addresses across 208 regions in 165 countries. Compared to general-purpose IPv6 target generation algorithms, our router-centric approach achieves near-complete coverage and, to the best of our knowledge, yields the most comprehensive known set of active IPv6 addresses for Starlink user routers. Based on the discovered user routers, we further propose an efficient method for mapping the Starlink backbone network and uncover a topology consisting of 49 Points of Presence (PoPs) interconnected by 98 links. We conduct a detailed statistical analysis of active Starlink user routers and PoPs, and further characterize the IPv6 address assignment strategy adopted by the Starlink network. Finally, we analyze the latency of Starlink user routers, propose a method to distinguish different types of users within the same region using outside-in measurement, and identify the ongoing V2 Mini satellite deployment as a potential driver of the performance improvements. The dataset of the Starlink backbone network is publicly available at https://ki3.org.cn/#/starlink-network.

cs.NI↗

arXiv2026

Prototype-Guided Non-Exemplar Continual Learning for Cross-subject EEG Decoding

Due to the significant variability in electroencephalo-gram (EEG) signals across individuals, knowledge acquired from previous subjects is often overwritten as new subjects are introduced in continual EEG decoding tasks. Existing methods mainly rely on storing historical data from seen subjects as replay buffers to mitigate forgetting, which is impractical under privacy or memory constraints. To address this issue, we propose a Prototype-guided Non-Exemplar Continual Learning (ProNECL) framework that preserves prior knowledge without accessing historical EEG samples. ProNECL summarizes subject-specific discriminative representations into class-level prototypes and incrementally aligns new subject representations with a global prototype memory through prototype-based feature regulariza-tion and cross-subject alignment. Experiments on the BCI Com-petition IV 2a and 2b datasets demonstrate that ProNECL effec-tively balances knowledge retention and adaptability, achieving superior performance in cross-subject continual EEG decoding tasks.

cs.LG↗

arXiv2026

Understanding and Preserving Safety in Fine-Tuned LLMs

Fine-tuning is an essential and pervasive functionality for applying large language models (LLMs) to downstream tasks. However, it has the potential to substantially degrade safety alignment, e.g., by greatly increasing susceptibility to jailbreak attacks, even when the fine-tuning data is entirely harmless. Despite garnering growing attention in defense efforts during the fine-tuning stage, existing methods struggle with a persistent safety-utility dilemma: emphasizing safety compromises task performance, whereas prioritizing utility typically requires deep fine-tuning that inevitably leads to steep safety declination. In this work, we address this dilemma by shedding new light on the geometric interaction between safety- and utility-oriented gradients in safety-aligned LLMs. Through systematic empirical analysis, we uncover three key insights: (I) safety gradients lie in a low-rank subspace, while utility gradients span a broader high-dimensional space; (II) these subspaces are often negatively correlated, causing directional conflicts during fine-tuning; and (III) the dominant safety direction can be efficiently estimated from a single sample. Building upon these novel insights, we propose safety-preserving fine-tuning (SPF), a lightweight approach that explicitly removes gradient components conflicting with the low-rank safety subspace. Theoretically, we show that SPF guarantees utility convergence while bounding safety drift. Empirically, SPF consistently maintains downstream task performance and recovers nearly all pre-trained safety alignment, even under adversarial fine-tuning scenarios. Furthermore, SPF exhibits robust resistance to both deep fine-tuning and dynamic jailbreak attacks. Together, our findings provide new mechanistic understanding and practical guidance toward always-aligned LLM fine-tuning.

cs.LG↗

arXiv2026

Manipulating Anomalous Transport via Crystal Symmetry in 2D Altermagnets

Anomalous transports, including the anomalous Hall effect (AHE) and anomalous Nernst effect (ANE), are typical manifestations of time-reversal-symmetry-breaking responses in materials. In general, the two Hall states with opposite Hall conductivities can be regarded as time-reversal pairs coupled to magnetic order, and switching between them relies on reversing the magnetization via an external magnetic field or electric current. Here, we introduce a approach for manipulating anomalous transport through crystal symmetry engineering in two-dimensional (2D) altermagnetic systems. Based on symmetry analysis, we demonstrate that 2D altermagnets (AM) with out-of-plane Néel vectors will not host any anomalous Hall transport. Remarkably, breaking the symmetry connecting the two magnetic sublattices, an anomalous Hall response can emerge immediately, and the signs of the anomalous Hall and anomalous Nernst conductivities can be flexibly controlled by the symmetry-breaking term, thereby realizing tunable sign-reversible anomalous transport. Furthermore, the feasibility of the theoretical scheme is further verified by explicit lattice-model construction. Using first-principles calculations, we investigate the realization of crystal symmetry-controlled anomalous transport in a 2D AM material Cr$_{2}$O$_{2}$. The results indicate that Cr$_{2}$O$_{2}$ with out-of-plane Néel vectors can sequentially exhibit the AHE and quantum anomalous Hall effect (QAHE) under continuous uniaxial strain. Interestingly, the sign reversal between these two effects can be achieved by simply rotating the strain direction by C$_{4z}$ symmetry. The corresponding ANE and its sign reversal are also revealed. Our findings provide a new strategy to manipulate anomalous transport, and should have significant potential applications.

cond-mat.mes-hall↗

arXiv2025

Effect of superconductivity by Nb and V substitution in kagome CaPd5

Materials featuring kagome lattices have attracted significant research interest due to their unique geometric frustration, which gives rise to rich physical phenomena such as non-trivial topology, spin fluctuations, and superconductivity. In this work, using CaPd5 as the prototype structure, we discover and systematically investigate a new class of kagome superconductors, CaMxPd5-x (M = Nb and V) alloys. First-principles calculations confirm that these compounds are non-magnetic metals, among which four are dynamically stable: CaNb5, CaV5, CaNb2Pd3, and CaV2Pd3. CaNb5 is identified as a strong electron-phonon coupling (EPC) superconductor with the highest superconducting transition temperature (Tc) of 10.1 K, which can be further increased to 12.8 K under external pressure. In contrast, CaV5, CaNb2Pd3, and CaV2Pd3 exhibit weaker EPC and correspondingly lower Tc values. Furthermore, by applying the method of symmetry indicators, we systematically classify the topological and nodal characteristics of CaNb5, providing valuable insights for determining its superconducting pairing symmetry. Our findings demonstrate that Nb and V substitution in kagome CaPd5 provides an effective route for designing a new type of kagome superconductor with relatively high Tc. This study also offers new perspectives on topological superconductivity in kagome systems and establishes a useful guideline for discovering other superconducting materials with unique properties.

cond-mat.supr-con↗

arXiv2025

Switchable Giant Spin Injection Current in Janus Altermagnet Fe$_2$SSeO

Generating and controlling spin current in miniaturized magnetic quantum devices remains a central objective of spintronics, due to its potential to enable future energy-efficient information technologies. Among the existing magnetic phases, altermagnetism have recently emerged as a highly promising platform for spin current generation and control, going beyond ferromagnetism and antiferromagnetism. Here, we propose a symmetry-allowed spin photovoltaic effect in two-dimensional (2D) altermagnetic semiconductors that enables predictable control of giant spin injection currents. Distinct from parity-time ($\mathcal{PT}$)-antiferromagnets, Janus altermagnetic semiconductors generate not only shift current but also a unique injection current with spin momentum locked in a specific direction under linearly polarized light -- a mechanism absent in $\mathcal{PT}$-antiferromagnets. Through symmetry analysis and first-principles calculations, we identify Janus Fe$_2$SSeO as a promising candidate. Specifically, the monolayer Fe$_2$SSeO exhibits a polarization-dependent injection conductivity reaching $\sim$1,200~$μ$A/V$^{2}\!\cdot\!\hbar/2e$, and the giant spin injection current can be effectively switched by rotating the magnetization direction and engineering strains. These findings underscore the potential of 2D altermagnets in spin photovoltaics and open avenues for innovative quantum devices.

cond-mat.mtrl-sci↗