Pure OAI Repository
Not a member yet
421387 research outputs found
Sort by
Designing Complexity:Virtual Growth Algorithms for Non-Periodic Bioinspired Material Architectures
No full textNon-periodic architectures observed in biological materials have been studied for their outstanding mechanical properties, such as high stiffness-to-weight ratio, energy absorption, and capacity to redistribute applied stresses. Taking inspiration from these architectures to generate engineering materials is still an open challenge. Irregular structures are challenging to model and fabricate using conventional design methods, yet they offer unique opportunities for creating functional and efficient material systems. One emerging approach is the use of tile-based computational algorithms that simulate growth processes to more effectively capture the structural irregularity of these materials. In this work, we discuss biological irregular architectures and the recent developments in computational tiling algorithms, with a particular emphasis on algorithms of virtual growth. These algorithms rely on simple tiles and a set of modifiable connection rules to generate countless complex, non-periodic structures with precise control over their geometry and topology. Recent studies have shown that material systems synthesized using tile-based designs inspired by non-periodic biological architectures can exhibit favorable properties, including enhanced impact absorbance and stress modulation. Despite this progress, integration of structure and function remains limited, highlighting the need for hybrid approaches that incorporate performance-based feedback and optimization strategies. In this context, these tools are uniquely positioned not only as generators of designs of increasing structural complexity for advanced architected materials but also as promising models for investigating fundamental questions in developmental biology.</p
Novel Wavelength Distribution Nodes Based on Photonic Integrated EDWA and SOA-Based WSS for Metro-Access Networks
No full textThis paper presents an experimental investigation of a scalable, reconfigurable wavelength distribution node (WDN) architecture for next-generation optical metro-access networks, capable of dynamically routing coherent Wavelength Division Multiplexing (WDM) channels at multi-Tbps levels, with 800 Gbps per channel. The proposed architecture leverages novel photonic integrated erbium-doped waveguide amplifiers (EDWAs) combined with Semiconductor Optical Amplifier (SOA)-based 1×N Wavelength Selective Switches (WSS), facilitating dynamic add/drop capabilities essential for flexible metro-access deployments. Three experiments validated the architecture’s transmission performance and scalability. The first experiment demonstrated successful add/drop and drop&continue dynamic switching of six 800 Gbps WDM channels, achieving an aggregate throughput of 4.8 Tbps with an average Optical Signal-to-Noise Ratio (OSNR) degradation of only 0.55 dB after one node, and 2.4 dB after cascading two nodes over a 55.2 km link. The second experiment confirmed error-free transmission of eight 800 Gbps WDM channels, recording a maximum OSNR penalty variation among channels of 3.56 dB. The third experiment evaluated scalability using a 1×4 for WDN1, 1×16 WSS for WDN2 configuration, achieving error-free transmission serving up to 64 access points at 800 Gbps per channel, with a cumulative OSNR penalty of 7.44 dB after cascading two WDN nodes. These results demonstrate the feasibility and high performance of integrated photonic solutions in meeting future 6G network demands.</p
Graph-Theoretic Bézier Curve Optimization over Safe Corridors for Safe and Smooth Motion Planning
No full textAs a parametric motion representation, Bézier curves have significant applications in polynomial trajectory optimization for safe and smooth motion planning of various robotic systems, including flying drones, autonomous vehicles, and robotic manipulators. An essential component of Bézier curve optimization is the optimization objective, as it significantly influences the resulting robot motion. Standard physical optimization objectives, such as minimizing total velocity, acceleration, jerk, and snap, are known to yield quadratic optimization of Bézier curve control points. In this paper, we present a unifying graph-theoretic perspective for defining and understanding Bézier curve optimization objectives using a consensus distance of Bézier control points derived based on their interaction graph Laplacian. In addition to demonstrating how standard physical optimization objectives define a consensus distance between Bézier control points, we also introduce geometric and statistical optimization objectives as alternative consensus distances, constructed using finite differencing and differential variance. To compare these optimization objectives, we apply Bézier curve optimization over convex polygonal safe corridors automatically constructed around a maximal-clearance minimal-length reference path. We provide an explicit analytical formulation for quadratic optimization of Bézier curves using Bézier matrix operations. We conclude that the norm and variance of the finite differences of Bézier control points lead to simpler and more intuitive interaction graphs and optimization objectives compared to Bézier derivative norms, despite having similar robot motion profiles
Exploring Polydimethylsiloxane Coating Strategies to Enhance Liquid Repellency of Carbonaceous Porous Media
No full textCarbonaceous porous diffusion media play a critical role in electrochemical technologies sustaining multiphase (gas and liquid) flows, including low-temperature fuel cells and CO2 electrolyzers. To resist liquid intrusion and preserve gas pathways, these materials are typically hydrophobized with polytetrafluoroethylene dispersions. However, the application of dispersion-based coatings limits uniformity and performance, while the persistence and toxicity associated with per- and polyfluoroalkyl compounds have prompted regulatory scrutiny. Here, we investigate polydimethylsiloxane (PDMS), a fluorine-free, low-surface-energy, and environmentally benign polymer, as a viable alternative. We assess four PDMS coating application strategies: (1) dip-coating, vapor deposition in (2) oxygen and (3) nitrogen atmospheres, and (4) electrografting of an amine-functionalized derivative. We perform spectroscopic, microscopic, wetting, and electrochemical double-layer measurements to correlate the surface chemical composition and morphology with the resulting wettability. Vapor deposition in an oxygen environment produces a superhydrophobic rough microstructure; however, the weak adhesion of the coating and substrate oxidation result in severe flooding when the substrate is exposed to an alkaline liquid flow. Notably, electrografting and vapor deposition in nitrogen yield thin uniform coatings with an ethanol–potassium hydroxide solution repellency in flow comparable to the polytetrafluoroethylene baseline. On the contrary, dip-coated samples feature thick unevenly distributed layers exhibiting poor liquid infiltration resistance. Under prolonged flow in the presence of 10 wt % ethanol, the desorption of the physisorbed layer vapor deposited in nitrogen led to a continuous increase in wetting, resulting in complete electrode flooding within 65 h. In contrast, the electrografted substrates showed enhanced durability under liquid flows with various compositions, retaining repellency for the same time scale comparably with the PTFE baseline, highlighting the importance of covalently attached hydrophobic coatings for long-term operation. These findings demonstrate the potential of PDMS-based coatings as sustainable fluorine-free alternatives for hydrophobizing carbonaceous porous media and provide practical guidelines for engineering the wetting behavior and coating stability.</p
Zero-shot load forecasting for integrated energy systems:A large language model-based framework with multi-task learning
No full textThe growing penetration of renewable energy sources in power systems has increased the complexity and uncertainty of load forecasting, especially for integrated energy systems with multiple energy carriers. Traditional forecasting methods heavily rely on historical data and demonstrate limited transferability across different scenarios, creating significant challenges for emerging applications in smart grids and energy internet. This paper proposes the TSLLM-Load Forecasting Mechanism, a novel zero-shot load forecasting framework based on large language models (LLMs) to address these challenges. The framework consists of three key components: a data preprocessing module that handles multi-source energy load data, a time series prompt generation module that bridges the semantic gap between energy data and LLMs through multi-task learning and similarity alignment, and a prediction module that leverages pre-trained LLMs for accurate forecasting. We validate the framework's effectiveness on real-world datasets comprising load profiles from Australian solar-powered households. In conventional testing, our method achieves competitive performance, ranking third among nine methods. In zero-shot prediction, our framework demonstrates 10.8 % MSE improvement and 12.5 % MAE improvement compared to Informer, ranking second overall after TimeMixer. Most significantly, few-shot learning experiments reveal exceptional capabilities under extreme data constraints, with our method achieving optimal performance when trained with only 1 % of available data, representing 40.8 % MSE improvement compared to the best conventional method and 78.9 % improvement compared to existing LLM-based approaches. Large-scale transferability analysis shows our model outperforms baselines for 82 % of households when trained with minimal data from a single household. These results demonstrate the framework's potential for accurate and transferable load forecasting in integrated energy systems, particularly beneficial for renewable energy integration and smart grid applications where historical data availability is limited.</p
Synthetic Data in MR Spectroscopy:Current Practices, Applications, and Considerations
No full textThe use of synthetic data has emerged as an essential tool in Magnetic Resonance Spectroscopy (MRS) research and applications, providing advantages for optimization of acquisition, software validation, deep learning applications, and enhanced reproducibility. Importantly, synthetic data addresses challenges of limited training data availability, particularly for clinical populations, and offers controlled solutions for investigating uncertainties and unexplained variance with in vivo data. This work provides a review and evaluation of current practices in the use and generation of synthetic data within the MRS field. Conducted by the MRS Synthetic Data Working Group under the Code & Data Sharing Committee of the MRS Study Group of the International Society for Magnetic Resonance in Medicine (ISMRM), this manuscript encompasses existing literature, supplemented by collective experience and in-house methodologies
An AI-based approach accelerates the discovery of protein–protein interaction modulators targeting NCS-1
No full textDrug repurposing is an efficient strategy to accelerate the identification of therapeutic compounds by finding new uses for existing drugs. Here, we leveraged artificial intelligence (AI) to optimize this process and discover protein–protein interaction (PPI) modulators targeting the Neuronal Calcium Sensor 1. AI models were trained on large datasets of known PPIs, protein structures, and small molecular graphs to predict binding score. Using virtual screening of an FDA-approved drug library, the models generated a prioritized list of candidates. The most promising compounds were selected for experimental validation. This integrated approach efficiently reduced experimental workload, time, and cost, leading to the identification of dipyridamole as a selective modulator of the interaction between NCS-1 and the dopamine D2 receptor. The crystal structure of NCS-1 in complex with dipyridamole elucidates its mechanism of modulation. Our findings highlight the potential of AI to streamline and accelerate the discovery of novel therapeutic PPI modulators
Sparseness-Optimized Feature Importance for Time Series Classification
No full textThe literature reports a wide variety of attribution methods for explaining the predictions made by time series classification (TSC) algorithms. These post-hoc explanation methods span from model-specific to agnostic procedures that operate at different granularity levels. Despite their relative success, they often fail to generate sparse explanations, thereby increasing the cognitive overhead for experts seeking to isolate the most relevant features associated with model performance. Another limitation of segment-based explanation methods for TSC problems is that they do not allow any expert intervention. In this paper, we present an agnostic explainer termed Sparseness-Optimized Feature Importance (SOFI) that can be used to explain the predictions generated by any black-box TSC model. In practice, the explanation takes the form of a ranking of time series segments whose cumulative perturbation leads to fast degradation in model performance. Those segments should ideally be provided or defined by experts to ensure that explanations are meaningful and aligned with domain knowledge. As a second contribution, we mathematically demonstrate that under the modularity assumption, the optimal segment ranking associated with SOFI is unique. If the modularity assumption is dropped, we prove that multiple segment importance rankings lead to the optimal model performance degradation. In our experiments, we study the effect of different strategies for computing time series segments and perturbation operators on the explanation results. Simulation results show that SOFI generates explanations that are equally robust, up to 16 times more faithful, and 1.4 times sparser than those generated by the state-of-the-art explainers used for comparison
Weak-jamming detection in IEEE 802.11 networks:Techniques, scenarios and mobility
No full textState-of-the-art solutions detect jamming attacks ex-post, i.e., only when jamming has already disrupted the wireless communication link. In many scenarios, e.g., mobile networks or static deployments distributed over a large geographical area, it is often desired to detect jamming at the early stage, when it affects the communication link enough to be detected but not sufficiently to disrupt it (detection of weak jamming signals). Under such assumptions, devices can enhance situational awareness and promptly apply mitigation, e.g., moving away from the jammed area in mobile scenarios or changing communication frequency in static deployments, before jamming fully disrupts the communication link. Although some contributions recently demonstrated the feasibility of detecting low-power and weak jamming signals, they make simplistic assumptions far from real-world deployments. Given the current state of the art, no evidence exists that detection of weak jamming can be considered with real-world communication technologies. In this paper, we provide and comprehensively analyze new general-purpose strategies for detecting weak jamming signals, compatible by design with one of the most relevant communication technologies used by commercial-off-the-shelf devices, i.e., IEEE 802.11. We describe two operational modes: (i) binary classification via Convolutional Neural Networks and (ii) one-class classification via Sparse Autoencoders. We evaluate and compare the proposed approaches with the current state-of-the-art using data collected through an extensive real-world experimental campaign in three relevant environments. At the same time, we made the dataset available to the public. Our results demonstrate that detecting weak jamming signals is feasible in all considered real-world environments, and we provide an in-depth analysis that considers different techniques, scenarios, and mobility patterns.</p
