JPEG is a lossy compression technique which is the default format for all major digital cameras and estimated to be used for over 80% of all digital images on the web [1], [2]. During JPEG encoding, quantization is used to approximate frequency coefficients by a set of integers called the quantization table. We introduce OptQT as a new framework for the optimal design of JPEG quantization tables.

Nowadays, pre-training models on largescale datasets and fine-tuning models on task-specific datasets have become common paradigms, achieving impressive success in natural language processing and 2D vision. Nonetheless, the potential of this paradigm has not been fully explored in 3D vision due to the scale of the datasets. To overcome this, we propose BoostPoint, a novel pipeline that uses large-scale rendered images as 3D point cloud model inputs for pretraining and uses general 3D tasks for fine-tuning. In BoostPoint, we propose a novel learning-free image-topoint (I2P) module to transform raw pixels into required inputs. Specifically, we view pixels as unorganized points, including essential raw features (e.g., color) and positional information (e.g., coordinates). Employing simple linear iterative clustering (SLIC), the I2P module effectively groups these unorganized points into superpixels, facilitating point cloud backbone pretraining. Furthermore, we employ a modality-agnostic debiasing mechanism during pre-training to prevent negative transfer in downstream tasks. Extensive finetuning experiments show that BoostPoint provides significant improvements to 3D point cloud backbones for 3D point cloud classification and part segmentation.

The Channel Knowledge Map (CKM) is an emerging technology for enabling future integrated sensing and communication (ISAC) services that have attracted significant research interests in recent years. Current methods for constructing CKM primarily include interpolation-based, model-based, and machine learning algorithms. However, these methods are often limited by their low estimation accuracy or high computational complexity. To address these challenges, we propose a novel approach to construct CKM based on Environment Partitioning and Interpolation Weight Learning (EPIWL). The proposed method leverages channel state information from partially known locations to interpolate and estimate the channel state in the target region, thus completing the CKM construction. To reduce computational complexity, we proposed a Graph Feature Aggregation and Community Detection based Partitioning (GFA-CD-P) algorithm, which selects representative anchor points through environmental partitioning, thereby decreasing the computational load. Furthermore, we propose an interpolation weight learning scheme based on Kolmogorov–Arnold Network (KAN) and Multi-Head Cross Attention (MHCA), namely KM-IWL algorithm, which automatically learn the weight between anchor points and target points, enhancing both the efficiency and accuracy of CKM construction. The experimental results demonstrate that the proposed EPIWL approach achieves a 10 dB improvement in normalized mean squared error (NMSE) and reduces computational complexity by over 60% compared with existing schemes, showcasing robust performance and excellent generalization capability.

To date, there is a notable lack of rigorous benchmarks that assess Multimodal Large Language Models (MLLMs) within the financial domain, a field characterized by specialized financial charts and complex domain-specific expertise. To address this gap, we introduce MME-Finance, the first comprehensive bilingual multimodal benchmark tailored for financial analysis. MME-Finance comprises 4,751 meticulously curated samples, encompassing 2,274 open-ended questions, 2,000 binary-choice questions, and 477 multi-turn questions. To mitigate bias when LLMs act as judges, we also created an evaluation framework that strengthens alignment with human judgments by embedding visual context into the multimodal assessment pipeline. A comprehensive evaluation of 31 popular MLLMs has been conducted to assess their perception, reasoning, and cognitive capabilities. Gemini2.5Pro achieves highest accuracy of 79.28% and 85.71% on the open-ended questions and multi-turn questions, respectively. Among open-source models, InternVL3-78B attains 71.24 % accuracy on the open-ended question, whereas Qwen2.5-VL-72B achieves an F1 score of 88.73 % on the binary-choice question. The results indicate that state-of-the-art MLLMs demonstrate considerable overall competence, yet exhibit significant deficiencies in fine-grained visual perception and the understanding of domain-specific financial images. Source code is available at https://github.com/HiThink-Research/MME-Finance.

JPEG is a widely used format for images. Most JPEG variants are based upon a block-based DCT transformation followed by quantization and entropy coding. Redundancy at row/column level is explored in [1]. Brunsli [2] and Lepton [3], lossless JPEG repacking libraries, explore redundancy at block level.

As one of the earliest algorithms for analyzing link relations of network, Pagerank algorithm finds the most influential nodes by iterating through a loop based on a strict mathematical background and the proposed quantitative and qualitative assumptions. In order to make the PR value of every node in the network reach a smooth convergence state, Pagerank assumes that all the other nodes have a link pointing to the current node, which uses a damping factor to represent the coefficient. We believe that this consideration is not very meaningful in practical terms and have therefore considered the possibility of relational transfer between nodes to make the meaning of the factor coefficients more practical and interpretable. In this paper, by changing the Pagerank algorithm to construct a smooth distributed Markov chain model, the idea of inter-neighboring nodes is introduced to obtain a complex network relationship analysis method. At the same time, this method combined with the idea of six degrees of separation theory, so as to obtain a sequence of key nodes. By experimenting on different kinds of data sets, we find that it has better results compared with the traditional Pagerank algorithm.

MOTIVATION: Biological rhythmicity is fundamental to almost all organisms on Earth and plays a key role in health and disease. Identification of oscillating signals could lead to novel biological insights, yet its investigation is impeded by the extensive computational and statistical knowledge required to perform such analysis. RESULTS: To address this issue, we present DiscoRhythm (Discovering Rhythmicity), a user-friendly application for characterizing rhythmicity in temporal biological data. DiscoRhythm is available as a web application or an R/Bioconductor package for estimating phase, amplitude, and statistical significance using four popular approaches to rhythm detection (Cosinor, JTK Cycle, ARSER, and Lomb-Scargle). We optimized these algorithms for speed, improving their execution times up to 30-fold to enable rapid analysis of -omic-scale datasets in real-time. Informative visualizations, interactive modules for quality control, dimensionality reduction, periodicity profiling, and incorporation of experimental replicates make DiscoRhythm a thorough toolkit for analyzing rhythmicity. AVAILABILITY AND IMPLEMENTATION: The DiscoRhythm R package is available on Bioconductor (https://bioconductor.org/packages/DiscoRhythm), with source code available on GitHub (https://github.com/matthewcarlucci/DiscoRhythm) under a GPL-3 license. The web application is securely deployed over HTTPS (https://disco.camh.ca) and is freely available for use worldwide. Local instances of the DiscoRhythm web application can be created using the R package or by deploying the publicly available Docker container (https://hub.docker.com/r/mcarlucci/discorhythm). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Systems in many safety-critical application domains are subject to certification requirements. For any given system, however, it may be the case that only a subset of its functionality is safety-critical and hence subject to certification; the rest of the functionality is non-safety-critical and does not need to be certified, or is certified to lower levels of assurance. The certification-cognizant runtime scheduling of such mixed-criticality systems is considered. An algorithm called EDF-VD (for Earliest Deadline First with Virtual Deadlines) is presented: this algorithm can schedule systems for which any number of criticality levels are defined. Efficient implementations of EDF-VD, as well as associated schedulability tests for determining whether a task system can be correctly scheduled using EDF-VD, are presented. For up to 13 criticality levels, analyses of EDF-VD, based on metrics such as processor speedup factor and utilization bounds, are derived, and conditions under which EDF-VD is optimal with respect to these metrics are identified. Finally, two extensions of EDF-VD are discussed that enhance its applicability. The extensions are aimed at scheduling a wider range of task sets, while preserving the favorable worst-case resource usage guarantees of the basic algorithm.

We investigate the problem of checking class attendance by detecting, tracking and recognizing multiple student faces in classroom videos taken by instructors. Instead of recognizing each individual face independently, first, we perform multi-object tracking to associate detected faces (including false positives) into face tracklets (each tracklet contains multiple instances of the same individual with variations in pose, illumination etc.) and then we cluster the face instances in each tracklet into a small number of clusters, achieving sparse face representation with less redundancy. Then, we formulate a unified optimization problem to (a) identify false positive face tracklets; (b) link broken face tracklets belonging to the same person due to long occlusion; and (c) recognize the group of faces simultaneously with spatial and temporal context constraints in the video. We test the proposed method on Honda/UCSD database and real classroom scenarios. The high recognition performance achieved by recognizing a group of multi-instance tracklets simultaneously demonstrates that multi-face recognition is more accurate than recognizing each individual face independently.

Systems in many safety-critical application domains are subject to certification requirements. For any given system, however, it may be the case that only a subset of its functionality is safety-critical and hence subject to certification, the rest of the functionality is non safety critical and does not need to be certified, or is certified to a lower level of assurance. An algorithm called EDF-VD (for Earliest Deadline First with Virtual Deadlines) is described for the scheduling of such mixed-criticality task systems. Analyses of EDF-VD significantly superior to previously-known ones are presented, based on metrics such as processor speedup factor (EDF-VD is proved to be optimal with respect to this metric) and utilization bounds.

The scheduling of mixed-criticality implicit-deadline sporadic task systems on identical multiprocessor platforms is considered, when inter-processor migration is permitted. A scheduling algorithm is derived and proved correct, and its properties investigated. Theoretical analysis (in the form of both a speedup factor and sufficient schedulability conditions) as well as extensive simulation experiments serve to demonstrate its effectiveness.

Many safety-critical embedded systems are subject to certification requirements; some systems may be required to meet multiple sets of certification requirements, from different certification authorities. Certification requirements in such "mixed-criticality” systems give rise to interesting scheduling problems, that cannot be satisfactorily addressed using techniques from conventional scheduling theory. In this paper, we study a formal model for representing such mixed-criticality workloads. We demonstrate first the intractability of determining whether a system specified in this model can be scheduled to meet all its certification requirements, even for systems subject to merely two sets of certification requirements. Then we quantify, via the metric of processor speedup factor, the effectiveness of two techniques, reservation-based scheduling and priority-based scheduling, that are widely used in scheduling such mixed-criticality systems, showing that the latter of the two is superior to the former. We also show that the speedup factors we obtain are tight for these two techniques.

Many safety-critical embedded systems are subject to certification requirements. However, only a subset of the functionality of the system may be safety-critical and hence subject to certification; the rest of the functionality is non safety-critical and does not need to be certified. Certification requirements in such mixed-criticality systems give rise to some interesting scheduling problems, that cannot be satisfactorily addressed using techniques from conventional scheduling theory. In prior work, we have proposed a priority-based algorithm for scheduling such mixed-criticality systems on preemptive uniprocessor platforms. In this paper, we derive a sufficient schedulability condition for efficiently determining whether a given mixed-criticality system can be successfully scheduled by this algorithm. We show that this algorithm (and the associated schedulability test) is strictly superior to prior algorithms that have been used for scheduling mixed-criticality systems needing certification.

Many safety-critical embedded systems are subject to certification requirements; some systems may be required to meet multiple sets of certification requirements, from different certification authorities. Certification requirements in such "mixed-criticality" systems give rise to some interesting scheduling problems, that cannot be satisfactorily addressed using techniques from conventional scheduling theory. In this paper, we propose a formal model for representing such mixed-criticality workloads. We demonstrate the intractability of determining whether a system specified in this model can be scheduled to meet all its certification requirements. For dual-criticality systems - systems subject to two sets of certification requirements - we quantify, via the metric of processor speedup factor, the effectiveness of 2 techniques (reservation-based scheduling and priority-based scheduling) that are widely used in scheduling such mixed-criticality systems.

Many safety-critical embedded systems are subject to certification requirements. However, only a subset of the functionality of the system may be safety-critical and hence subject to certification, the rest of the functionality is non safety-critical and does not need to be certified. Certification requirements in such "mixed-criticality" systems give rise to some interesting scheduling problems, that cannot be satisfactorily addressed using techniques from conventional scheduling theory. In prior work, we have studied the scheduling and analysis of mixed criticality systems that are specified as finite collections of jobs executing on a single shared preemptive processor. In this paper, we consider mixed criticality systems that are comprised of finite collections of recurrent tasks, specified using a mixed-criticality generalization of the widely-used sporadic tasks model. We design a priority-based algorithm for scheduling such systems, derive an algorithm for computing priorities, and obtain a sufficient schedulability condition for efficiently determining whether a given mixed-criticality system can be successfully scheduled by this algorithm.