Accurate low dimension chemical kinetic models for methane are an essential component in the design of efficient gas turbine combustors. Kinetic models coupled to computational fluid dynamics (CFD) provide quick and efficient ways to test the effect of operating conditions, fuel composition and combustor design compared to physical experiments. However, detailed chemical kinetic models are too computationally expensive for use in CFD. We propose a novel data orientated three-step methodology to produce compact models that replicate a target set of detailed model properties to a high fidelity. In the first step, a reduced kinetic model is obtained by removing all non-essential species from the detailed model containing 118 species using path flux analysis (PFA). It is then numerically optimised to replicate the detailed model's prediction in two rounds; First, to selected species (OH,H,CO and CH4) profiles in perfectly stirred reactor (PSR) simulations and then re-optimised to the detailed model's prediction of the laminar flame speed. This is implemented by a purposely developed Machine Learned Optimisation of Chemical Kinetics (MLOCK) algorithm. The MLOCK algorithm systematically perturbs all three Arrhenius parameters for selected reactions and assesses the suitability of the new parameters through an objective error function which quantifies the error in the compact model's calculation of the optimisation target. This strategy is demonstrated through the production of a 19 species and a 15 species compact model for methane/air combustion. Both compact models are validated across a range of 0D and 1D calculations across both lean and rich conditions and shows good agreement to the parent detailed mechanism. The 15 species model is shown to outperform the current state-of-art models in both accuracy and range of conditions the model is valid over.

With the rapid development of AI and robotics, transporting a large swarm of networked robots has foreseeable applications in the near future. Existing research in swarm robotics has mainly followed a bottom-up philosophy with predefined local coordination and control rules. However, it is arduous to verify the global requirements and analyze their performance. This motivates us to pursue a top-down approach, and develop a provable control strategy for deploying a robotic swarm to achieve a desired global configuration. Specifically, we use mean-field partial differential equations (PDEs) to model the swarm and control its mean-field density (i.e., probability density) over a bounded spatial domain using mean-field feedback. The presented control law uses density estimates as feedback signals and generates corresponding velocity fields that, by acting locally on individual robots, guide their global distribution to a target profile. The design of the velocity field is therefore centralized, but the implementation of the controller can be fully distributed -- individual robots sense the velocity field and derive their own velocity control signals accordingly. The key contribution lies in applying the concept of input-to-state stability (ISS) to show that the perturbed closed-loop system (a nonlinear and time-varying PDE) is locally ISS with respect to density estimation errors. The effectiveness of the proposed control laws is verified using agent-based simulations.

This study presents PRISM, a probabilistic simplex component analysis approach to identifying the vertices of a data-circumscribing simplex from data. The problem has a rich variety of applications, the most notable being hyperspectral unmixing in remote sensing and non-negative matrix factorization in machine learning. PRISM uses a simple probabilistic model, namely, uniform simplex data distribution and additive Gaussian noise, and it carries out inference by maximum likelihood. The inference model is sound in the sense that the vertices are provably identifiable under some assumptions, and it suggests that PRISM can be effective in combating noise when the number of data points is large. PRISM has strong, but hidden, relationships with simplex volume minimization, a powerful geometric approach for the same problem. We study these fundamental aspects, and we also consider algorithmic schemes based on importance sampling and variational inference. In particular, the variational inference scheme is shown to resemble a matrix factorization problem with a special regularizer, which draws an interesting connection to the matrix factorization approach. Numerical results are provided to demonstrate the potential of PRISM.

We propose HyperDynamics, a dynamics meta-learning framework that conditions on an agent's interactions with the environment and optionally its visual observations, and generates the parameters of neural dynamics models based on inferred properties of the dynamical system. Physical and visual properties of the environment that are not part of the low-dimensional state yet affect its temporal dynamics are inferred from the interaction history and visual observations, and are implicitly captured in the generated parameters. We test HyperDynamics on a set of object pushing and locomotion tasks. It outperforms existing dynamics models in the literature that adapt to environment variations by learning dynamics over high dimensional visual observations, capturing the interactions of the agent in recurrent state representations, or using gradient-based meta-optimization. We also show our method matches the performance of an ensemble of separately trained experts, while also being able to generalize well to unseen environment variations at test time. We attribute its good performance to the multiplicative interactions between the inferred system properties--captured in the generated parameters-- and the low-dimensional state representation of the dynamical system. Humans learn dynamics models that predict results of their interactions with the environment, and use such predictions for selecting actions to achieve intended goals (Miall & Wolpert, 1996; Haruno et al., 1999). These models capture intuitive physics and mechanics of the world and are remarkably versatile: they are expressive and can be applied to all kinds of environments that we encounter in our daily lives, with varying dynamics and diverse visual and physical properties. In addition, humans do not consider these models fixed over the course of interaction; we observe how the environment behaves in response to our actions and quickly adapt our model for the situation at hand based on new observations. Let us consider the scenario of moving an object on the ground. We can infer how heavy the object is by simply looking at it, and we can then decide how hard to push.

Real-world decision-making tasks are generally complex, requiring trade-offs between multiple, often conflicting, objectives. Despite this, the majority of research in reinforcement learning and decision-theoretic planning either assumes only a single objective, or that multiple objectives can be adequately handled via a simple linear combination. Such approaches may oversimplify the underlying problem and hence produce suboptimal results. This paper serves as a guide to the application of multi-objective methods to difficult problems, and is aimed at researchers who are already familiar with single-objective reinforcement learning and planning methods who wish to adopt a multi-objective perspective on their research, as well as practitioners who encounter multi-objective decision problems in practice. It identifies the factors that may influence the nature of the desired solution, and illustrates by example how these influence the design of multi-objective decision-making systems for complex problems.

Machine learning on sets towards sequential output is an important and ubiquitous task, with applications ranging from language modelling and meta-learning to multi-agent strategy games and power grid optimization. Combining elements of representation learning and structured prediction, its two primary challenges include obtaining a meaningful, permutation invariant set representation and subsequently utilizing this representation to output a complex target permutation. This paper provides a comprehensive introduction to the field as well as an overview of important machine learning methods tackling both of these key challenges, with a detailed qualitative comparison of selected model architectures.

In December 2020 we launched a focus series AI for Good: UN sustainable development goals (SDGs). Each month we pick a different sustainable development goal (SDG) and highlight work in that area. February was the turn of UN SDG number 13: climate action. In this summary article we highlight some of work at the intersection of AI and climate science. Climate Change AI (CCAI) is a volunteer-led effort bringing together people from academia, industry, and the public sector.

Six months after the Trump administration dealt a crushing blow to Huawei Technologies Co.'s smartphone business, the Chinese telecommunications giant is turning to less glamorous alternatives that may eventually offset the decline of its biggest revenue contributor. Among its newest customers is a fish farm in eastern China that's twice the size of New York's Central Park. The farm is covered with tens of thousands of solar panels outfitted with Huawei's inverters to shield its fish from excessive sunlight while generating power. About 370 miles to the west in coal-rich Shanxi province, wireless sensors and cameras deep beneath the earth monitor oxygen levels and potential machine malfunctions in mine pit -- all supplied by the tech titan. And next month, a shiny new electric car featuring its lidar sensor will debut at China's largest auto show.

There is growing interest in data-driven weather prediction (DDWP), for example using convolutional neural networks such as U-NETs that are trained on data from models or reanalysis. Here, we propose 3 components to integrate with commonly used DDWP models in order to improve their physical consistency and forecast accuracy. These components are 1) a deep spatial transformer added to the latent space of the U-NETs to preserve a property called equivariance, which is related to correctly capturing rotations and scalings of features in spatio-temporal data, 2) a data-assimilation (DA) algorithm to ingest noisy observations and improve the initial conditions for next forecasts, and 3) a multi-time-step algorithm, which combines forecasts from DDWP models with different time steps through DA, improving the accuracy of forecasts at short intervals. To show the benefit/feasibility of each component, we use geopotential height at 500~hPa (Z500) from ERA5 reanalysis and examine the short-term forecast accuracy of specific setups of the DDWP framework. Results show that the equivariance-preserving networks (U-STNs) clearly outperform the U-NETs, for example improving the forecast skill by $45\%$. Using a sigma-point ensemble Kalman (SPEnKF) algorithm for DA and U-STN as the forward model, we show that stable, accurate DA cycles are achieved even with high observation noise. The DDWP+DA framework substantially benefits from large ($O(1000)$) ensembles that are inexpensively generated with the data-driven forward model in each DA cycle. The multi-time-step DDWP+DA framework also shows promises, e.g., it reduces the average error by factors of 2-3.

Picture of a polymer:non-fullerene acceptor solar cell device, for which the polymer was designed by machine learning. Researchers at Osaka University use machine learning to design and virtually test molecules for organic solar cells, which can lead to higher efficiency functional materials for renewable energy applications. Osaka University researchers employed machine learning to design new polymers for use in photovoltaic devices. After virtually screening over 200,000 candidate materials, they synthesized one of the most promising and found its properties were consistent with their predictions. This work may lead to a revolution in the way functional materials are discovered.