Generative AI has caused a paradigm shift in the area of Artificial Intelligence (AI) and as such has inspired much new research, especially on Large Language Models (LLMs). LLMs are transforming how people interact with computers in service-oriented fields in both the consumer (for example: retail, travel, education, healthcare) and enterprise (customer care, field service, sales, marketing, etc.) spaces. One barrier to widespread adoption is the current unpredictability of LLM behavior: users must trust that LLM-based services and systems are accurate, fair, and unbiased. Model responses that exhibit biases related to race, social status, and other sensitive topics can have serious consequences, ranging from lack of trust in the model to adverse social implications for consumers, all the way to damage to the reputations of the corporations that provide them. This study explores how to uncover biases related to social stigmas in LLM output, by using an adversarial prompt-based approach. Discovering model vulnerabilities of this type is a nontrivial task due to the large search space, making it resource-intensive. We present an evaluation framework for probing and analyzing the behaviors of multiple LLMs systematically. We use a curated set of adversarial prompts with a focus on uncovering biased responses to prompts associated with social attributes.

Graph Neural Networks (GNNs) have enjoyed wide spread applications in graph-structured data. However, existing graph based applications commonly lack annotated data. GNNs are required to learn latent patterns from a limited amount of training data to perform inferences on a vast amount of test data. The increased complexity of GNNs, as well as a single point of model parameter initialization, usually lead to overfitting and sub-optimal performance. In addition, it is known that GNNs are vulnerable to adversarial attacks. In this paper, we push one step forward on the ensemble learning of GNNs with improved accuracy, generalization, and adversarial robustness. Following the principles of stochastic modeling, we propose a new method called GNN-Ensemble to construct an ensemble of random decision graph neural networks whose capacity can be arbitrarily expanded for improvement in performance. The essence of the method is to build multiple GNNs in randomly selected substructures in the topological space and subfeatures in the feature space, and then combine them for final decision making. These GNNs in different substructure and subfeature spaces generalize their classification in complementary ways. Consequently, their combined classification performance can be improved and overfitting on the training data can be effectively reduced. In the meantime, we show that GNN-Ensemble can significantly enhance the adversarial robustness against attacks on GNNs.

The increased digitization of commercial and consumer workflows, accelerated cloud adoption, and the growing sophistication of cyber criminals have resulted in ransomware emerging as a major threat of cyber attacks in cloud and data services. While research in malware detection can be partially adapted for ransomware, specific ransomware infection patterns can be leveraged to improve the detection efficiency. In this paper, we focus on identifying vulnerable targets in ransomware attacks, aiming to accelerate the ransomware detection process as well as enable better data backup policies design. Specifically, we make three contributions. First, we characterize lexical features and hierarchical file structure features on those ransomware infected files and folders. Second, we model the data backup as an attributed tree graph, learn a new feature representation of the nodes with graph neural networks, and train a classifier based on the new features. Third, with real-world snapshot backup instances, we demonstrate the superior performance of the graph representation learning based approach over several baselines. Compared to the traditional full-scan approaches, our finding suggests that the vulnerable ransomware attack target identification can result in a more efficient ransomware detection process via focused inspection of most vulnerable data in the backups. Our method can also be easily integrated into existing ransomware detection systems for accelerated cyber resiliency.

This paper introduces the concept of Trustless AutoML, and proposes a framework that combines AutoML techniques with blockchain to fully decentralize the design and training process of machine learning models. The goal is to introduce full transparency and trust in the model design pipeline to establish a root-of-trust.

Multivariate time series prediction has applications in a wide variety of domains and is considered to be a very challenging task, especially when the variables have correlations and exhibit complex temporal patterns, such as seasonality and trend. Many existing methods suffer from strong statistical assumptions, numerical issues with high dimensionality, manual feature engineering efforts, and scalability. In this work, we present a novel deep learning architecture, known as Temporal Tensor Transformation Network, which transforms the original multivariate time series into a higher order of tensor through the proposed Temporal-Slicing Stack Transformation. This yields a new representation of the original multivariate time series, which enables the convolution kernel to extract complex and non-linear features as well as variable interactional signals from a relatively large temporal region. Experimental results show that Temporal Tensor Transformation Network outperforms several state-of-the-art methods on window-based predictions across various tasks. The proposed architecture also demonstrates robust prediction performance through an extensive sensitivity analysis.

IBM's Watson Internet of Things (IoT) initiative is the next step of IBM's exploration of new ways for systems and humans to learn and interact synergistically. As the IoT ecosystems continues mature, new devices and protocols are being developed. Our team is specifically interested in Bluetooth Low Energy (BLE) IoT beacons. As BLE beacons become more popular and widely adopted, interesting use cases have been developed. For example, Location-Based Systems that use BLE beacon technology to help smartphone users navigate through buildings. In this paper, we will provide examples of what has been done with BLE beacons, and an overview of a work-in-progress prototype that leverages BLE beacons. Our prototype, BeaSmart, is a mobile application that is being deployed and tested in the IBM Almaden Research Center (ARC). The BeaSmart application has four features that provide a more contextualized experience for employees and visitors as they navigate through ARC. We plan to use BeaSmart as a tool to evaluate the technical and human factors to consider when making a building smarter with BLE Beacons.

The unprecedented volume of data generated by contemporary business users and consumers has created enormous data storage and management challenges. In order to control data storage cost, many users are moving their data to online storage clouds, and applying capacity usage reducing data transformation techniques like de-duplication, compression, and transcoding. These give rise to several challenges, such as which cloud to choose, and what data transformation techniques to apply for optimizing cost. This paper presents an integrated storage service called 'iCostale' that reduces the overall cost of data storage through automatic selection and placement of users' data into one of many storage clouds. Further, it intelligently transforms data based on its type, access frequency, transformation overhead, and the cost model of the storage cloud providers. We demonstrate the efficacy of iCostale through a series of micro- and application-level benchmarks. Our experimental results show that, through intelligent data placement and transformation, 'iCostale' can reduce overall cost of data storage by more than 50%.

Automatic discovery and monitoring of IT resources is a critical part of enterprise systems management. In addition to ascertaining internal device configurations, this discovery process may also need to capture the capabilities, usage, connectivity, availability, and other information related to various IT components. Systems resource management (SRM) tools typically implement this discovery process using device specific APIs, custom agents and/or some standard-based solution (like WBEM and CIM). The discovery actions need to be systematically planned; an inefficient implementation or scheduling may easily take from a few minutes to several hours to complete in a large heterogeneous enterprise datacenter. This paper discusses the various challenges associated in discovering the configuration of a datacenter environment and presents an autonomic configuration monitoring middleware called Magellan that builds upon industry best practices and standards. Magellan reduces the overall discovery time by more than 50% in our micro-benchmark experiments as well as in a large datacenter configuration of a major financial organization.

Current data centers require storage capacities of hundreds of terabytes to petabytes. Time-critical applications such as online transactions processing depend on getting adequate performance from the storage subsystem: otherwise, they fail. It is difficult to provide predictable quality of service at this level of complexity, because I/O workloads are extremely variable and device behavior is poorly understood. Ensuring that unrelated but competing workloads do not affect each other's performance is still more difficult, and equally necessary. We present SLEDS (Service Level Enforcement Discipline for Storage), a distributed controller that provides statistical performance guarantees on a storage system built from commodity components. SLEDS can adaptively handle unpredictable workload variations so that each client continues to get the performance it needs even in the presence of misbehaving, competing peers. After evaluating the SLEDS on a heterogeneous mid-range storage system, we found that it is vastly superior to the raw system in its ability to provide performance guarantees, while only introducing a negligible overhead.

High-performance servers and high-speed networks will form the backbone of the infrastructure required for distributed multimedia information systems. A server for an interactive distributed multimedia system may require thousands of gigabytes of storage space and a high I/O bandwidth. In order to maximize the system utilization, and thus minimize the cost, it is essential that the load be balanced among each of the server's components, viz. the disks, the interconnection network and the scheduler. Many algorithms for maximizing retrieval capacity from the storage system have been proposed in the literature. This paper presents techniques for improving the server capacity by assigning media requests to the nodes of a server so as to balance the load on the interconnection network and the scheduling nodes. Five policies for request assignment-round-robin (RR), minimum link allocation (MLA), minimum contention allocation (MCA), weighted minimum link allocation (WMLA) and weighted minimum contention allocation (WMCA)-are developed. The performance of these policies on a server model developed by the authors (1995) is presented. We also consider the issue of file replication, and develop two schemes for storing the replicas: the parent group-based round-robin placement (PGBRRP) scheme, and the group-wide round-robin placement (GWRRP) scheme. The performance of the request assignment policies in the presence of file replication is presented.

A server for an interactive distributed multimedia system may require thousands of gigabytes of storage space and high I/O bandwidth. In order to maximize system utilization, and thus minimize cost, the load must be balanced among the server's disks, interconnection network and scheduler. Many algorithms for maximizing retrieval capacity from the storage system have been proposed. This paper presents techniques for improving server capacity by assigning media requests to the nodes of a server so as to balance the load on the interconnection network and the scheduling nodes. Five policies for request assignment are developed. The performance of these policies on a server model developed earlier is presented.

One of the key components of a multiuser multimedia-on-demand system is the data server. Digitalization of traditionally analog data such as video and audio, and the feasibility of obtaining network bandwidths above the gigabit-per-second range, are two important advances that have made possible the realization, in the near future, of interactive distributed multimedia systems. Secondary-to-main memory I/O technology has not kept pace with advances in networking, main memory, and CPU processing power. Consequently, the performance of the server has a direct bearing on the overall performance of such a system. In this paper, we present a highperformance solution to the I/O retrieval problem in a distributed multimedia system. We develop a model for the architecture of a server for such a system. Parallelism of data retrieval is achieved by striping the data across multiple disks. We present the algorithms for server operation when servicing a constant number of streams, as well as the admission control policy for accepting requests for new streams. The performance of any server ultimately depends on the data access patterns. Two modifications of the basic retrieval algorithm are presented to exploit data access patterns in order to improve system throughput and response time. Finally, we present preliminary performance results of these algorithms on the IBM SP1 and Intel Paragon parallel computers.