Advances in brain science and computer technology in the past decade have led to exciting developments in brain–computer interface (BCI), thereby making BCI a top research area in applied science. The renaissance of BCI opens new methods of neurorehabilitation for physically disabled people (e.g. paralyzed patients and amputees) and patients with brain injuries (e.g. stroke patients). Recent technological advances such as wireless recording, machine learning analysis, and real-time temporal resolution have increased interest in electroencephalographic (EEG) based BCI approaches. Many BCI studies have focused on decoding EEG signals associated with whole-body kinematics/kinetics, motor imagery, and various senses. Thus, there is a need to understand the various experimental paradigms used in EEG-based BCI systems. Moreover, given that there are many available options, it is essential to choose the most appropriate BCI application to properly manipulate a neuroprosthetic or neurorehabilitation device. The current review evaluates EEG-based BCI paradigms regarding their advantages and disadvantages from a variety of perspectives. For each paradigm, various EEG decoding algorithms and classification methods are evaluated. The applications of these paradigms with targeted patients are summarized. Finally, potential problems with EEG-based BCI systems are discussed, and possible solutions are proposed.
Abstract Vehicular Ad Hoc Networks (VANETs) are wireless networks established between vehicles and their surrounding infrastructure, enabling the exchange of information. Consequently, many applications that can enhance passengers’ safety and traffic flow are built upon this information. However, malicious nodes can manipulate the exchanged data to attack other nodes and disrupt the network’s normal behavior. For example, if an attacker broadcasts a falsified location for a vehicle, the functionality of applications that rely on accurate location sharing will be compromised, potentially leading to deadly accidents. Although numerous Misbehavior Detection Schemes (MDSs) have been proposed to detect position falsification attacks, their effectiveness remains limited for certain attack types, raising concerns given the safety-critical nature of VANET applications. This paper proposes a machine learning-based method for detecting position falsification attacks. The proposed approach evaluates four machine-learning algorithms using three feature vectors (FV1, FV2, and FV3) composed of selected and derived features extracted from Basic Safety Messages (BSMs), in addition to a novel confidence-based Received Signal Strength Indicator feature, termed RSSIConf. The RSSIConf feature assesses the reliability of a sender’s claimed position by comparing the measured RSSI with confidence intervals corresponding to the claimed sender–receiver distance. Experimental results show that the Random Forest classifier trained with FV2 features achieves the best overall performance, outperforming existing approaches with improvements ranging from 0.76% to 13.26% in accuracy and from 0.74% to 12.71% in F1-score across different position spoofing attack types. These improvements enhance the reliability of misbehavior detection and contribute to safer and more trustworthy VANET communications.
Abhiraam Eranti, Yogesh Tewari, Rafael Palacios
et al.
Deep-learning methods have boosted the analytical power of Raman spectroscopy, yet they still require large, task-specific, labeled datasets and often fail to transfer across application domains. The study explores pre-trained encoders as a solution. Pre-trained encoders have significantly impacted Natural Language Processing and Computer Vision with their ability to learn transferable representations that can be applied to a variety of datasets, significantly reducing the amount of time and data required to create capable models. The following work puts forward a new approach that applies these benefits to Raman Spectroscopy. The proposed approach, RSPTE (Raman Spectroscopy Pre-Trained Encoder), is designed to learn generalizable spectral representations without labels. RSPTE employs a novel domain adaptation strategy using unsupervised Barlow Twins decorrelation objectives to learn fundamental spectral patterns from multi-domain Raman Spectroscopy datasets containing samples from medicine, biology, and mineralogy. Transferability is demonstrated through evaluation on several models created by fine-tuning RSPTE for different application domains: Medicine (detection of Melanoma and COVID), Biology (Pathogen Identification), and Agriculture. As an example, using only 20% of the dataset, models trained with RSPTE achieve accuracies ranging 50%–86% (depending on the dataset used) while without RSPTE the range is 9%–57%. Using the full dataset, accuracies with RSPTE range 81%–97%, and without pre-training 51%–97%. Current methods and state-of-the-art models in Raman Spectroscopy are compared to RSPTE for context, and RSPTE exhibits competitive results, especially with less data as well. These results provide evidence that the proposed RSPTE model can effectively learn and transfer generalizable spectral features across different domains, achieving accurate results with less data in less time (both data collection time and training time).
LSST Dark Energy Science Collaboration, Eric Aubourg, Camille Avestruz
et al.
The Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST) will produce unprecedented volumes of heterogeneous astronomical data (images, catalogs, and alerts) that challenge traditional analysis pipelines. The LSST Dark Energy Science Collaboration (DESC) aims to derive robust constraints on dark energy and dark matter from these data, requiring methods that are statistically powerful, scalable, and operationally reliable. Artificial intelligence and machine learning (AI/ML) are already embedded across DESC science workflows, from photometric redshifts and transient classification to weak lensing inference and cosmological simulations. Yet their utility for precision cosmology hinges on trustworthy uncertainty quantification, robustness to covariate shift and model misspecification, and reproducible integration within scientific pipelines. This white paper surveys the current landscape of AI/ML across DESC's primary cosmological probes and cross-cutting analyses, revealing that the same core methodologies and fundamental challenges recur across disparate science cases. Since progress on these cross-cutting challenges would benefit multiple probes simultaneously, we identify key methodological research priorities, including Bayesian inference at scale, physics-informed methods, validation frameworks, and active learning for discovery. With an eye on emerging techniques, we also explore the potential of the latest foundation model methodologies and LLM-driven agentic AI systems to reshape DESC workflows, provided their deployment is coupled with rigorous evaluation and governance. Finally, we discuss critical software, computing, data infrastructure, and human capital requirements for the successful deployment of these new methodologies, and consider associated risks and opportunities for broader coordination with external actors.
Home healthcare has become more and more central in the last decades, due to the advantages it can bring to both healthcare institutions and patients. Planning activities in this context, however, presents significant challenges related to route planning and mutual synchronization of caregivers.In this paper we propose a new compact model for the combined optimization of scheduling (of the activities) and routing (of the caregivers) characterized by fewer variables and constraints when compared with the models previously available in the literature. The new model is solved by a constraint programming solver and compared experimentally with the exact and metaheuristic approaches available in the literature on the common datasets adopted by the community. The results show that the new model provides improved lower bounds for the vast majority of the instances, while producing at the same time high quality heuristic solutions, comparable to those of tailored metaheuristics, for small/medium size instances.
Abstract Mobile ad hoc networks (MANETs) facilitate data communication across multiple nodes and hop stations, characterized by their dynamic topology. This inherent flexibility, however, makes MANETs vulnerable to various security threats, notably blackhole and wormhole attacks, where malicious nodes can intercept and manipulate data. This study investigates the security vulnerabilities of MANETs, particularly against blackhole, Sybil, and wormhole attacks, and introduces the Advanced Blockchain Dynamic Source Routing (ABCD) algorithm to address these challenges. Motivated by the need for robust and decentralized security solutions in MANETs, the proposed algorithm integrates blockchain technology and homomorphic encryption to secure data communication without intermediate decryption. The ABCD algorithm leverages Dijkstra’s algorithm for optimal routing and employs a tamper-proof, decentralized data storage approach. Comparative analysis under attack scenarios reveals that the ABCD algorithm outperforms the standard DSR protocol across multiple quality of service metrics, demonstrating a significant improvement in MANET security over equivalent studies. The packet delivery rate is also improved from 81 to 92% using the modified ABCD algorithm.
This EPTCS volume contains the post-proceedings of the Twelfth International Workshop on Fixed Points in Computer Science, presenting a selection of the works presented during the workshop that took place in Naples (Italy) on the 19th and 20th of February 2024 as a satellite of the International Conference on Computer Science Logic (CSL 2024).
Emanuele Aiello, Mirko Agarla, Diego Valsesia
et al.
Large-scale self-supervised pretraining of deep learning models is known to be critical in several fields, such as language processing, where its has led to significant breakthroughs. Indeed, it is often more impactful than architectural designs. However, the use of self-supervised pretraining lags behind in several domains, such as hyperspectral images, due to data scarcity. This paper addresses the challenge of data scarcity in the development of methods for spatial super-resolution of hyperspectral images (HSI-SR). We show that state-of-the-art HSI-SR methods are severely bottlenecked by the small paired datasets that are publicly available, also leading to unreliable assessment of the architectural merits of the models. We propose to capitalize on the abundance of high resolution (HR) RGB images to develop a self-supervised pretraining approach that significantly improves the quality of HSI-SR models. In particular, we leverage advances in spectral reconstruction methods to create a vast dataset with high spatial resolution and plausible spectra from RGB images, to be used for pretraining HSI-SR methods. Experimental results, conducted across multiple datasets, report large gains for state-of-the-art HSI-SR methods when pretrained according to the proposed procedure, and also highlight the unreliability of ranking methods when training on small datasets.
Abstract The growing focus on enhancing color quality in liquid crystal displays (LCDs) and organic light-emitting diodes (OLEDs) has spurred significant advancements in color-conversion materials. Furthermore, color conversion is also important for the development and commercialization of Micro-LEDs. This article provides a comprehensive review of different types of color conversion methods as well as different types of color conversion materials. We summarize the current status of patterning process, and discuss key strategies to enhance display performance. Finally, we speculate on the future prospects and roles that color conversion will play in ultra-high-definition micro- and projection displays.
We survey two key problems-Multi-Winner Determination and Hedonic Games in Computational Social Choice, with a special focus on their parameterized complexity, and propose some research challenges in the field.
In order to generalise semi-invariant Riemannian maps, Sahin first introduced the idea of “Generic Riemannian maps”. We extend the idea of generic Riemannian maps to the case in which the total manifold is a nearly Kaehler manifold. We study the integrability conditions for the horizontal distribution although vertical distribution is always integrable. We also study the geometry of foliations of two distributions and obtain the necessary and sufficient condition for generic Riemannian maps to be totally geodesic. Additionally, we study the generic Riemannian map with umbilical fibers.
Andi Kurniawan Nugroho, Dinar Mutiara Kusumo Nugraheni, Terawan Agus Putranto
et al.
When the blood flow to the arteries in brain is blocked, its known as Ischemic stroke or blockage stroke. Ischemic stroke can occur due to the formation of blood clots in other parts of the body. Plaque buildup in arteries, on the other hand, can cause blockages because if it ruptures, it can form blood clots. The b-1000 Diffusion Weighted (DW) Magnetic Resonance Imaging (MRI) image was used in a general examination to obtain an image of the part of the brain that had a stroke. In this study, classifications used several variations of layer convolution to obtain high accuracy and high computational consumption using b-1000 Diffusion Weighted (DW) MR in ischemic stroke types: acute, sub-acute and chronic. Ischemic stroke was classified using five variants of the Convolutional Neural Network (CNN) architectural design, i.e., CNN1–CNN5. The test results show that the CNN5 architectural design provides the best ischemic stroke classification compared to other architectural designs tested, with an accuracy of 99.861%, precision 99.862%, recall 99.861, and F1-score 99.861%.
Computer Science education has been evolving over the years to reflect applied realities. Until about a decade ago, theory of computation, algorithm design and system software dominated the curricula. Most courses were considered core and were hence mandatory; the programme structure did not allow much of a choice or variety. This column analyses why this changed Circa 2010 when elective subjects across scores of topics become part of mainstream education to reflect the on-going lateral acceleration of Computer Science. Fundamental discoveries in artificial intelligence, machine learning, virtualization and cloud computing are several decades old. Many core theories in data science are centuries old. Yet their leverage exploded only after Circa 2010, when the stage got set for people-centric problem solving in massive scale. This was due in part to the rush of innovative real-world applications that reached the common man through the ubiquitous smart phone. AI/ML modules arrived in popular programming languages; they could be used to build and train models on powerful - yet affordable - compute on public clouds reachable through high-speed Internet connectivity. Academia responded by adapting Computer Science curricula to align it with the changing technology landscape. The goal of this experiential piece is to trigger a lively discussion on the past and future of Computer Science education.