Marianne Winslett

Modern power grids are undergoing significant changes driven by information and communication technologies (ICTs), and evolving into smart grids with higher efficiency and lower operation cost. Using ICTs, however, comes with an inevitable side effect that makes the power system more vulnerable to cyber attacks. In this paper, we propose a self-supervised learning-based framework to detect and identify various types of cyber attacks. Different from existing approaches, the proposed framework does not rely on large amounts of well-curated labeled data but makes use of the massive unlabeled data in the wild which are easily accessible. Specifically, the proposed framework adopts the BERT model from the natural language processing domain and learns generalizable and effective representations from the unlabeled sensing data, which capture the distinctive patterns of different attacks. Using the learned representations, together with a very small amount of labeled data, we can train a task-specific classifier to detect various types of cyber attacks. Experiment results in a 3-area power grid system with 37 buses demonstrate the superior performance of our framework over existing approaches, especially when a very limited amount of labeled data are available. We believe such a framework can be easily adopted to detect a variety of cyber attacks in other power grid scenarios.

Pre-trained Transformer-based models have achieved state-of-the-art performance for various Natural Language Processing (NLP) tasks. However, these models often have billions of parameters, and thus are too resource- hungry and computation-intensive to suit low- capability devices or applications with strict latency requirements. One potential remedy for this is model compression, which has attracted considerable research attention. Here, we summarize the research in compressing Transformers, focusing on the especially popular BERT model. In particular, we survey the state of the art in compression for BERT, we clarify the current best practices for compressing large-scale Transformer models, and we provide insights into the workings of various methods. Our categorization and analysis also shed light on promising future research directions for achieving lightweight, accurate, and generic NLP models.

Pre-trained Transformer-based models have achieved state-of-the-art performance for various Natural Language Processing (NLP) tasks. However, these models often have billions of parameters, and, thus, are too resource-hungry and computation-intensive to suit low-capability devices or applications with strict latency requirements. One potential remedy for this is model compression, which has attracted a lot of research attention. Here, we summarize the research in compressing Transformers, focusing on the especially popular BERT model. In particular, we survey the state of the art in compression for BERT, we clarify the current best practices for compressing large-scale Transformer models, and we provide insights into the workings of various methods. Our categorization and analysis also shed light on promising future research directions for achieving lightweight, accurate, and generic NLP models.

Control and communication technologies are key building blocks of cyber-physical systems (CPSes) that can improve the efficiency of the physical processes. However, they also make a CPS vulnerable to cyberattacks that can cause disruptions or even severe damage. This article focuses on one particular type of CPS cyberattack, namely the time delay attack (TDA), which exploits vulnerabilities in the communication channels to cause potentially serious harm to the system. Much work proposed for TDA detection is tested offline only and under strong assumptions. In order to construct a practical solution to deal with real-world scenarios, we propose a deep learning-based method to detect and characterize TDA. Specifically, we design a hierarchical long short-term memory model to process raw data streams from relevant CPS sensors online and continually monitor embedded signals in the data to detect and characterize the attack. Moreover, various strategies of interpreting the outputs of the model are proposed, which allow the user to tune the performance based on different objectives. We evaluate our model on two representative types of CPS, namely power plant control system (PPCS) and automatic generation control (AGC).Code and dataset can be found at: https://github.com/prakharg24/tda For TDA detection, our solution achieves an accuracy of 92% in PPCS, compared with 81% by random forests (RFs) and 72% by k-nearest neighbours (kNNs). For AGC, our solution achieves 98% accuracy, compared with 74% by RFs and 71% by kNNs. It also reduces the mean absolute error in the delay value characterization from about six to two seconds in the PPCS, and from about three seconds to half a second in the AGC, with about 3x to 4x shorter reaction latency in both systems.

The pervasive adoption of Deep Learning (DL) and Graph Processing (GP) makes it a de facto requirement to build large-scale clusters of heterogeneous accelerators including GPUs and FPGAs. The OpenCL programming framework can be used on the individual nodes of such clusters but is not intended for deployment in a distributed manner. Fortunately, the original OpenCL semantics naturally fit into the programming environment of heterogeneous clusters. In this paper, we propose a heterogeneity-aware OpenCL-like (HaoCL) programming framework to facilitate the programming of a wide range of scientific applications including DL and GP workloads on large-scale heterogeneous clusters. With HaoCL, existing applications can be directly deployed on heterogeneous clusters without any modifications to the original OpenCL source code and without awareness of the underlying hardware topologies and configurations. Our experiments show that HaoCL imposes a negligible overhead in a distributed environment, and provides near-liner speedups on standard benchmarks when computation or data size exceeds the capacity of a single node. The system design and the evaluations are presented in this demo paper.

Data-driven models are becoming essential parts in modern mechanical systems, commonly used to capture the behavior of various equipment and varying environmental characteristics. Despite the advantages of these data-driven models on excellent adaptivity to high dynamics and aging equipment, they are usually hungry to massive labels over historical data, mostly contributed by human engineers at an extremely high cost. The label demand is now the major limiting factor to modeling accuracy, hindering the fulfillment of visions for applications. Fortunately, domain adaptation enhances the model generalization by utilizing the labelled source data as well as the unlabelled target data and then we can reuse the model on different domains. However, the mainstream domain adaptation methods cannot achieve ideal performance on time series data, because most of them focus on static samples and even the existing time series domain adaptation methods ignore the properties of time series data, such as temporal causal mechanism. In this paper, we assume that causal mechanism is invariant and present our Causal Mechanism Transfer Network(CMTN) for time series domain adaptation. By capturing and transferring the dynamic and temporal causal mechanism of multivariate time series data and alleviating the time lags and different value ranges among different machines, CMTN allows the data-driven models to exploit existing data and labels from similar systems, such that the resulting model on a new system is highly reliable even with very limited data. We report our empirical results and lessons learned from two real-world case studies, on chiller plant energy optimization and boiler fault detection, which outperforms the existing state-of-the-art method.

Welcome to this installment of ACM Sigmod Records series of interviews with distinguished members of the database community. I'm Marianne Winslett and today we're at the 2017 SIGMOD and PODS Conference in Chicago. I have here with me Peter Bailis who's a professor at Stanford University. Peter won the 2017 ACM SIGMOD Jim Gray Dissertation award for his thesis entitled "Coordination Avoidance in Distributed Databases." Peter's advisors were Joseph Hellerstein, Ion Stoica, and Ali Ghodsi at Berkeley.

We present LiveTraj, a novel system for tracking trajectories in a live video stream in real time, backed by a cloud platform. Although trajectory tracking is a well-studied topic in computer vision, so far most attention has been devoted to improving the accuracy of trajectory tracking, rather than the efficiency. To our knowledge, LiveTraj is the first that achieves real-time efficiency in trajectory tracking, which can be a key enabler in many important applications such as video surveillance, action recognition and robotics. LiveTraj is based on a state-of-the-art approach to (offline) trajectory tracking; its main innovation is to adapt this base solution to run on an elastic cloud platform to achieve real-time tracking speed at an affordable cost. The video demo shows the offline base solution and LiveTraj side by side, both running on a video stream containing human actions. Besides demonstrating the real-time efficiency of LiveTraj, our video demo also exhibits important system parameters to the audience such as latency and cloud resource usage for different components of the system. Further, if the conference venue provides sufficiently fast Internet connection to our cloud platform, we also plan to demonstrate LiveTraj on-site, during which we will show LiveTraj identifying and tracking trajectories from a live video stream captured by a camera.

A cloud-based data stream management system (DSMS) handles fast data by utilizing the massively parallel processing capabilities of the underlying platform. An important property of such a DSMS is elasticity, meaning that nodes can be dynamically added to or removed from an application to match the latter's workload, which may fluctuate in an unpredictable manner. For an application involving stateful operations such as aggregates, the addition / removal of nodes necessitates the migration of operator states. Although the importance of migration has been recognized in existing systems, two key problems remain largely neglected, namely how to migrate and what to migrate, i.e., the migration mechanism that reduces synchronization overhead and result delay during migration, and the selection of the optimal task assignment that minimizes migration costs. Consequently, migration in current systems typically incurs a high spike in result delay caused by expensive synchronization barriers and suboptimal task assignments. Motivated by this, we present the first comprehensive study on efficient operator states migration, and propose designs and algorithms that enable live, progressive, and optimized migrations. Extensive experiments using real data justify our performance claims.