Plans to beam 5G signals to the public via drones that stay airborne for nine days at a time have been announced by two UK firms. They want to use antenna-equipped aircraft powered by hydrogen to deliver high-speed connectivity to wide areas. Stratospheric Platforms and Cambridge Consultants say they could cover the whole of the UK with about 60 drones. But telecoms analysts question whether the economic case for this scheme is quite as simple as it sounds. The Cambridge-based companies say they would run the service in partnership with existing mobile operators. They are already backed by Deutsche Telekom, which hopes to trial the technology in rural southern Germany in 2024.

The oil and gas industry are dealing with massive disruption on several fronts from increasing oil price volatility due to Coronavirus and the failed OPEC deal. Combined with complexity of a rapidly changing energy sector where digital technologies, the drive for greener energy and demand for more consumer-centric services are putting shareholder returns at risk and reconfiguring policy mandates, industry players are forced to make a significant re-evaluation of energy value chains, assets and operations. The way we produce and consume oil & gas is shifting. Renewable energy sources, such as wind and solar, are growing exponentially and are expected to account for nearly 70% of global electricity production in 2050. Transport is being electrified, with 50% of all new cars sold globally forecasted to be electric by 2033.

This article was first published in Branch magazine, an online collaboration between EIT Climate-KIC, Mozilla Foundation and Climate Action.tech A global pandemic has shocked the world, leading to thousands of deaths, economic hardship and profound social disruption. While we worry about our immediate needs, we should remember that another crisis is looming: climate change. The lockdown made it clear that staying at home and slowing down the economy is far from enough to solve the climate crisis. We're still emitting more than 80 per cent as much CO2 as normal, despite having 17 per cent fewer emissions compared to 2019 -- which is one of the most significant drops in recent years (1).

Static recommendation methods like collaborative filtering suffer from the inherent limitation of performing real-time personalization for cold-start users. Online recommendation, e.g., multi-armed bandit approach, addresses this limitation by interactively exploring user preference online and pursuing the exploration-exploitation (EE) trade-off. However, existing bandit-based methods model recommendation actions homogeneously. Specifically, they only consider the items as the arms, being incapable of handling the item attributes, which naturally provide interpretable information of user's current demands and can effectively filter out undesired items. In this work, we consider the conversational recommendation for cold-start users, where a system can both ask the attributes from and recommend items to a user interactively. This important scenario was studied in a recent work. However, it employs a hand-crafted function to decide when to ask attributes or make recommendations. Such separate modeling of attributes and items makes the effectiveness of the system highly rely on the choice of the hand-crafted function, thus introducing fragility to the system. To address this limitation, we seamlessly unify attributes and items in the same arm space and achieve their EE trade-offs automatically using the framework of Thompson Sampling. Our Conversational Thompson Sampling (ConTS) model holistically solves all questions in conversational recommendation by choosing the arm with the maximal reward to play. Extensive experiments on three benchmark datasets show that ConTS outperforms the state-of-the-art methods Conversational UCB (ConUCB) and Estimation-Action-Reflection model in both metrics of success rate and average number of conversation turns.

We investigate the use of data-driven likelihoods to bypass a key assumption made in many scientific analyses, which is that the true likelihood of the data is Gaussian. In particular, we suggest using the optimization targets of flow-based generative models, a class of models that can capture complex distributions by transforming a simple base distribution through layers of nonlinearities. We call these flow-based likelihoods (FBL). We analyze the accuracy and precision of the reconstructed likelihoods on mock Gaussian data, and show that simply gauging the quality of samples drawn from the trained model is not a sufficient indicator that the true likelihood has been learned. We nevertheless demonstrate that the likelihood can be reconstructed to a precision equal to that of sampling error due to a finite sample size. We then apply FBLs to mock weak lensing convergence power spectra, a cosmological observable that is significantly non-Gaussian (NG). We find that the FBL captures the NG signatures in the data extremely well, while other commonly used data-driven likelihoods, such as Gaussian mixture models and independent component analysis, fail to do so. This suggests that works that have found small posterior shifts in NG data with data-driven likelihoods such as these could be underestimating the impact of non-Gaussianity in parameter constraints. By introducing a suite of tests that can capture different levels of NG in the data, we show that the success or failure of traditional data-driven likelihoods can be tied back to the structure of the NG in the data. Unlike other methods, the flexibility of the FBL makes it successful at tackling different types of NG simultaneously. Because of this, and consequently their likely applicability across datasets and domains, we encourage their use for inference when sufficient mock data are available for training.

Dynamic Uncertain Causality Graph(DUCG) is a recently proposed model for diagnoses of complex systems. It performs well for industry system such as nuclear power plants, chemical system and spacecrafts. However, the variable state combination explosion in some cases is still a problem that may result in inefficiency or even disability in DUCG inference. In the situation of clinical diagnoses, when a lot of intermediate causes are unknown while the downstream results are known in a DUCG graph, the combination explosion may appear during the inference computation. Monte Carlo sampling is a typical algorithm to solve this problem. However, we are facing the case that the occurrence rate of the case is very small, e.g. $10^{-20}$, which means a huge number of samplings are needed. This paper proposes a new scheme based on conditional stochastic simulation which obtains the final result from the expectation of the conditional probability in sampling loops instead of counting the sampling frequency, and thus overcomes the problem. As a result, the proposed algorithm requires much less time than the DUCG recursive inference algorithm presented earlier. Moreover, a simple analysis of convergence rate based on a designed example is given to show the advantage of the proposed method. % In addition, supports for logic gate, logic cycles, and parallelization, which exist in DUCG, are also addressed in this paper. The new algorithm reduces the time consumption a lot and performs 3 times faster than old one with 2.7% error ratio in a practical graph for Viral Hepatitis B.

A group of Skoltech scientists developed machine learning (ML) algorithms that can teach artificial intelligence (AI) to determine oil viscosity based on nuclear magnetic resonance (NMR) data. The new method can come in handy for the petroleum industry and other sectors, which have to rely on indirect measurements to characterize a substance. The research was published in the Energy and Fuels journal. An important parameter of oil and petrochemicals, viscosity has implications for production and processing, while helping to better understand and model the natural processes in the reservoir. Standard oil viscosity assessment and monitoring techniques are very time and money consuming and sometimes technically unfeasible.

A group of Skoltech scientists have developed machine learning (ML) algorithms that can teach artificial intelligence (AI) to determine oil viscosity based on nuclear magnetic resonance (NMR) data. The new method can come in handy for the petroleum industry and other sectors that have to rely on indirect measurements to characterize a substance. The research was published in the Energy and Fuels journal. An important parameter of oil and petrochemicals, viscosity has implications for production and processing, while helping to better understand and model the natural processes in the reservoir. Standard oil viscosity assessment and monitoring techniques are very time and money consuming and sometimes technically unfeasible.

Deep-learning A.I. is the machine learning technology that powers everything from cutting-edge natural language processing to machine vision tools. It may also be powering climate change -- as a result of the massive energy consumption and CO2 emissions associated with training these deep-learning models. As the use of deep learning has exploded, so has the compute power associated with them, although this effect is rarely studied. Researchers at the University of Copenhagen's Department of Computer Science are working to change that, however. They've developed a tool called Carbontracker, which works out the energy consumption associated with deep-learning algorithms and then converts this into a prediction about CO2 emissions.

Rectifier (ReLU) deep neural networks (DNN) and their connection with piecewise affine (PWA) functions is analyzed. The paper is an effort to find and study the possibility of representing explicit state feedback policy of model predictive control (MPC) as a ReLU DNN, and vice versa. The complexity and architecture of DNN has been examined through some theorems and discussions. An approximate method has been developed for identification of input-space in ReLU net which results a PWA function over polyhedral regions. Also, inverse multiparametric linear or quadratic programs (mp-LP or mp-QP) has been studied which deals with reconstruction of constraints and cost function given a PWA function.