Artificial Intelligence (AI) models, encompassing both traditional machine learning (ML) and more advanced approaches such as deep learning and large language models (LLMs), play a central role in modern applications. AI model lifecycle management involves the end-to-end process of managing these models, from data collection and preparation to model building, evaluation, deployment, and continuous monitoring. This process is inherently complex, as it requires the coordination of diverse services that manage AI artifacts such as datasets, dataflows, and models, all orchestrated to operate seamlessly. In this context, it is essential to isolate applications from the complexity of interacting with heterogeneous services, datasets, and AI platforms. In this paper, we introduce Gypscie, a cross-platform AI artifact management system. By providing a unified view of all AI artifacts, the Gypscie platform simplifies the development and deployment of AI applications. This unified view is realized through a knowledge graph that captures application semantics and a rule-based query language that supports reasoning over data and models. Model lifecycle activities are represented as high-level dataflows that can be scheduled across multiple platforms, such as servers, cloud platforms, or supercomputers. Finally, Gypscie records provenance information about the artifacts it produces, thereby enabling explainability. Our qualitative comparison with representative AI systems shows that Gypscie supports a broader range of functionalities across the AI artifact lifecycle. Our experimental evaluation demonstrates that Gypscie can successfully optimize and schedule dataflows on AI platforms from an abstract specification.

Scientific workflows are used to model experiments that rely on computer simulations. Because these workflows are typically data-intensive, they commonly require execution in distributed environments. The cloud, with on-demand and elastic resources, has emerged as a cost-effective environment for workflow execution. However, the efficiency and cost-effectiveness of cloud workflow execution depend on how the workflow is executed on these resources. In particular, reusing cached data to avoid re-executing parts of the workflow is critical for performance but hard to make effective. This manuscript introduces MemoirGRASP, a workflow execution strategy based on the GRASP metaheuristic that optimizes execution while exploring memoization. The memoization technique involves caching intermediate results, reducing workflow execution redundancy, and enhancing efficiency across multiple workflow executions. The experimental evaluation of MemoirGRASP using synthetic workflows and two real-world workflows demonstrates the advantages and benefits of the MemoirGRASP strategy in improving workflow execution efficiency.

Federated Learning (FL) is a promising distributed machine learning approach that enables collaborative training of a global model using multiple edge devices. The data distributed among the edge devices is highly heterogeneous. Thus, FL faces the challenge of data distribution and heterogeneity, where non-Independent and Identically Distributed (non-IID) data across edge devices may yield in significant accuracy drop. Furthermore, the limited computation and communication capabilities of edge devices increase the likelihood of stragglers, thus leading to slow model convergence. In this paper, we propose the FedDHAD FL framework, which comes with two novel methods: Dynamic Heterogeneous model aggregation (FedDH) and Adaptive Dropout (FedAD). FedDH dynamically adjusts the weights of each local model within the model aggregation process based on the non-IID degree of heterogeneous data to deal with the statistical data heterogeneity. FedAD performs neuron-adaptive operations in response to heterogeneous devices to improve accuracy while achieving superb efficiency. The combination of these two methods makes FedDHAD significantly outperform state-of-the-art solutions in terms of accuracy (up to 6.7% higher), efficiency (up to 2.02 times faster), and computation cost (up to 15.0% smaller).

This paper presents Gypscie-KG, a system that integrates heterogeneous machine learning (ML) data into a knowledge graph (KG) using logicbased rules to enable semantic queries and reasoning. It also explores the use of ChatGPT to support relational-to-graph schema mapping, demonstrating AI’s potential in declarative data integration for complex ML systems.

Predictive queries over spatiotemporal (ST) stream data pose significant data processing and analysis challenges. ST data streams involve a set of time series whose data distributions may vary in space and time, exhibiting multiple distinct patterns. In this context, assuming a single machine learning model would adequately handle such variations is likely to lead to failure. To address this challenge, we propose StreamEnsemble, a novel approach to predictive queries over ST data that dynamically selects and allocates Machine Learning models according to the underlying time series distributions and model characteristics. Our experimental evaluation reveals that this method markedly outperforms traditional ensemble methods and single model approaches in terms of accuracy and time, demonstrating a significant reduction in prediction error of more than 10 times compared to traditional approaches.

Federated Learning (FL) has achieved significant achievements recently, enabling collaborative model training on distributed data over edge devices. Iterative gradient or model exchanges between devices and the centralized server in the standard FL paradigm suffer from severe efficiency bottlenecks on the server. While enabling collaborative training without a central server, existing decentralized FL approaches either focus on the synchronous mechanism that deteriorates FL convergence or ignore device staleness with an asynchronous mechanism, resulting in inferior FL accuracy. In this paper, we propose an Asynchronous Efficient Decentralized FL framework, i.e., AEDFL, in heterogeneous environments with three unique contributions. First, we propose an asynchronous FL system model with an efficient model aggregation method for improving the FL convergence. Second, we propose a dynamic staleness-aware model update approach to achieve superior accuracy. Third, we propose an adaptive sparse training method to reduce communication and computation costs without significant accuracy degradation. Extensive experimentation on four public datasets and four models demonstrates the strength of AEDFL in terms of accuracy (up to 16.3% higher), efficiency (up to 92.9% faster), and computation costs (up to 42.3% lower).

Multivariate time series find extensive applications in conjunction with machine learning methodologies for scenario forecasting across various domains. Nevertheless, certain domains exhibit inherent complexities and diversities, which detrimentally impact the predictive efficacy of global models. This ongoing study introduces a Subset Modeling Framework designed to acknowledge the inherent diversity within a domain’s multivariate space. Comparative assessments between subset models and global models are conducted in terms of performance, revealing compelling findings and suggesting the potential for further exploration and refinement of this novel framework.

Spatio-temporal Predictive Queries encompass a spatio-temporal constraint, defining a region, a target variable, and an evaluation metric. The output of such queries presents the future values for the target variable computed by predictive models at each point of the spatio-temporal region. Unfortunately, especially for large spatio-temporal domains with millions of points, training temporal models at each spatial domain point is prohibitive. In this work, we propose a data-driven approach for selecting pre-trained temporal models to be applied at each query point. The chosen approach applies a model to a point according to the training and input time series similarity. The approach avoids training a different model for each domain point, saving model training time. Moreover, it provides a technique to decide on the best-trained model to be applied to a point for prediction. In order to assess the applicability of the proposed strategy, we evaluate a case study for temperature forecasting using historical data and auto-regressive models. Computational experiments show that the proposed approach, compared to the baseline, achieves equivalent predictive performance using a composition of pre-trained models at a fraction of the total computational cost.

As a promising paradigm to collaboratively train models with decentralized data, Federated Learning (FL) can be exploited to fine-tune Large Language Models (LLMs). While LLMs correspond to huge size, the scale of the training data significantly increases, which leads to tremendous amounts of computation and communication costs. The training data is generally non-Independent and Identically Distributed (non-IID), which requires adaptive data processing within each device. Although Low Rank Adaptation (LoRA) can significantly reduce the scale of parameters to update in the fine-tuning process, it still takes unaffordable time to transfer the low-rank parameters of all the layers in LLMs. In this paper, we propose a Fisher Information-based Efficient Curriculum Federated Learning framework (FibecFed) with two novel methods, i.e., adaptive federated curriculum learning and efficient sparse parameter update. First, we propose a fisher information-based method to adaptively sample data within each device to improve the effectiveness of the FL fine-tuning process. Second, we dynamically select the proper layers for global aggregation and sparse parameters for local update with LoRA so as to improve the efficiency of the FL fine-tuning process. Extensive experimental results based on 10 datasets demonstrate that FibecFed yields excellent performance (up to 45.35% in terms of accuracy) and superb fine-tuning speed (up to 98.61% faster) compared with 17 baseline approaches).

Climate trends and weather indicators are used in several research fields due to their importance in statistical modeling, frequently used as covariates. Usually, climate indicators are available as grid files with different spatial and time resolutions. The availability of a time series of climate indicators compatible with administrative boundaries is scattered in Brazil, not fully available for several years, and produced with diverse methodologies. In this paper, we propose time series of climate indicators for the Brazilian municipalities produced using zonal statistics derived from the ERA5-Land reanalysis indicators. As a result, we present datasets with zonal statistics of climate indicators with daily data, covering the period from 1950 to 2022.

Data driven science requires manipulating large datasets coming from various data sources through complex workflows based on a variety of models and languages. With the increasing number of big data sources and models developed by different groups, it is hard to relate models and data and use them in unanticipated ways for specific data analysis. Current solutions are typically ad-hoc, specialized for particular data, models and workflow systems. In this paper, we focus on data driven life science and propose an open service-based architecture, Life Science Workflow Services (LifeSWS), which provides data analysis workflow services for life sciences. We illustrate our motivations and rationale for the architecture with real use cases from life science.

Distributed infrastructures for computation and analytics are now evolving towards an interconnected ecosystem allowing complex scientific workflows to be executed across hybrid systems spanning from IoT Edge devices to Clouds, and sometimes to supercomputers (the Computing Continuum). Understanding the performance trade-offs of large-scale workflows deployed on such complex Edge-to-Cloud Continuum is challenging. To achieve this, one needs to systematically perform experiments, to enable their reproducibility and allow other researchers to replicate the study and the obtained conclusions on different infrastructures. This breaks down to the tedious process of reconciling the numerous experimental requirements and constraints with low-level infrastructure design choices. To address the limitations of the main state-of-the-art approaches for distributed, collaborative experimentation, such as Google Co-lab, Kaggle, and Code Ocean, we propose KheOps, a collaborative environment specifically designed to enable cost-effective reproducibility and replicability of Edge-to-Cloud experiments. KheOps is composed of three core elements: (1) an experiment repository; (2) a notebook environment; and (3) a multi-platform experiment methodology. We illustrate KheOps with a real-life Edge-to-Cloud application. The evaluations explore the point of view of the authors of an experiment described in an article (who aim to make their experiments reproducible) and the perspective of their readers (who aim to replicate the experiment). The results show how KheOps helps authors to systematically perform repeatable and reproducible experiments on the Grid5000 + FIT IoT LAB testbeds. Furthermore, KheOps helps readers to cost-effectively replicate authors experiments in different infrastructures such as Chameleon Cloud + CHI@Edge testbeds, and obtain the same conclusions with high accuracies (>88% for all performance metrics).

The success of machine learning (ML) systems depends on data availability, volume, quality, and efficient computing resources. A challenge in this context is to reduce computational costs while maintaining adequate accuracy of the models. This paper presents a framework to address this challenge. The idea is to identify “subdomains” within the input space, train local models that produce better predictions for samples from that specific subdomain, instead of training a single global model on the full dataset. We experimentally evaluate our approach on two real-world datasets. Our results indicate that subset modelling (i) improves the predictive performance compared to a single global model and (ii) allows data-efficient training.

Although more layers and more parameters generally improve the accuracy of the models, such big models generally have high computational complexity and require big memory, which exceed the capacity of small devices for inference and incurs long training time. In addition, it is difficult to afford long training time and inference time of big models even in high performance servers, as well. As an efficient approach to compress a large deep model (a teacher model) to a compact model (a student model), knowledge distillation emerges as a promising approach to deal with the big models. Existing knowledge distillation methods cannot exploit the elastic available computing resources and correspond to low efficiency. In this paper, we propose an Elastic Deep Learning framework for knowledge Distillation, that is, EDL-Dist. The advantages of EDL-Dist are threefold. First, the inference and the training process is separated. Second, elastic available computing resources can be utilized to improve the efficiency. Third, fault-tolerance of the training and inference processes is supported. We take extensive experimentation to show that the throughput of EDL-Dist is up to 3.125 times faster than the baseline method (online knowledge distillation) while the accuracy is similar or higher.

Modern scientific workflows require hybrid infrastructures combining numerous decentralized resources on the IoT/Edge interconnected to Cloud/HPC systems (aka the Computing Continuum) to enable their optimized execution. Understanding and optimizing the performance of such complex Edgeto-Cloud workflows is challenging. Capturing the provenance of key performance indicators, with their related data and processes, may assist in understanding and optimizing workflow executions. However, the capture overhead can be prohibitive, particularly in resource-constrained devices, such as the ones on the IoT/Edge. To address this challenge, based on a performance analysis of existing systems, we propose ProvLight, a tool to enable efficient provenance capture on the IoT/Edge. We leverage simplified data models, data compression and grouping, and lightweight transmission protocols to reduce overheads. We further integrate ProvLight into the E2Clab framework to enable workflow provenance capture across the Edge-to-Cloud Continuum. This integration makes E2Clab a promising platform for the performance optimization of applications through reproducible experiments. We validate ProvLight at a large scale with synthetic workloads on 64 real-life IoT/Edge devices in the FIT IoT LAB testbed. Evaluations show that ProvLight outperforms state-of-the-art systems like ProvLake and DfAnalyzer in resource-constrained devices. ProvLight is 26-37x faster to capture and transmit provenance data; uses 5-7x less CPU; 2x less memory; transmits 2x less data; and consumes 2-2.5x less energy. ProvLight [1] and E2Clab [2] are available as open-source tools.

Cooperative Intelligent Transport Systems (C-ITS) is a promising technology to make transportation safer and more efficient. Ridesharing for long-distance is becoming a key means of transportation in C-ITS. In this paper, we focus on private long-distance ridesharing, which reduces the total cost of vehicle utilization for long-distance journeys. In this context, we investigate journey scheduling problem with shared vehicles to reduce the total cost of vehicle utilization. Most of the existing works directly schedule journeys to vehicles with long scheduling time and only consider the cost of driving travellers instead of the total cost. In contrast, to reduce the total cost and scheduling time, we propose a comprehensive cost model and a two-phase journey scheduling approach, which includes path generation and path scheduling. On this basis, we propose two path generation methods: a simple near optimal method and a reset near optimal method as well as a greedy based path scheduling method. Finally, we present an experimental evaluation with different path generation and path scheduling methods with synthetic data generated based on real-world data. The results reveal that the proposed scheduling approach significantly outperforms baseline methods in terms of total cost (up to 69.8%) and scheduling time (up to 84.0%) and the scheduling time is reasonable (up to 0.16s). The results also show that our approach has higher efficiency (up to 141.7%) than baseline methods.

The integration of machine learning algorithms into database systems has brought new opportunities in different areas from indexing to query optimization. In this paper, we describe the integration of an approach for the automatic computation of model ensembles to answer a predictive query. We have extended the SAVIME multi-dimensional array DBMS by adding a new function to its query language and implementing the selection and allocation ensemble model dataflow into the query processing component of SAVIME. We show some initial experimental results depicting its performance against a pure Python implementation of the ensemble approach. Interestingly enough the C++ implementation within SAVIME is up to 4 times faster than its competitor.

The explosion of data volumes generated by an increasing number of applications is strongly impacting the evolution of distributed digital infrastructures for data analytics and machine learning (ML). While data analytics used to be mainly performed on cloud infrastructures, the rapid development of IoT infrastructures and the requirements for low-latency, secure processing has motivated the development of edge analytics. Today, to balance various trade-offs, ML-based analytics tends to increasingly leverage an interconnected ecosystem that allows complex applications to be executed on hybrid infrastructures where IoT Edge devices are interconnected to Cloud/HPC systems in what is called the Computing Continuum, the Digital Continuum, or the Transcontinuum.Enabling learning-based analytics on such complex infrastructures is challenging. The large scale and optimized deployment of learning-based workflows across the Edge-to-Cloud Continuum requires extensive and reproducible experimental analysis of the application execution on representative testbeds. This is necessary to help understand the performance trade-offs that result from combining a variety of learning paradigms and supportive frameworks. A thorough experimental analysis requires the assessment of the impact of multiple factors, such as: model accuracy, training time, network overhead, energy consumption, processing latency, among others.This review aims at providing a comprehensive vision of the main state-of-the-art libraries and frameworks for machine learning and data analytics available today. It describes the main learning paradigms enabling learning-based analytics on the Edge-to-Cloud Continuum. The main simulation, emulation, deployment systems, and testbeds for experimental research on the Edge-to-Cloud Continuum available today are also surveyed. Furthermore, we analyze how the selected systems provide support for experiment reproducibility. We conclude our review with a detailed discussion of relevant open research challenges and of future directions in this domain such as: holistic understanding of performance; performance optimization of applications;efficient deployment of Artificial Intelligence (AI) workflows on highly heterogeneous infrastructures; and reproducible analysis of experiments on the Computing Continuum.

Deep neural networks proved to be useful to learn representations and perform classification on many different modalities of data. Traditional approaches work well on the closed set problem. For learning tasks involving novel classes, known as the open set problem, the metric learning approach has been proposed. However, while promising, common metric learning approaches require pairwise learning, which significantly increases training cost while adding additional challenges. In this paper we present a method in which the similarity of samples projected onto a feature space is enforced by a metric learning approach without requiring pairwise evaluation. We compare our approach against known methods in different datasets, achieving results up to $81\%$ more accurate.