Viral diseases have had a significant impact on millions of people worldwide. Utilizing time series forecasting methods allows for the estimation of cases, facilitating the control of disease spread and the allocation of necessary resources in medical facilities. Traditional estimation methods use model per dataset methodology where a model is trained only on a single dataset for a disease. However, foundation models including large language models have shown improved results by training on multiple datasets before being applied to a target dataset. Following this strategy, we aim to train a time series model using multiple datasets from diverse domains and viral diseases. We utilize a pretrained large language model and adapt it for estimating Influenza-Like Illness. We propose a novel network architecture called LLM4cast that encodes the input patches through a bidirectional encoder for rich embedding extraction and passes the encoded patches to a pretrained TinyLlama for fine-tuning. The output from TinyLlama is then flattened and projected to estimate the probable cases. The framework is trained in two stages. In the first stage, we train using time series data from diverse domains with 2.56M timesteps. The second stage involves training using domain specific time series data about viral diseases. The trained network is used in the evaluation of disease cases. The results demonstrate significant improvement in the accuracy of forecasts for foundation models compared to state-of-the-art models trained from scratch.

The success of the Transformer model has promoted recent advances in time series forecasting.This adoption sparked an interest in developing efficient Transformer models that scale well for forecasting long sequences.This involves maintaining non-autoregressive one-time decoding.However, the role of autoregressive decoding is less explored.To address this gap, we revisit an essential idea of the vanilla Transformer model and show that autoregressive decoding works well compared to non-autoregressive decoding.It also becomes vital for critical forecasting tasks, such as pandemic forecasting, where the stakes are high.Our code and data are publicly available at https://github.com/maldosari1/ar_transformer_tf.

We propose a new framework for 2-D interpreting (features and samples) black-box machine learning models via a metamodeling technique, by which we study the output and input relationships of the underlying machine learning model. The metamodel can be estimated from data generated via a trained complex model by running the computer experiment on samples of data in the region of interest. We utilize a Gaussian process as a surrogate to capture the response surface of a complex model, in which we incorporate two parts in the process: interpolated values that are modeled by a stationary Gaussian process Z governed by a prior covariance function, and a mean function μ that captures the known trends in the underlying model. The optimization procedure for the variable importance parameter θ is to maximize the likelihood function. This θ corresponds to the correlation of individual variables with the target response. There is no need for any pre-assumed models since it depends on empirical observations. Experiments demonstrate the potential of the interpretable model through quantitative assessment of the predicted samples.

The deep learning revolution, driven especially by incredible achievements in image recognition technology, encourages a predominately data-driven approach to machine learning and data science techniques. However, the emphasis on statistical inference from big data sources alone has diminished the role of existing knowledge and well-established theory. For example, vehicle traffic patterns are highly regular on a weekly period but can deviate unexpectedly in certain situations such as inclement weather, accidents, road work, etc. A purely datadriven approach would simply ask the model to handle all the different situations, while knowledge and situation-aware systems will take advantage of existing knowledge to help guide the machine learning process. This work aims to illustrate how knowledge and situational awareness can help data scientists to build more effective models in the field of vehicle traffic forecasting. The special circumstances considered here include holidays, special weather conditions, accidents, and location awareness that facilitates transfer learning. In addition, we present a novel modeling technique, Quadratic Extreme Learning Machine, that generally improves upon the standard Extreme Learning Machine model while remaining relatively efficient. The Quadratic Extreme Learning Machine can be potentially used as an alternative to Neural Networks, which generally entails higher computational costs.

Efforts to improve early identification of aggressive high grade breast cancers, which pose the greatest risk to patient health if not detected early, are hindered by the rarity of these events. To address this problem, we proposed an accurate and efficient deep transfer learning method to handle the imbalanced data problem that is prominent in breast cancer data. In contrast to existing approaches based primarily on large image databases, we focused on structured data, which has not been commonly used for deep transfer learning. We used a number of publicly available breast cancer data sets to generate a "pre-trained" model and transfer learned concepts to predict high grade malignant tumors in patients diagnosed with breast cancer at Mayo Clinic. We compared our results with state-of-the-art techniques for addressing the problem of imbalanced learning and confirmed the superiority of the proposed method. To further demonstrate the ability of the proposed method to handle different degrees of class imbalance, a series of experiments were performed on publicly available breast cancer data under simulated class imbalanced settings. Based on the experimental results, we concluded that the proposed deep transfer learning on structured data can be used as an efficient method to handle imbalanced class problems in clinical research.

As research and development in Web Services and Services Computing has only existed for approximately a decade and half, it is remarkably that it has lead to a multi-billion dollar services industry. The first major research conference on the subject, the International Conference on Web Services, just started in 2003. Research in services computing is active and multifaceted. This special issue of IEEE Transactions on Services Computing considers two areas in services computing that have experienced “Recent Advances in Web Services Research”: Security, Privacy, and Trust; and New Approaches to Services Composition. It consists of the best extended papers from the two premier research conferences on services computing: the 2016 IEEE International Conference on Web Services and the 2016 IEEE Conference on Services Computing.

One of the biggest challenges for today's data scientists is to be able to make an informed decision among an exhaustive number of different modeling techniques. As no single algorithm can perform optimally in all cases, the context for the modeling task including the dataset characteristics plays an unsurprisingly important role in deciding which modeling algorithm to choose. In our previous work, we have presented an ontology-based automated model-selection system extending the Scala-based SCALATION data framework. In this study, we present a meta-learning based model-selection system and provide an evaluation of the system.

Open Science Big Data is emerging as an important area of research and software development. Although there are several high quality frameworks for Big Data, additional capabilities are needed for Open Science Big Data. These include data provenance, citable reusable data, data sources providing links to research literature, relationships to other data and theories, transparent analysis/reproducibility, data privacy, new optimizations/advanced algorithms, data curation, data storage and transfer. An important part of science is explanation of results, ideally leading to theory formation. In this paper, we examine means for supporting the use of theory in big data analytics as well as using big data to assist in theory formation. One approach is to fit data in a way that is compatible with some theory, existing or new. Functional Data Analysis allows precise fitting of data as well as penalties for lack of smoothness or even departure from theoretical expectations. This paper discusses principal differential analysis and related techniques for fitting data where, for example, a time-based process is governed by an ordinary differential equation. Automation in theory formation is also considered. Case studies in the fields of computational economics and finance are considered.

Research on Web Services and Services Computing began shortly after the start of the new millennium and with the first major research conference on the subject, the International Conference on Web Services, starting in 2003. Due to the new and practical utility of Web service technology, many researchers began to work in this field. New ideas, applications and related technologies continue to invigorate this discipline. This special issue of IEEE Transactions on Services Computing considers four such topics that lead to “Advances in Web Services Research”: Cloud Services; Polices and Agreements; Services Engineering; and Service/Process Mining. It consists of the best extended papers from the two premier research conferences on services computing: the 2015 IEEE International Conference on Web Services and the 2015 IEEE Conference on Services Computing.

Key foundational components of Big Data frameworks include efficient large-scale storage and high-performance linear algebra. This paper discusses efficient implementations that utilize compression techniques inspired by columnar relational databases for improving space and time profiles for vector and matrix operations. In addition, linear algebra operations are integrated with columnar relational algebra operations both in dense and compressed forms. For several of the operations substantial speedups are obtained by operating directly on the compressed relations, vectors and matrices. Advantages of mixing and matching relational and linear algebra operations are also pointed out. Both serial and parallel implementations are provided in the ScalaTion Big Data Analytics Framework.

MOTIVATION: In the field of glycomics research, several different techniques are used for structure elucidation. Although multiple techniques are often used to increase confidence in structure assignments, most glycomics databases allow storing of only a single type of experimental data. In addition, the methods used to prepare a sample for analysis is seldom recorded making it harder to reproduce the analytical data and results. RESULTS: We have extended the freely available EUROCarbDB framework to allow the submission of experimental data and the reporting of several orthogonal experimental datasets. The features aim to increase the understandability and reproducibility of the reported data. AVAILABILITY AND IMPLEMENTATION: The installation with the glycan standards is available at http://glycomics.ccrc.uga.edu/eurocarb/. The source code of the project is available at https://code.google.com/p/ucdb/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

In the big data era, scalability has become a crucial requirement for any useful computational model. Probabilistic graphical models are very useful for mining and discovering data insights, but they are not scalable enough to be suitable for big data problems. Bayesian Networks particularly demonstrate this limitation when their data is represented using few random variables with a massive set of outcome values for each of them. With hierarchical data - data that is arranged in a treelike structure with several levels - one would expect to see hundreds of thousands or millions of values distributed over even just a small number of levels. When modeling this kind of hierarchical data across large data sets, Bayesian networks become unsuitable for representing the probability distributions for the following reasons: i) each level represents a single random variable with hundreds of thousands of values, ii) the number of levels is usually small, so there are also few random variables, and iii) the structure of the network is predefined since the dependency is modeled top-down from each parent to each of its child nodes. In this paper we propose a scalable probabilistic graphical model to overcome these limitations for massive hierarchical data. We believe the proposed model will lead to an easily-scalable, more readable, and expressive implementation for problems that require probabilistic-based solutions for massive amounts of hierarchical data. We successfully applied this model to solve two different challenging probabilistic-based problems on massive hierarchical data sets for different domains, namely, bioinformatics and latent semantic discovery over search logs.

Using caching techniques to improve response time of queries is a proven approach in many contexts. However, it is not well explored for subgraph pattern matching queries, mainly because of subtleties enforced by traditional pattern matching models. Indeed, efficient caching can greatly impact the query answering performance for massive graphs in any query engine whether it is centralized or distributed. This paper investigates the capabilities of the newly introduced pattern matching models in graph simulation family for this purpose. We propose a novel caching technique, and show how the results of a query can be used to answer the new similar queries according to the similarity measure that is introduced. Using large real-world graphs, we experimentally verify the efficiency of the proposed technique in answering subgraph pattern matching queries.

Graph pattern matching is fundamentally important to many applications such as analyzing hyper-links in the World Wide Web, mining associations in online social networks, and substructure search in biochemistry. Most existing graph pattern matching algorithms are highly computation intensive, and do not scale to extremely large graphs that characterize many emerging applications. In recent years, graph processing frameworks such as Pregel have sought to harness shared nothing clusters for processing massive graphs through a vertex-centric, Bulk Synchronous Parallel (BSP) programming model. However, developing scalable and efficient BSP-based algorithms for pattern matching is very challenging because this problem does not naturally align with a vertex-centric programming paradigm. This paper presents novel distributed algorithms based on the vertex-centric programming paradigm for a set of pattern matching models, namely, graph simulation, dual simulation and strong simulation. Our algorithms are fine-tuned to consider the challenges of pattern matching on massive data graphs. Furthermore, we introduce a new pattern matching model, called strict simulation, which outperforms strong simulation in terms of scalability while preserving its important properties. We investigate potential performance bottlenecks and propose several techniques to mitigate them. This paper also presents an extensive set of experiments involving massive graphs (millions of vertices and billions of edges) to study the effects of various parameters on the scalability and performance of the proposed algorithms. The results demonstrate that our techniques are highly effective in alleviating performance bottlenecks and yield significant scalability benefits.

Time evolving graph (TEG) is increasingly being used as a paradigm for modeling and analyzing dynamic relationships in many emerging domains such as online social networks, World Wide Web and evolutionary genomics. A time-evolving graph consists of a sequence of snapshots of the graph as it evolves

The wide-scale development of ontologies in the bioinformatics domain facilitates their use in the creation of scientific workflows. To speed up the design of workflows, a Service Suggestion Engine is interfaced to the Galaxy Tool Integration and Workflow Platform. This enables users to ask for suggestions (e.g., what operation should go next) while designing workflows with the Galaxy user interface. The Service Suggest Engine utilizes semantic annotations to suggest appropriate Web service operations to plug into the workflow under design. The enriched Ontology for Biomedical Investigation (OBI) is used as a target for the annotations. The effectiveness of the suggestions provided is evaluated against a consensus of domain experts.

This paper presents a semi-automatic Web service composition approach. This approach ranks all available candidate Web service operations based on semantic annotations and suggests service operations to a human designer during the process of Web service composition. The ranking scores are based on data mediation, functionality and formal service specifications. A formal graph model, an IODAG, is defined to formalize an input/output schema of a Web service operation. Three data mediation algorithms are developed to handle the data heterogeneities arising during Web service composition. The data mediation algorithms analyze the schemas of the inputs and outputs of service operations and consider the structures of the schemas. A typed representation for our data mediation approach, which formalizes the data mediation problem as a subtype-checking problem, is presented. An evaluation is performed to study the effectiveness of different data mediation and service suggestion algorithms used to assist designers composing Web services.

Recently, mashups are gaining tremendous popularity as an important Web 2.0 application. Mashups provide end-users with an opportunity to create personalized Web services which aggregate and manipulate data from multiple diverse sources distributed across the Web. However, this increase in personalization also results in new scalability and performance challenges. Surprisingly, there are very few studies on the performance aspect of mashups. In this paper, we propose two novel techniques to enhance the scalability and performance of mashup platforms. The first is an efficient mashup merging scheme that avoids duplicate computations and unnecessary data retrievals by detecting common operator sequences in different mashups and executing them together. Second, we propose a canonical form-based mashup reordering scheme that not only transforms individual mashups to their most efficient forms but also increases the effectiveness of mashup merging. This paper also reports a number of experiments studying the benefits and costs of the proposed techniques.

The process of selecting Web services from a large number of potential services available on the Web is a challenging task for users engaged in Web service composition. This work is devoted to resolving this issue by suggesting Web services to the user. Our suggestion algorithm ranks all the available services for the user based on the semantic annotations of a service's inputs, outputs and functionality, as well as pre-conditions and effects, if available. This paper presents multiple algorithms for making suggestions during Web service composition. These algorithms extend traditional Web service discovery algorithms; in particular, they include new techniques for ranking the effectiveness of data mediation.

The recent surge of popularity has established mashups as an important category of Web 2.0 applications. Mashups are essentially Web services that are often created by end-users. They aggregate and manipulate data from sources around the World Wide Web. Surprisingly, there are very few studies on the scalability and performance of mashups. In this paper, we study caching as a vehicle for enhancing the scalability and the efficiency of mashups. Although caching has long been used to improve the performance of Web services, mashups pose some unique challenges that necessitate a more dynamic approach to caching. Towards this end, we present MACE - a cache specifically designed for mashups. In designing the MACE framework this paper makes three technical contributions. First, we present a model for representing mashups and analyzing their performance. Second, we propose an indexing scheme that enables efficient reuse of cached data for newly created mashups. Finally, this paper also describes a novel caching policy that analyzes the costs and benefits of caching data at various stages of different mashups and selectively stores data that is most effective in improving system scalability. We report experiments studying the performance of the MACE system.

With the rising popularity of Web services, both academia and industry have invested considerably in Web service description standards, discovery, and composition techniques. The standards based approach utilized by Web services has supported interoperability at the syntax level. However, issues of structural and semantic heterogeneity between messages exchanged by Web services are far more complex and crucial to interoperability. It is for these reasons that we recognize the value that schema/data mappings bring to Web service descriptions. In this paper, we examine challenges to interoperability; classify the types of heterogeneities that can occur between interacting services and present a possible solution for data mediation using the mapping support provided by WSDL-S, the extensibility features of WSDL and the popular SOAP engine, Axis 2

We present methods for optimally adapting Web processes to exogenous events while preserving inter-service constraints that necessitate coordination. For example, in a supply chain process, orders placed by a manufacturer may get delayed in arriving. In response to this event, the manufacturer has the choice of either waiting out the delay or changing the supplier. Additionally, there may be compatibility constraints between the different orders, thereby introducing the problem of coordination between them if the manufacturer chooses to change the suppliers. We focus on formulating the decision making models of the managers, who must adapt to external events while satisfying the coordination constraints, using Markov decision processes. Our methods range from being centralized and globally optimal in their adaptation but not scalable, to decentralized that is suboptimal but scalable to multiple managers. We also develop a hybrid approach that improves on the performance of the decentralized approach with a minimal loss of optimality

The growing popularity of service oriented computing based on Web services standards is creating a need for paradigms to represent and design business processes. Significant work has been done in the representation aspects with regards to WSBPEL. However, design and modeling of business processes is still an open issue. In this paper, we present a novel designer for business processes, which allows for intuitive modeling of Web processes, as well as using a template based approach for semi-automatically integrating partners either at design time or at deployment time. This work has been done as part of the METEOR-S project, which concentrates on adding semantics to the entire Web process lifecycle.

Abstract Database visualization helps users process, interpret and act upon large stored data sets. In this paper, we present a Java‐based 3D database visualization tool called J3DV. The J3DV tool successfully solved the problem of data management faced by many other visualization systems by integrating multiple data sources with the visualization tool. The tool utilizes two‐level mapping to transform the data into intermediate data that can be used to render graphs. Intermediate data offers better performance with a two‐tier cache. This visualization tool presents a sound framework, which has good extensibility for plugging in new data sources, supporting new data models and visual presentation types and allowing new graph layout algorithms. Copyright © 2002 John Wiley & Sons, Ltd.

Genetic algorithms have demonstrated considerable success in providing good solutions to many NP-hard optimization problems. For such problems, exact algorithms that always find an optimal solution are only useful for small toy problems, so heuristic algorithms such as the genetic algorithm must be used in practice. In this paper, we apply the genetic algorithm to the NP-hard problem of multiple fault diagnosis (MFD). We compare a pure genetic algorithm with several variants that include local improvement operators. These operators, which are often domain-specific, are used to accelerate the genetic algorithm in converging on optimal solutions. Our empirical results indicate that by using the appropriate local improvement operator, the genetic algorithm is able to find an optimal solution in all but a tiny fraction of the cases and at a speed orders of magnitude faster than exact algorithms.>

The performance modeling described in this paper compares several protocols which ensure that a database can be recovered to a consistent state after a transaction failure or system crash. The contributions of the paper include a collection of simple analytic models, based on Markov processes, for these protocols and some surprising results on the relative performance of the protocols. We consider only two-stage transactions (all reads before writes) and ignore effects of serializing transactions. The most interesting performance result presented is that, for systems obeying the assumptions of this paper, the "pessimistic" policy of holding write locks to commit point is considerably less efficient than the "optimistic" policy which allows reading of uncommitted data, but risks cascading aborts. A multiversion policy introduced in [2] was also studied and found always to be nearly as good as the optimistic policy and sometimes much better.