While spot instances offer a cost-effective alternative to on-demand cloud resources, they introduce reliability challenges for latency-sensitive microservices due to preemption risks and unpredictable provisioning delays. Conventional resource management systems, which often rely on assumptions of immediate instance availability, fail to account for these operational realities—resulting in increased risk of SLO violations when deployed in spot-based environments.

Creating high-fidelity, coherent long videos is a sought-after aspiration. While recent video diffusion models have shown promising potential, they still grapple with spatiotemporal inconsistencies and high computational resource demands. We propose Global-Local Collaborative Diffusion (GLC-Diffusion), a tuning-free method for long video generation. It models the long video denoising process by establishing denoising trajectories through Global-Local Collaborative Denoising (GLCD) to ensure overall content consistency and temporal coherence between frames. Additionally, we introduce a Noise Reinitialization strategy which combines local noise shuffling with frequency fusion to improve global content consistency and visual diversity. Further, we propose a Video Motion Consistency Refinement (VMCR) module that computes the gradient of pixel-wise and frequency-wise losses to enhance visual consistency and temporal smoothness. Extensive experiments, including quantitative and qualitative evaluations on videos of varying lengths (e.g., 3× and 6× longer), demonstrate that our method effectively integrates with existing video diffusion models, producing coherent, high-fidelity long videos superior to previous approaches.

The goal of FedCSIS 2025 Challenge is to build a model to predict the difficulty (measured as Lichess rating) of given chess puzzles.To address this task, we propose a three-stage joint visual-statistical framework for predicting Glicko-based difficulty ratings.In the first stage, a convolutional model based on MobileNetV2 integrates FEN-rendered board images with structured features, including engine-predicted success probabilities, move count, and piece counts, to generate baseline predictions.The second stage employs LightGBM to perform residual refinement, explicitly learning the residual errors of the baseline predictions to correct systematic biases, particularly for extreme difficulty levels.Finally, a domain-informed refinement adjusts the outputs toward interpretable difficulty estimates derived from failure probability distributions and rating-bucket inflection points.Our model ranked 9th in the challenge.Experimental results show that residual refinement and domain-informed adjustment significantly reduce mean squared error compared to the baseline visual-statistical model.

With the explosive growth of multimedia data storage, multimodal learning is an inevitable trend for Information Extraction (IE). However, the noise and irrelevance of web- crawled samples cause adverse degradation in each modality. Additionally, previous researches inadequately address the above issue, and the potential of cross-modal fusion remains underexplored. We propose a strengthened alignment module, using a generative text augmentation submodule to reduce noise contamination and emphasize incorporating visual features into texts. We further propose a fusion adapter utilizing a soft-prompt structure for profound fusion. To activate logical capabilities, we apply prompts with a multi-turn dialogue. For the Multimodal Relation Extraction task over the MNRE dataset, our method exceeds the previous SOTA model with a 7% increase in F1-score. It has superior generalization for other multimodal IE tasks, achieving SOTA on Named Entity Recognition over Twitter-15/17 datasets and on Event Extraction over M2E2dataset.

We introduce a model named DreamLight for universal image relighting in this work, which can seamlessly composite subjects into a new background while maintaining aesthetic uniformity in terms of lighting and color tone. The background can be specified by natural images (image-based relighting) or generated from unlimited text prompts (text-based relighting). Existing studies primarily focus on image-based relighting, while with scant exploration into text-based scenarios. Some works employ intricate disentanglement pipeline designs relying on environment maps to provide relevant information, which grapples with the expensive data cost required for intrinsic decomposition and light source. Other methods take this task as an image translation problem and perform pixel-level transformation with autoencoder architecture. While these methods have achieved decent harmonization effects, they struggle to generate realistic and natural light interaction effects between the foreground and background. To alleviate these challenges, we reorganize the input data into a unified format and leverage the semantic prior provided by the pretrained diffusion model to facilitate the generation of natural results. Moreover, we propose a Position-Guided Light Adapter (PGLA) that condenses light information from different directions in the background into designed light query embeddings, and modulates the foreground with direction-biased masked attention. In addition, we present a post-processing module named Spectral Foreground Fixer (SFF) to adaptively reorganize different frequency components of subject and relighted background, which helps enhance the consistency of foreground appearance. Extensive comparisons and user study demonstrate that our DreamLight achieves remarkable relighting performance.

Continual learning requires the model to continually capture novel information without forgetting prior knowledge. Nonetheless, existing studies predominantly address the catastrophic forgetting, often neglecting enhancements in model robustness. Consequently, these methodologies fall short in real-time applications, such as autonomous driving, where data samples frequently exhibit noise due to environmental and lighting variations, thereby impairing model efficacy and causing safety issues. In this paper, we address robustness in continual learning systems by introducing an innovative approach, the Dynamic Siamese Expansion Framework (DSEF) that employs a Siamese backbone architecture, comprising static and dynamic components, to facilitate the learning of both global and local representations over time. Specifically, the proposed framework dynamically generates a lightweight expert for each novel task, leveraging the Siamese backbone to enable rapid adaptation. A novel Robust Dynamic Representation Optimization (RDRO) approach is proposed to incrementally update the dynamic backbone by maintaining all previously acquired representations and prediction patterns of historical experts, thereby fostering new task learning without inducing detrimental knowledge transfer. Additionally, we propose a novel Robust Feature Fusion (RFF) approach to incrementally amalgamate robust representations from all historical experts into the expert construction process. A novel mutual information-based technique is employed to derive adaptive weights for feature fusion by assessing the knowledge relevance between historical experts and the new task, thus maximizing positive knowledge transfer effects. A comprehensive experimental evaluation, benchmarking our approach against established baselines, demonstrates that our method achieves state-of-the-art performance even under adversarial attacks.

Image super-resolution (SR) aims to generate a high-resolution (HR) image from a low-resolution (LR) input. Traditionally, full-reference image quality assessment (FR-IQA) models have been widely used to evaluate the perceptual quality of super-resolved images, relying on pristine reference images as the gold standard. However, in real-world SR applications, such reference images are often unavailable, posing challenges for the use of FR-IQA. While blind image quality assessment (BIQA) models can assess the perceptual quality of super-resolved images without requiring a reference, there remains a lack of comprehensive studies evaluating the effectiveness of existing BIQA models for real-world SR tasks. This dilemma can largely be attributed to the high cost of subjective testing required to collect sufficient human quality annotations, which in turn hinders the development of effective SR-IQA models. In this study, we tackle this challenge with a data-efficient approach. We first generate super-resolved images from LR inputs using state-of-the-art real-world SR methods. Then, we use the maximum differentiation competition (MAD) to select a diverse set of images for subjective testing, allowing us to efficiently gather human preferences and assess the alignment between BIQA model predictions and human judgments. The resulting global ranking of SR methods not only indicates the relative performance of recent real-world SR models, but also gives us an opportunity to develop a new BIQA model tailored for real-world SR-IQA. By utilizing the global rankings of SR algorithms as prior knowledge, we can refine pretrained BIQA models using vast amounts of super-resolved images without any supervisory signal. Experimental results show that our approach substantially enhances IQA performance for real-world SR while preserving robust predictive accuracy across various distortion scenarios. The dataset and the code are available at https://github.com/cschenjunlin/SR-IQA-SMC25.

Gas sensor drift has consistently been a bottleneck in the progression of electronic noses (E-noses) systems. In contrast to target-domain-adaptive anti-drift classification algorithms , target-domain-free methods exhibit independence from target domain information, thereby possessing broader applicability. Support vector machine (SVM), as a popular target-domain-free classifier, is widely employed to solve the E-noses drift problem. However, SVM-based methods have some inherent flaws in drift resistance performance: (1) The objective of maximizing minimum margin, rendering it highly sensitive to disturbances; (2) The utilization of hinge loss, resulting in poor robustness; (3) The absence of sample reliability measurement, contributing to noise susceptibility. To tackle these problems, we present a novel twin-center intuitionistic fuzzy large margin classifier with unified pinball loss (TC-IFUPLMC), which effectively improves the anti-drift performance in E-noses system. Specifically, the model achieves strong disturbance resistance by adopting marginal mean and variance as the optimization objective. Furthermore, the unified pinball loss is utilized to measure the distance between two categories of sample sets, which essentially further enhances the robustness. Additionally, a novel twin-center based IF function is used to obtain the confidence level of each sample, which further weakens the noise susceptibility. Comparative experiments based on different sensor drift datasets demonstrate the effectiveness of the proposed model in improving the anti-drift performance. The high stability of TC-IFUPLMC is also substantiated by the parameter sensitivity analysis experiments.

Ride-hailing has become a popular service in recent years. For each ride-hailing request, after the platform determines the fare for the passenger and the commission for the driver, it is not uncommon for the platform to set aside a promotional budget and give instant incentives to both sides, that is, a discount to the passenger and a bonus to the driver, to further improve the match between the two sides. Although there is a proliferation of studies on the determination of the fare and the commission in ride-hailing services, they cannot address the instant incentive problem because their approaches do not deal with budget constraints. In this research, we investigate this two-sided instant incentive problem under a shared promotional budget, which is new to the literature. We formulate this problem as a binary integer linear programming problem, whose goal is to find the optimal incentives for both sides given predicted trip completion probabilities. We first assume that the predicted trip completion probabilities are accurate and develop a Lagrangian-dual-based approach to decompose the problem into a series of subproblems that can be efficiently solved. We then proceed to accommodate the inaccuracy in the predictions and develop a robust instant incentive optimization approach that exploits the prediction error reflected by historical data. We conduct numerical experiments using real data in the city of Nanjing from a leading ride-hailing platform in China. Results show that compared to the baseline approach: (i) Before we account for prediction inaccuracy, our solution approach can improve the number of completed requests by at most 8.30% with a decision error of 8.31%; and (ii) After we account for prediction inaccuracy, our solution approach can improve the number of completed requests by at most 9.81% while reducing the decision error to 7.03%.

Protection of intellectual property (IP) is increasingly critical for IP vendors in the semiconductor industry. However, advanced reverse engineering techniques can physically disassemble the chip and derive the IPs at a much lower cost than the value of IP design that chips carry. This invasive hardware attack—obtaining information from IC chips—always violates the IP rights of vendors. The intent of this article is to present a chip-level reverse engineering resilient design technique. In the proposed technique, transformable interconnects enable an IC chip to maintain functioning in normal use and to transform its physical structure into another pattern when exposed to invasive attacks. The newly created pattern will significantly increase the difficulty of reverse engineering. Furthermore, to improve the effectiveness of the proposed technique, a systematic design method is developed targeting integrated circuits with multiple design constraints. Simulations have been conducted to demonstrate the capability of the proposed technique, which generates extremely large complexity for reverse engineering with manageable overhead.

Off-highway dump truck contains amounts of potential energy and kinetic energy. During the braking process, these two parts are consumed in the form of heat on the braking resistance, and the energy consumption will be about 24 percent of total energy consumption of the Truck. Hence, it is necessary to recover the braking energy. This paper, aiming at the characteristics of electric drive system in off-highway dump truck, proposes a new type of DC/DC main circuit which is a series connection topology that includes super-capacitor and battery hybrid energy storage, the DC bus's rate voltage is 1800v. The bottom controller is designed based on DSP + FPGA, which realizes the efficient control and protection of the system, and completes the recovery and utilization function of the braking energy in off-highway dump trucks. In the host computer, the monitoring system designed by utilizing of C#.net language, and the relay protection 103 communication protocol has been used to improve system stability. In this paper, the dynamic test platform of braking energy recovery system has been completed, after the dynamic test, the system has been install in Komatsu 730E dump truck of Pingshuo Anjialing surface coal mine to complete the field operation. The experimental result shows the effectiveness and reliability of the system.

The reverse engineering (RE) of electronic chips and systems can be used with honest and dishonest intentions. To inhibit RE for those with dishonest intentions (e.g., piracy and counterfeiting), it is important that the community is aware of the state-of-the-art capabilities available to attackers today. In this article, we will be presenting a survey of RE and anti-RE techniques on the chip, board, and system levels. We also highlight the current challenges and limitations of anti-RE and the research needed to overcome them. This survey should be of interest to both governmental and industrial bodies whose critical systems and intellectual property (IP) require protection from foreign enemies and counterfeiters who possess advanced RE capabilities.

This brief presents a system design technique for improving the energy utilization of radio frequency (RF) circuits powered by renewable energy sources. Different from conventional systems, the operation of self-powered RF circuits is largely constrained by two factors: time-varying channel conditions and nondeterministic renewable energy levels. The proposed technique dynamically adjusts the modulation scheme to deal with these two factors in a coherent manner. This is the effective way to maximize the data rate of RF circuits while maintaining the required performance under unstable energy supplies. A detailed VLSI implementation is developed with negligible energy overheads. Simulation results demonstrate that the proposed technique outperforms conventional RF circuits based on the fixed modulation scheme under variable channel and energy conditions.

A new link and energy adaptive UWB-based sensing system is proposed to maximize the detection time coverage of UWB-based pulse radar in embedded sensing applications powered by renewable energy. By jointly considering the link information between the transmitter and receiver of the UWB radar as well as the non-deterministic characteristics of the renewable energy, the proposed system dynamically adjusts the pulse repetition frequency of the UWB radar to enhance the energy efficiency. Simulation results demonstrate that the proposed system can significantly improve the average detection time coverage as compared with the conventional system. The proposed technique is also insensitive to battery issues such as limited battery capacity.