The COVID-19 pandemic’s severe impact highlighted the need for accurate and timely hospitalization forecasting to support effective healthcare planning. However, most forecasting models struggled, particularly during variant surges, when they were most needed. This study introduces a novel parallel-stream Long Short-Term Memory (LSTM) framework to forecast daily state-level incident hospitalizations in the United States. Our framework incorporates a spatiotemporal feature, Social Proximity to Hospitalizations (SPH), derived from Meta’s Social Connectedness Index, to improve forecasts. SPH serves as a proxy for interstate population interaction, capturing transmission dynamics across space and time. Our architecture captures both short- and long-term temporal dependencies, and a multi-horizon ensembling strategy balances forecasting consistency and error. An evaluation against the COVID-19 Forecast Hub ensemble models during the Delta and Omicron surges reveals the superiority of our model. On average, our model surpasses the ensemble by 27, 42, 54, and 69 hospitalizations per state at the 7-, 14-, 21-, and 28-day horizons, respectively, during the Omicron surge. Data-ablation experiments confirm SPH’s predictive power, highlighting its effectiveness in enhancing forecasting models. This research not only advances hospitalization forecasting but also underscores the significance of spatiotemporal features, such as SPH, in modeling the complex dynamics of infectious disease spread.

Abstract Static maps and animations remain popular in spatial epidemiology of dengue, limiting the analytical depth and scope of visualizations. Over half of the global population live in dengue endemic regions. Understanding the spatiotemporal dynamics of the four closely related dengue serotypes, and their immunological interactions, remains a challenge at a global scale. To facilitate this understanding, we worked with dengue epidemiologists in a user‐centred design framework to create GeoDEN, an exploratory visualization tool that empowers experts to investigate spatiotemporal patterns in dengue serotype reports. The tool has several linked visualizations and filtering mechanisms, enabling analysis at a range of spatial and temporal scales. To identify successes and failures, we present both insight‐based and value‐driven evaluations. Our domain experts found GeoDEN valuable, verifying existing hypotheses and uncovering novel insights that warrant further investigation by the epidemiology community. The developed visual exploration approach can be adapted for exploring other epidemiology and disease incident datasets.

Sea ice type classification is essential for climate change research and maritime safety. Traditionally, this process relies on manual ice charting, which is time-consuming, expensive, and requires expert knowledge, making it difficult to scale up for current demands. Automating sea ice type classification is essential to keep pace with rapidly changing sea ice conditions. However, two main challenges limit the development of effective automated classifiers. First, while ice charts provide valuable labeled data, they only offer large-area (polygon) annotations rather than pixel-level labels, leading to a lack of precise training data. Second, although there are additional datasets with useful sea ice information, effectively combining these different data sources remains difficult. To tackle the first challenge, we employed co-training and label propagation, two semi-supervised learning methods, to learn from a small amount of labeled data and a large pool of unlabeled data, thereby improving the accuracy of sea ice classifiers despite limited labeled data. To address the second challenge, we leveraged co-training’s built-in ability to integrate multiple data sources during the training process for the small labeled data. Additionally, we further enhanced data integration by using an ensemble of these co-trained models after training. Our approach demonstrates significant improvements over traditional supervised methods, showcasing the potential of semi-supervised learning methods in addressing two major challenges in developing automated sea ice classification solutions. Our study shows that semi-supervised learning improved F1 scores by 17% for SAR data and 33% for AMSR2 with limited labels, compared to supervised methods, while ensembling further boosted accuracy by 33%.

Deploying deep learning on Synthetic Aperture Radar (SAR) data is becoming more common for mapping purposes. One such case is sea ice, which is highly dynamic and rapidly changes as a result of the combined effect of wind, temperature, and ocean currents. Therefore, frequent mapping of sea ice is necessary to ensure safe marine navigation. However, there is a general shortage of expert-labeled data to train deep learning algorithms. Fine-tuning a pre-trained model on SAR imagery is a potential solution. In this paper, we compare the performance of deep learning models trained from scratch using randomly initialized weights against pre-trained models that we fine-tune for this purpose. Our results show that pre-trained models lead to better results, especially on test samples from the melt season.

Up-to-date sea ice charts are crucial for safer navigation in ice-infested waters. Recently, Convolutional Neural Network (CNN) models show the potential to accelerate the generation of ice maps for large regions. However, results from CNN models still need to undergo scrutiny as higher metrics performance not always translate to adequate outputs. Sea ice type classes are imbalanced, requiring special treatment during training. We evaluate how three different loss functions, some developed for imbalanced class problems, affect the performance of CNN models trained to predict the dominant ice type in Sentinel-1 images. Despite the fact that Dice and Focal loss produce higher metrics, results from cross-entropy seem generally more physically consistent.

Aerosols play a critical role in atmospheric chemistry, and affect clouds, climate, and human health. However, the spatial coverage of satellite-derived aerosol optical depth (AOD) products is limited by cloud cover, orbit patterns, polar night, snow, and bright surfaces, which negatively impacts the coverage and accuracy of particulate matter modeling and health studies relying on air pollution characterization. We present a random forest model trained to capture spatial dependence of AOD and produce higher coverage through imputation. By combining the models with and without the mean filters, we are able to create full-coverage high-resolution daily AOD in the conterminous U.S., which can be used for aerosol estimation and other studies leveraging air pollutant concentration levels.

Despite the growing use of deep learning in sea ice mapping with SAR imagery, the study of model uncertainty and segmentation results remains limited. Deep learning models often produce overconfident predictions, a concern in sea ice mapping where misclassification can impact marine navigation safety. We incorporate and compare dropout and model ensemble within a convolutional neural network segmentation architecture to highlight regions with prediction uncertainty, and explore the impact of loss function choice. We evaluate model generalization and uncertainty characterization by training and evaluating models on the AI4Arctic Sea Ice Challenge Dataset (primary). We further explore model uncertainty by testing the trained models on the Extreme Earth version 2 Dataset (secondary). The primary and secondary datasets vary in number of scenes as well as in the available data and preprocessing. We obtain test F1 results higher than 0.97 for the primary dataset. Although the F1 performance for the secondary dataset is reduced to 0.93, the generated sea ice maps are reasonable across several Sentinel-1 scenes, and our proposed strategy helps in identification of misclassified and uncertain regions for human quality control. Our models seem to be robust against banding noise in Sentinel-1 SAR, and the prediction uncertainty frequently highlights ice regions misclassified as water, indicating its potential for real-world applications. Our study advances the field of machine learning-based sea ice mapping and highlights the importance of uncertainty estimation and cross-dataset evaluation for model development and deployment. Our approach can be adopted for other remote sensing applications as well.

Many evaluation methods have been used to assess the usefulness of Visual Analytics (VA) solutions. These methods stem from a variety of origins with different assumptions and goals, which cause confusion about their proofing capabilities. Moreover, the lack of discussion about the evaluation processes may limit our potential to develop new evaluation methods specialized for VA. In this paper, we present an analysis of evaluation methods that have been used to summatively evaluate VA solutions. We provide a survey and taxonomy of the evaluation methods that have appeared in the VAST literature in the past two years. We then analyze these methods in terms of validity and generalizability of their findings, as well as the feasibility of using them. We propose a new metric called summative quality to compare evaluation methods according to their ability to prove usefulness, and make recommendations for selecting evaluation methods based on their summative quality in the VA domain.

Social media platforms are filled with social spambots. Detecting these malicious accounts is essential, yet challenging, as they continually evolve to evade detection techniques. In this article, we present VASSL, a visual analytics system that assists in the process of detecting and labeling spambots. Our tool enhances the performance and scalability of manual labeling by providing multiple connected views and utilizing dimensionality reduction, sentiment analysis and topic modeling, enabling insights for the identification of spambots. The system allows users to select and analyze groups of accounts in an interactive manner, which enables the detection of spambots that may not be identified when examined individually. We present a user study to objectively evaluate the performance of VASSL users, as well as capturing subjective opinions about the usefulness and the ease of use of the tool.

Various domain users are increasingly leveraging real-time social media data to gain rapid situational awareness. However, due to the high noise in the deluge of data, effectively determining semantically relevant information can be difficult, further complicated by the changing definition of relevancy by each end user for different events. The majority of existing methods for short text relevance classification fail to incorporate users' knowledge into the classification process. Existing methods that incorporate interactive user feedback focus on historical datasets. Therefore, classifiers cannot be interactively retrained for specific events or user-dependent needs in real-time. This limits real-time situational awareness, as streaming data that is incorrectly classified cannot be corrected immediately, permitting the possibility for important incoming data to be incorrectly classified as well. We present a novel interactive learning framework to improve the classification process in which the user iteratively corrects the relevancy of tweets in real-time to train the classification model on-the-fly for immediate predictive improvements. We computationally evaluate our classification model adapted to learn at interactive rates. Our results show that our approach outperforms state-of-the-art machine learning models. In addition, we integrate our framework with the extended Social Media Analytics and Reporting Toolkit (SMART) 2.0 system, allowing the use of our interactive learning framework within a visual analytics system tailored for real-time situational awareness. To demonstrate our framework's effectiveness, we provide domain expert feedback from first responders who used the extended SMART 2.0 system.

Evaluating employee performance in organizations with varying workloads and tasks is challenging. Specifically, it is important to understand how quantitative measurements of employee achievements relate to supervisor expectations, what the main drivers of good performance are, and how to combine these complex and flexible performance evaluation metrics into an accurate portrayal of organizational performance in order to identify shortcomings and improve overall productivity. To facilitate this process, we summarize common organizational performance analyses into four visual exploration task categories. Additionally, we develop MetricsVis, a visual analytics system composed of multiple coordinated views to support the dynamic evaluation and comparison of individual, team, and organizational performance in public safety organizations. MetricsVis provides four primary visual components to expedite performance evaluation: (1) a priority adjustment view to support direct manipulation on evaluation metrics; (2) a reorderable performance matrix to demonstrate the details of individual employees; (3) a group performance view that highlights aggregate performance and individual contributions for each group; and (4) a projection view illustrating employees with similar specialties to facilitate shift assignments and training. We demonstrate the usability of our framework with two case studies from medium-sized law enforcement agencies and highlight its broader applicability to other domains.

Abstract Communication‐minded visualizations are designed to provide their audience—managers, decision‐makers, and the public—with new knowledge. Authoring such visualizations effectively is challenging because the audience often lacks the expertise, context, and time that professional analysts have at their disposal to explore and understand datasets. We present a novel summarized line graph visualization technique designed specifically for data analysts to communicate data to decision‐makers more effectively and efficiently. Our summarized line graph reduces a large and detailed dataset of multiple quantitative time‐series into (1) representative data that provides a quick takeaway of the full dataset; (2) analytical highlights that distinguish specific insights of interest; and (3) a data envelope that summarizes the remaining aggregated data. Our summarized line graph achieved the best overall results when evaluated against line graphs, band graphs, stream graphs, and horizon graphs on four representative tasks.

In this article, we present the GeoCorpora corpus building framework and software tools as well as a geo-annotated Twitter corpus built with these tools to foster research and development in the areas of microblog/Twitter geoparsing and geographic information retrieval. The developed framework employs crowdsourcing and geovisual analytics to support the construction of large corpora of text in which the mentioned location entities are identified and geolocated to toponyms in existing geographical gazetteers. We describe how the approach has been applied to build a corpus of geo-annotated tweets that will be made freely available to the research community alongside this article to support the evaluation, comparison and training of geoparsers. Additionally, we report lessons learned related to corpus construction for geoparsing as well as insights about the notions of place and natural spatial language that we derive from application of the framework to building this corpus.

Network Function Virtualization involves implementing network functions (e.g., virtualized LTE component) in software that can run on a range of industry standard server hardware, and can be migrated or instantiated on demand. A prediction service hosted on cloud infrastructures enables consumers to request the prediction information on-demand and respond accordingly. In this paper we introduce MOBaaS, which is a network function of Mobility and Bandwidth prediction cloudified over the cloud computing infrastructure. We implemented the service orchestration framework of MOBaaS, which can easily be setup and integrated with any other cloud-based LTE entities to provide prediction information about the future location of mobile user(s) as well as the network link(s) bandwidth availability. This information can be used to generate required triggers for on-demand deployment or scaling-up/down of virtualized network components as well as for the self-adaptation procedures and optimal network function configuration. We also describe the performance evaluation of the MOBaaS cloudification procedures and present an example of the benefit of such a prediction service.

In this paper we discuss the major modifications required in the current LTE network to realize a decentralized LTE architecture and develop a novel IP mobility management solution for it. The proposed solution can handle traffic redirecting and IP address continuity above the distributed anchor points in a scalable and resource efficient manner. Our approach is based on the NAT (Network Address Translation) mechanism, which is a well- known and widely used procedure in the current Internet. We extend the NS3-LENA to implement a decentralized LTE network as well as the proposed scheme. The evaluation results show that the proposed solution efficiently fulfills the functionality and performance requirements (e.g.,latency and signaling load) related to the mobility management.

The core architecture of current mobile networks does not scale well to cope with future traffic demands owing to its highly centralized composition. Typically, it is believed that decentralization of the network architecture would be a sustainable approach to deal with ever growing amount of mobile data traffic. Nevertheless, the decentralization strategy of network architecture has not been properly examined through quantitative performance studies. Given that LTE will be the leading mobile networking technology in the coming 5–10 years, we conduct a hybrid study model to compare performance of current and future (decentralized) LTE network architectures. Particularly, our analysis presents numerical results quantifying impact of the number of attached nodes on the load at network routers and links, on the latency, and on the processing cost of the user’s data and control planes. Analytical results demonstrate that decentralization of the LTE network architecture achieves higher performance compared to the current architecture and improves the latency and cost of data packet delivery more than 10 and 6 times, respectively. Furthermore, it is also observed that GTP outperforms PMIP for all studied performance metrics in the decentralized architecture and provides about twofold better latency and cost for data packet delivery and roughly 6 times lower data traffic load on the network routers.

The location of data centres is crucial when mobile network operators are moving towards cloudified mobile networks to optimize resource utilization and to improve performance of services. Quality of Experience (QoE) can be enhanced in terms of content access latency, by placing user content at locations where they will be present in the future. The Follow-Me Cloud (FMC) concept aims at optimising operations of moving Mobile Network Operators Services towards cloudified environments, where Information Centric Networking (ICN) and the appropriate content migration policies are of paramount importance. However, several factors need to be considered, including user movements and mobility prediction (MP), content popularity, and migration. This paper addresses all these aspects by implementing a fully integrated multi-criteria FMC and mobility prediction mechanisms (MP-FMC) on a cloud infrastructure. Experimental evaluation shows that MP-FMC can be orchestrated on-demand within a reasonable time frame, and it could deliver ≈ 33% improvement of content retrieval time.

One of the most important concepts used in mobile networks, like LTE (Long Term Evolution) is service continuity. A mobile user moving from one network to another network should not lose an on-going service. In cloud-based (virtualized) LTE systems, services are hosted on Virtual Machines (VMs) that can be moved and migrated across multiple networks to such locations where these services can be well delivered to mobile users. The migration of the (1) VMs and (2) the services running on such VMs, should happen in such a way that the disruption of an on-going service is minimized. In this paper we argue that a technology that can efficiently be used for supporting service and VM migration is the ICN/CCN (Information Centric Networking / Content Centric Networking) technology.

Distributed Mobility Management (DMM) is a mobility management solution, where the mobility anchors are distributed instead of being centralized. The use of DMM can be applied in cloud-based (virtualized) Long Term Evolution (LTE) mobile network environments to (1) provide session continuity to users across personal, local, and wide area networks without interruption and (2) support traffic redirection when a virtualized LTE entity like a virtualized Packet Data Network Gateway (P-GW) running on an virtualization platform is migrated to another virtualization platform and the on-going sessions supported by this P-GW need to be maintained. In this paper we argue that the enabling technology that can efficiently be used for supporting DMM in virtualized LTE systems is the Software Defined Networking (SDN)/OpenFlow technology.