Integrated Software (United States)

The glacial isostatic adjustment (GIA) and sea level modeling communities have historically lagged other fields in adhering to the FAIR principles of making model outputs findable, accessible, interoperable, and reusable – a delay that has slowed scientific discovery. While sharing model outputs has improved recently, usability of available outputs continues to be hindered by lack of standardization. Meanwhile, legacy model outputs can be lost as the technology storing them grows obsolete and their creators retire or leave academia.The GIAmachine initiative addresses this problem. GIAmachine aims to make accessible as many published GIA and sea level model outputs as are retrievable bycataloguing and standardizing published GIA and sea level model outputs; contacting authors of published-but-inaccessible models to encourage them to upload their outputs to DOI-minting repositories;partnering with the GHub science gateway to make a long-term home for these newly available outputs;building Jupyter notebooks on GHub that make these models interoperable and easy to use; and encouraging the GIA and sea level modeling communities to follow the FAIR principles. Our poster will introduce the GIAmachine online portal and outline outstanding challenges. We appreciate community input for designing a living resource that meets the specific needs of current and future scientists.

Microservices have become a foundational architectural style for building scalable, resilient, and independently deployable enterprise systems. Spring Boot significantly accelerates the adoption of microservices by providing opinionated configuration, embedded servers, streamlined dependency management, and seamless integration with Spring Cloud for service discovery, routing, configuration, and resilience. Despite these advantages, microservice environments often degrade in quality when architectural patterns are ignored and anti-patterns emerge, such as improper service decomposition, excessive synchronous communication, and shared persistence layers. These issues create tightly coupled distributed monoliths that undermine agility and scalability. This paper analyzes the essential patterns and anti-patterns for designing effective Spring Boot microservices, supported by implementation strategies, architectural diagrams, and a real-world case study in retail-fuel transactions. The goal is to help practitioners build highly observable, fault-tolerant, maintainable systems while avoiding common pitfalls encountered during microservice modernization programs.

ABSTRACT: Modern energy platforms orchestrate dozens of external APIs-grid market interfaces, fleet telematics, billing systems, weather services, regulatory reporting endpoints-into complex workflows that power demand response, fleet charging, energy trading, and customer operations. As these platforms scale to manage thousands of EVs across multiple grid markets, the orchestration layer itself becomes a critical bottleneck: manual workflow configuration is slow, rule-based error handling misses cascading failures, and static API routing wastes both latency budget and cost. This paper presents an AI-powered API orchestration engine that replaces rule-based workflow management with eight specialized ML models: an intent classifier that routes incoming requests, a graph neural network workflow planner that generates optimal API call sequences, an anomaly detector that predicts API failures 15 minutes before they occur, an RL-based SLA optimizer that dynamically tunes timeouts and retries, and four additional models for cost allocation, schema mediation, failure prediction, and load balancing. Through a 16-month production deployment managing 42 external API integrations and 85 internal microservices for a fleet of 10,200 EVs across four grid ISO markets, the platform achieves a 98.7% workflow success rate (up from 72.4% with manual orchestration), reduces average workflow execution time from 42 minutes to 2.4 minutes, auto-resolves 94.2% of API errors with a mean time to recovery of 12 seconds, and reduces per-workflow cost from $18.50 to $1.15-a 93.8% reduction. The platform manages $45.6M in annual energy transaction revenue and achieves break-even ROI in 7 months. Five energy vertical case studies validate cross-domain effectiveness.

Automatic Target Recognition (ATR) constitutes a fundamental capability of Maritime Autonomous Vehicles, with corresponding services manifesting in a variety of forms, while relying on common core concepts - such as an overarching process and the main tasks forming it. Expanding beyond maritime contexts, such services find further application across different domains. For instance, radar systems serve both the classification of items in maritime environments and the identification of flying objects in air traffic control [1,2].

Cells process dynamic signaling inputs to regulate fate decisions during development. While oscillations or waves in key developmental pathways, such as Wnt, have been widely observed, the principles governing how cells decode these signals remain unclear. By leveraging optogenetic control of the Wnt signaling pathway in both HEK293T cells and H9 human embryonic stem cells, we systematically map the relationship between signal frequency and downstream pathway activation. We find that cells exhibit a minimal response to Wnt at certain frequencies, a behavior we term anti-resonance. We developed both detailed biochemical and simplified hidden variable models that explain how anti-resonance emerges from the interplay between fast and slow pathway dynamics. Remarkably, we find that frequency directly influences cell fate decisions involved in human gastrulation; signals delivered at anti-resonant frequencies result in dramatically reduced mesoderm differentiation. Our work reveals a previously unknown mechanism of how cells decode dynamic signals and how anti-resonance may filter against spurious activation. These findings establish new insights into how cells decode dynamic signals with implications for tissue engineering, regenerative medicine, and cancer biology.

As the e-commerce platforms grew exponentially, the volume of multilingual customer reviews increased, indicating that sentiment analysis is a priceless tool for finding consumer sentiment, enhancing marketing strategies, and improving customer experience. Nevertheless, emotion classification in multilingual reviews is very hard, and for one causes the variability of the language, the ambiguity of the sentiment, hierarchical word dependencies, and class imbalance, which can skew traditional models. In order to resolve such challenges, this paper introduces a T5-CapsNet ensemble model, which combines the T5 transformer for context-embedded feature extraction with Capsule Networks (CapsNet) for hierarchical sentiment learning. Furthermore, the model is further enhanced by a GAN-based data augmentation technique, which increases the number of minority class reviews in a dataset by adding synthetic minority class reviews in an effort to correct dataset imbalance and promote classification fairness. As an ensemble fusion strategy, weighted voting and stacking ensemble learning are used to improve sentiment prediction by making good use of the advantages of T5 and CapsNet. Experimental evaluations on the Multilingual Amazon Reviews Corpus (MARC) confirm that the proposed model surpasses the best sentiment classifier to reach an accuracy of 97.56%. It turns out that this hybrid deep learning approach very well captures the complex sentiment structures, or to put it differently, the multilingual e-commerce sentiment analysis largely benefited from such a hybrid deep learning approach. The findings from this study will be a foundation for building more advanced emotion classification models that can assist in improving customer sentiment analysis, automated feedback systems, as well as decision-making in global e-commerce ecosystems.

The effective processing of API calls by cloud providers such as AWS, Microsoft Azure, or GCP has become increasingly challenging as multi-cloud adoption has increased. Reliability, cost maintenance, latency reduction, and performance improvement are some advantages of using several clouds. Traditional approaches frequently fail because they don't respond well to the dynamic and shifting nature of workloads in multi-cloud environment. We solve this by introducing a deep Q-network-based AI approach for autonomous API request routing. This provides reinforcement learning for picking the most relevant and valuable cloud provider based on live metric parameters such as latency, cost, and failure rates. The system remains versatile, subject to changing conditions, to ensure that performance optimization is achieved across AWS, Azure, and GCP. These findings show significant improvements and accomplishments in key performance indicators (KPI). API latency has been reduced from 250 ms to 140 ms, approximately a 44% reduction, with a corresponding decrease in the cost per request of 20%. The success rate has increased from 92% to 98%, and the failure rate has decreased from 8% to 2%. There is also a balanced load across the three clouds <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(34 \%, 33 \%$</tex>, and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$33 \%)$</tex> with an energy savings of 10%. Thus, our results demonstrate that AI-enhanced reinforcement learning can be highly beneficial in enhancing multi-cloud API orchestration. Besides these, they will also be resource-efficient and therefore very scalable solutions for the modern-day cloud architectures.

Undetected anomalous incidents on the freeway, such as crashes, can cause increased congestion and secondary crashes. To address this problem, we recently released a real freeway traffic anomalous event detection dataset. However, our evaluation was limited to a static week of traffic data. This demo presents a live anomaly detection dashboard, processing sensor data with a 1-minute delay. It has interactive components to adjust the anomaly detection performance, providing a realistic evaluation of graph anomaly detection methods for long-term analysis and method improvements.

This article presents experimental evidence of the ability of a single automated vehicle acting as a controller to effectively dissipate stop-and-go waves in real traffic. The automated vehicle succeeded in stabilizing the speed profile by reducing oscillations in time and speed variations between vehicles during rush hour on I-24 in the Nashville area. We detail the control design, deployment and results obtained in this experiment, conducted as part of the CIRCLES consortium’s “MegaVanderTest” 2022, which involved a total of 100 automated vehicles.

Abstract The paper reports about the work performed in customizing the RELAP/SCDAPSIM code features for an easier assessment of a conceptual design of a Small Modular Reactor (SMR) operating at supercritical pressure proposed by Schulenberg and Otic [13] in the frame of the EU ECC-SMART Project. First unsuccessful attempts to make use of an NRC version of the RELAP5 code in the analysis of the thermal-hydraulic behaviour of the SCW SMR revealed numerical difficulties at the transition from supercritical to subcritical pressures. This suggested to activate a cooperation for customising an already existing version of the RELAP/SCDAPSIM code, aiming to overcome similar numerical problems and to implement heat transfer correlations in principle more appropriate for supercritical pressure conditions. The improved version of the code, featuring a user defined generalised correlation for heat transfer encompassing several of the presently available literature proposals, was applied in predicting rod-bundle experiments performed by Razumovskiy et al. [16–17] and in analysing the behaviour of the seven core flow passage reactor proposed in [13]. In the paper, the present capabilities of the code are discussed in front of the obtained results and perspectives for future developments are suggested.