We often hear about robots working in controlled, predictable environments, like an assembly line or an operating theatre, but as a former field biologist, Claire is most interested in what happens when we take robots outside into wild environments, where everything — from the elements to the local wildlife — seems to be against you. In this special live recording for the UK Festival of Robotics, Claire chatted to Sophie Armanini (TU Munich / Imperial College London), Ben Scott-Robinson (Small Robot Company) and Matthew Ryan Tucker (University of Bristol). Find out more about the UK Festival of Robotics here: https://www.ukras.org/robotics-festival/ is an assistant professor at the Technical University of Munich, Germany, and a guest researcher at Imperial College London, where she previously worked as a research associate. She obtained her PhD from Delft University of Technology (the Netherlands), and has been a visiting researcher at Cranfield University and Cornell University (USA). Sophie’s research focuses on the dynamics and control of unconventional and bioinspired aerial vehicles, including flapping-wing and aerial-aquatic robots. Ben Scott-Robinson is an accomplished digital entrepreneur focused on geospatial and mobility technologies. Ben co-founded the in 2017 which endeavours to replace tractors with accurate, smart, lightweight robots. With 20 years experience in digital innovation, including the digital transformation of Ordnance Survey, Ben is also an experienced technology entrepreneur having founded two agencies, two consultancies, an app start-up and a phone for the blind. is a Physics PhD Student at the University of Bristol, researching the use of ground based mobile robots for mapping radiation and finding radiation hotspots. Last year he was part of a University field trip to the Chernobyl Exclusion zone, where he deployed a Boston Dynamics Spot robot in a variety of different locations, including beneath the New Safe Confinement at the Chernobyl Power plant.

Farming robots may hold a promise of a cleaner and safer agricultural future. Potential downsides may arise from the loss of much-needed jobs to the safety of those working alongside the robots. Therefore, a process of responsible development is required. In the project called Robot Highways, multiple uses for autonomous robots made by Saga Robotics are currently demonstrated on a fruit farm in southeast England. Robots are now treating plant diseases in fields and glasshouses and will be mapping terrain, picking, packing, and providing logistical support to workers over the course of the project.

Machine learning techniques have contributed to progress in science and technology fields ranging from health care to high-energy physics. Now, machine learning is poised to help accelerate the development of stronger alloys, particularly stainless steels, for America's thermal power generation fleet. Stronger materials are key to producing energy efficiently, resulting in economic and decarbonization benefits. "The use of ultra-high-strength steels in power plants dates back to the 1950s and has benefited from gradual improvements in the materials over time," says Osman Mamun, a postdoctoral research associate at Pacific Northwest National Laboratory (PNNL). "If we can find ways to speed up improvements or create new materials, we could see enhanced efficiency in plants that also reduces the amount of carbon emitted into the atmosphere."

Machine learning, a technique used in the artificial intelligence (AI) software behind self-driving cars and digital assistants, now enables scientists to address key challenges to harvesting on Earth the fusion energy that powers the sun and stars. The technique recently empowered physicist Dan Boyer of the U.S. Department of Energy's (DOE) Princeton Plasma Physics Laboratory (PPPL) to develop fast and accurate predictions for advancing control of experiments in the National Spherical Torus Experiment-Upgrade (NSTX-U)--the flagship fusion facility at PPPL that is currently under repair. Such AI predictions could improve the ability of NSTX-U scientists to optimize the components of experiments that heat and shape the magnetically confined plasma that fuels fusion experiments. By optimizing the heating and shaping of the plasma scientists will be able to more effectively study key aspects of the development of burning plasmas--largely self-heating fusion reactions--that will be critical for ITER, the international experiment under construction in France, and future fusion reactors. "This is a step toward what we should do to optimize the actuators," said Boyer, author of a paper in Nuclear Fusion that describes the machine learning tactics.

Recent years have seen stagnating improvements to branch predictor (BP) efficacy and a dearth of fresh ideas in branch predictor design, calling for fresh thinking in this area. This paper argues that looking at BP from the viewpoint of Reinforcement Learning (RL) facilitates systematic reasoning about, and exploration of, BP designs. We describe how to apply the RL formulation to branch predictors, show that existing predictors can be succinctly expressed in this formulation, and study two RL-based variants of conventional BPs.

Contextual Bandits find important use cases in various real-life scenarios such as online advertising, recommendation systems, healthcare, etc. However, most of the algorithms use flat feature vectors to represent context whereas, in the real world, there is a varying number of objects and relations among them to model in the context. For example, in a music recommendation system, the user context contains what music they listen to, which artists create this music, the artist albums, etc. Adding richer relational context representations also introduces a much larger context space making exploration-exploitation harder. To improve the efficiency of exploration-exploitation knowledge about the context can be infused to guide the exploration-exploitation strategy. Relational context representations allow a natural way for humans to specify knowledge owing to their descriptive nature. We propose an adaptation of Knowledge Infused Policy Gradients to the Contextual Bandit setting and a novel Knowledge Infused Policy Gradients Upper Confidence Bound algorithm and perform an experimental analysis of a simulated music recommendation dataset and various real-life datasets where expert knowledge can drastically reduce the total regret and where it cannot.

While multifidelity modeling provides a cost-effective way to conduct uncertainty quantification with computationally expensive models, much greater efficiency can be achieved by adaptively deciding the number of required high-fidelity (HF) simulations, depending on the type and complexity of the problem and the desired accuracy in the results. We propose a framework for active learning with multifidelity modeling emphasizing the efficient estimation of rare events. Our framework works by fusing a low-fidelity (LF) prediction with an HF-inferred correction, filtering the corrected LF prediction to decide whether to call the high-fidelity model, and for enhanced subsequent accuracy, adapting the correction for the LF prediction after every HF model call. The framework does not make any assumptions as to the LF model type or its correlations with the HF model. In addition, for improved robustness when estimating smaller failure probabilities, we propose using dynamic active learning functions that decide when to call the HF model. We demonstrate our framework using several academic case studies and two finite element (FE) model case studies: estimating Navier-Stokes velocities using the Stokes approximation and estimating stresses in a transversely isotropic model subjected to displacements via a coarsely meshed isotropic model. Across these case studies, not only did the proposed framework estimate the failure probabilities accurately, but compared with either Monte Carlo or a standard variance reduction method, it also required only a small fraction of the calls to the HF model.

Utility staff spend hours each week routinely reviewing corrective action, maintenance, and operational reports. What if we could create an artificial intelligence model that could review and understand a report that typically takes a human 10-15 minutes to read in just half a second? It would save considerable time, money and improve operations for plant staff and grid operators around the world. But building an AI solution for this task is an intricate and time-consuming process due to complex industry jargon and unique component and system names that are different from traditional word usage. The first step to begin creating an electric power industry dictionary to use with natural language processing (NLP) is to build and evaluate a smaller-scale proof of concept to evaluate its functionality for a subset of technical terms and phrases that will not be correctly interpreted by a traditional NLP tool using a standard off-the-shelf dictionary.

Synopsys' Manuel Mota shows how splitting SoCs into smaller dies for advanced packaging and using die-to-die interfaces to enable high bandwidth, low latency, and low power connectivity can benefit hyperscale data centers. Siemens EDA's Chris Spear explains the relationship between classes and objects in SystemVerilog with a handy visualization and notes the difference between SystemVerilog variables and class variables. Cadence's Paul McLellan listens in as Waylon Grange of Stage 2 Security demonstrates hacking the embedded software in a home's solar power controller and how it points to areas where embedded security still needs improvement. In a blog for Arm, Alp Acar of Boston University explains the concept of federated learning, a privacy-preserving paradigm to train machine learning models in a decentralized fashion, leaving a user's data on the device. In a video, Infineon's Thomas Aichinger dives into how to make the gate oxide of SiC MOSFETs more reliable in the field through voltage screening and marathon stress tests.

This paper describes a new library for learning Bayesian networks from data containing discrete and continuous variables (mixed data). In addition to the classical learning methods on discretized data, this library proposes its algorithm that allows structural learning and parameters learning from mixed data without discretization since data discretization leads to information loss. This algorithm based on mixed MI score function for structural learning, and also linear regression and Gaussian distribution approximation for parameters learning. The library also offers two algorithms for enumerating graph structures - the greedy Hill-Climbing algorithm and the evolutionary algorithm. Thus the key capabilities of the proposed library are as follows: (1) structural and parameters learning of a Bayesian network on discretized data, (2) structural and parameters learning of a Bayesian network on mixed data using the MI mixed score function and Gaussian approximation, (3) launching learning algorithms on one of two algorithms for enumerating graph structures - Hill-Climbing and the evolutionary algorithm. Since the need for mixed data representation comes from practical necessity, the advantages of our implementations are evaluated in the context of solving approximation and gap recovery problems on synthetic data and real datasets.