With the increasing computational power of current supercomputers, the size of data produced by scientific simulations is rapidly growing. To reduce the storage footprint and facilitate scalable post-hoc analyses of such scientific data sets, various data reduction/summarization methods have been proposed over the years. Different flavors of sampling algorithms exist to sample the high-resolution scientific data, while preserving important data properties required for subsequent analyses. However, most of these sampling algorithms are designed for univariate data and cater to post-hoc analyses of single variables. In this work, we propose a multivariate sampling strategy which preserves the original variable relationships and enables different multivariate analyses directly on the sampled data. Our proposed strategy utilizes principal component analysis to capture the variance of multivariate data and can be built on top of any existing state-of-the-art sampling algorithms for single variables. In addition, we also propose variants of different data partitioning schemes (regular and irregular) to efficiently model the local multivariate relationships. Using two real-world multivariate data sets, we demonstrate the efficacy of our proposed multivariate sampling strategy with respect to its data reduction capabilities as well as the ease of performing efficient post-hoc multivariate analyses.

Energy theft causes large economic losses to utility companies around the world. In recent years, energy theft detection approaches based on machine learning (ML) techniques, especially neural networks, become popular in the research literature and achieve state-of-the-art detection performance. However, in this work, we demonstrate that the well-perform ML models for energy theft detection are highly vulnerable to adversarial attacks. In particular, we design an adversarial measurement generation algorithm that enables the attacker to report extremely low power consumption measurements to the utilities while bypassing the ML energy theft detection. We evaluate our approach with three kinds of neural networks based on a real-world smart meter dataset. The evaluation result demonstrates that our approach can significantly decrease the ML models' detection accuracy, even for black-box attackers.

Recent oil spills in Russia and Mauritius have shown that the industry still needs better methods for cleaning up accidents. Researchers are working on some unlikely-sounding solutions, including oil-absorbing wood chips, a solar-powered robot and a reusable sponge. The oil industry is controlled by large companies and their suppliers, which together have often been the cause of spills, but university researchers and small firms are playing a key role in promoting new ways to clean up. Researchers at Northwestern University have developed a reusable sponge coated in a mixture containing iron and carbon that can absorb 30 times its weight in oil. The sponge, similar to sponges in everyday items such as furniture cushions and packaging, has attracted interest for further testing from several major oil companies, according to the researchers.

Extremely energy-efficient artificial intelligence is now closer to reality after a study by UCL researchers found a way to improve the accuracy of a brain-inspired computing system. The system, which uses memristors to create artificial neural networks, is at least 1,000 times more energy efficient than conventional transistor-based AI hardware, but has until now been more prone to error. Existing AI is extremely energy-intensive--training one AI model can generate 284 tons of carbon dioxide, equivalent to the lifetime emissions of five cars. Replacing the transistors that make up all digital devices with memristors, a novel electronic device first built in 2008, could reduce this to a fraction of a ton of carbon dioxide--equivalent to emissions generated in an afternoon's drive. Since memristors are so much more energy-efficient than existing computing systems, they can potentially pack huge amounts of computing power into hand-held devices, removing the need to be connected to the Internet.

Abstract--Smart Manufacturing refers to optimization techniques that are implemented in production operations by utilizing advanced analytics approaches. With the widespread increase in deploying Industrial Internet of Things (IIoT) sensors in manufacturing processes, there is a progressive need for optimal and effective approaches to data management. Embracing Machine Learning and Artificial Intelligence to take advantage of manufacturing data can lead to efficient and intelligent automation. In this paper, we conduct a comprehensive analysis based on Evolutionary Computing and Deep Learning algorithms toward making semiconductor manufacturing smart. Computing, in manufacturing, provides access to valuable data at different levels, i.e., manufacturing enterprise, manufacturing equipment, and manufacturing processes. VER recent decades, the manufacturing industry witnessed tremendous advances in the form of four major manufacturing insights. Manufacturing, then, can be controlled paradigm shifts. In the latest industrial revolution, Industry 4.0, by leading-edge CI and Artificial Intelligence (AI), and tasks manufacturing has embraced the Industrial Internet of Things are modelled based on experimental observations, to enhance (IIoT) [1]-[3] and Machine Learning (ML) to enable machinery productivity while reducing costs. In doing so, it is of so can make industry processes smart. Broadly speaking, Smart great importance to identify which factors play a pivotal role in Manufacturing (SM) can be defined as a data-driven approach process outcomes.

This snapshot of turbulence density and vorticity from a simulation using SDSC's'Comet' supercomputer illustrates a notable physics concept: the formation of zonal (i.e. For more than four decades, UC San Diego Professor of Physics Patrick H. Diamond and his research group have been advancing fundamental concepts in plasma physics, which is an important aspect of furthering advancements in fusion energy. Most recently, Diamond worked with graduate student Robin Heinonen on a model reduction study that used the Comet supercomputer at the San Diego Supercomputer Center at the University of California San Diego to show how machine learning produced a novel model for plasma turbulence. Diamond and Heinonen say that advances in machine learning, such as new deep learning techniques, have provided them with new tools to better understand the self-organization process that emerges from what the researchers term as a seemingly chaotic process. "Turbulence and its transport is chaotic in a sense, but this chaos is ordered and constrained," said Heinonen, who co-authored Turbulence Model Reduction by Deep Learning with Diamond in the academic journal entitled Physical Review E. "Moreover, in certain turbulent systems, the chaos conspires to spontaneously form large, long-lived coherent structures and in many cases, we only have a tenuous understanding of why and now. There are definitely aspects of structure formation and self-organization which we do understand, but it's still an active area of research."

Model-based reinforcement learning approaches add explicit domain knowledge to agents in hopes of improving the sample-efficiency in comparison to model-free agents. However, in practice model-based methods are unable to achieve the same asymptotic performance on challenging continuous control tasks due to the complexity of learning and controlling an explicit world model. In this paper we investigate the stochastic value gradient (SVG), which is a well-known family of methods for controlling continuous systems which includes model-based approaches that distill a model-based value expansion into a model-free policy. We consider a variant of the model-based SVG that scales to larger systems and uses 1) an entropy regularization to help with exploration, 2) a learned deterministic world model to improve the short-horizon value estimate, and 3) a learned model-free value estimate after the model's rollout. This SVG variation captures the model-free soft actor-critic method as an instance when the model rollout horizon is zero, and otherwise uses short-horizon model rollouts to improve the value estimate for the policy update. We surpass the asymptotic performance of other model-based methods on the proprioceptive MuJoCo locomotion tasks from the OpenAI gym, including a humanoid. We notably achieve these results with a simple deterministic world model without requiring an ensemble.

Time-series forecasting is one of the most active research topics in predictive analysis. A still open gap in that literature is that statistical and ensemble learning approaches systematically present lower predictive performance than deep learning methods as they generally disregard the data sequence aspect entangled with multivariate data represented in more than one time series. Conversely, this work presents a novel neural network architecture for time-series forecasting that combines the power of graph evolution with deep recurrent learning on distinct data distributions; we named our method Recurrent Graph Evolution Neural Network (ReGENN). The idea is to infer multiple multivariate relationships between co-occurring time-series by assuming that the temporal data depends not only on inner variables and intra-temporal relationships (i.e., observations from itself) but also on outer variables and inter-temporal relationships (i.e., observations from other-selves). An extensive set of experiments was conducted comparing ReGENN with dozens of ensemble methods and classical statistical ones, showing sound improvement of up to 64.87% over the competing algorithms. Furthermore, we present an analysis of the intermediate weights arising from ReGENN, showing that by looking at inter and intra-temporal relationships simultaneously, time-series forecasting is majorly improved if paying attention to how multiple multivariate data synchronously evolve.

Modularity is a central principle throughout the design process for cyber-physical systems. Modularity reduces complexity and increases reuse of behavior. In this paper we pose and answer the following question: how can we identify independent `modules' within the structure of reactive control architectures? To this end, we propose a graph-structured control architecture we call a decision structure, and show how it generalises some reactive control architectures which are popular in Artificial Intelligence (AI) and robotics, specifically Teleo-Reactive programs (TRs), Decision Trees (DTs), Behavior Trees (BTs) and Generalised Behavior Trees ($k$-BTs). Inspired by the definition of a module in graph theory, we define modules in decision structures and show how each decision structure possesses a canonical decomposition into its modules. We can naturally characterise each of the BTs, $k$-BTs, DTs and TRs by properties of their module decomposition. This allows us to recognise which decision structures are equivalent to each of these architectures in quadratic time. Our proposed concept of modules extends to formal verification, under any verification scheme capable of verifying a decision structure. Namely, we prove that a modification to a module within a decision structure has no greater flow-on effects than a modification to an individual action within that structure. This enables verification on modules to be done locally and hierarchically, where structures can be verified and then repeatedly locally modified, with modules replaced by modules while preserving correctness. To illustrate the findings, we present an example of a solar-powered drone controlled by a decision structure. We use a Linear Temporal Logic-based verification scheme to verify the correctness of this structure, and then show how one can modify modules while preserving its correctness.

Wide area networking infrastructures (WANs), particularly science and research WANs, are the backbone for moving large volumes of scientific data between experimental facilities and data centers. With demands growing at exponential rates, these networks are struggling to cope with large data volumes, real-time responses, and overall network performance. Network operators are increasingly looking for innovative ways to manage the limited underlying network resources. Forecasting network traffic is a critical capability for proactive resource management, congestion mitigation, and dedicated transfer provisioning. To this end, we propose a nonautoregressive graph-based neural network for multistep network traffic forecasting. Specifically, we develop a dynamic variant of diffusion convolutional recurrent neural networks to forecast traffic in research WANs. We evaluate the efficacy of our approach on real traffic from ESnet, the U.S. Department of Energy's dedicated science network. Our results show that compared to classical forecasting methods, our approach explicitly learns the dynamic nature of spatiotemporal traffic patterns, showing significant improvements in forecasting accuracy. Our technique can surpass existing statistical and deep learning approaches by achieving approximately 20% mean absolute percentage error for multiple hours of forecasts despite dynamic network traffic settings.