Current deep neural networks are highly overparameterized (up to billions of connection weights) and nonlinear. Yet they can fit data almost perfectly through variants of gradient descent algorithms and achieve unexpected levels of prediction accuracy without overfitting. These are formidable results that escape the bias-variance predictions of statistical learning and pose conceptual challenges for non-convex optimization. In this paper, we use methods from statistical physics of disordered systems to analytically study the computational fallout of overparameterization in nonconvex neural network models. As the number of connection weights increases, we follow the changes of the geometrical structure of different minima of the error loss function and relate them to learning and generalisation performance. We find that there exist a gap between the SAT/UNSAT interpolation transition where solutions begin to exist and the point where algorithms start to find solutions, i.e. where accessible solutions appear. This second phase transition coincides with the discontinuous appearance of atypical solutions that are locally extremely entropic, i.e., flat regions of the weight space that are particularly solution-dense and have good generalization properties. Although exponentially rare compared to typical solutions (which are narrower and extremely difficult to sample), entropic solutions are accessible to the algorithms used in learning. We can characterize the generalization error of different solutions and optimize the Bayesian prediction, for data generated from a structurally different network. Numerical tests on observables suggested by the theory confirm that the scenario extends to realistic deep networks.

Rotary machine breakdown detection systems are outdated and dependent upon routine testing to discover faults. This is costly and often reactive in nature. Real-time monitoring offers a solution for detecting faults without the need for manual observation. However, manual interpretation for threshold anomaly detection is often subjective and varies between industrial experts. This approach is ridged and prone to a large number of false positives. To address this issue, we propose a Machine Learning (ML) approach to model normal working operation and detect anomalies. The approach extracts key features from signals representing known normal operation to model machine behaviour and automatically identify anomalies. The ML learns generalisations and generates thresholds based on fault severity. This provides engineers with a traffic light system were green is normal behaviour, amber is worrying and red signifies a machine fault. This scale allows engineers to undertake early intervention measures at the appropriate time. The approach is evaluated on windowed real machine sensor data to observe normal and abnormal behaviour. The results demonstrate that it is possible to detect anomalies within the amber range and raise alarms before machine failure.

We look into Generative Adversarial Network (GAN), its prevalent variants and applications in a number of sectors. GANs combine two neural networks that compete against one another using zero-sum game theory, allowing them to create much crisper and discrete outputs. GANs can be used to perform image processing, video generation and prediction, among other computer vision applications. GANs can also be utilised for a variety of science-related activities, including protein engineering, astronomical data processing, remote sensing image dehazing, and crystal structure synthesis. Other notable fields where GANs have made gains include finance, marketing, fashion design, sports, and music. Therefore in this article we provide a comprehensive overview of the applications of GANs in a wide variety of disciplines. We first cover the theory supporting GAN, GAN variants, and the metrics to evaluate GANs. Then we present how GAN and its variants can be applied in twelve domains, ranging from STEM fields, such as astronomy and biology, to business fields, such as marketing and finance, and to arts, such as music. As a result, researchers from other fields may grasp how GANs work and apply them to their own study. To the best of our knowledge, this article provides the most comprehensive survey of GAN's applications in different fields.

Geometric deep learning has gained tremendous attention in both academia and industry due to its inherent capability of representing arbitrary structures. Due to exponential increase in interest towards renewable sources of energy, especially wind energy, accurate wind speed forecasting has become very important.. In this paper, we propose a novel deep learning architecture, Multi Scale Graph Wavenet for wind speed forecasting. It is based on a graph convolutional neural network and captures both spatial and temporal relationships in multivariate time series weather data for wind speed forecasting. We especially took inspiration from dilated convolutions, skip connections and the inception network to capture temporal relationships and graph convolutional networks for capturing spatial relationships in the data. We conducted experiments on real wind speed data measured at different cities in Denmark and compared our results with the state-of-the-art baseline models. Figure 1.Weather stations in Denmark [7] Our novel architecture outperformed the state-of-the-art methods on wind speed forecasting for multiple forecast horizons by 4-5%.

We present the winning strategy of an electricity demand forecasting competition. This competition was organized to design new forecasting methods for unstable periods such as the one starting in Spring 2020. We rely on state-space models to adapt standard statistical and machine learning models. We claim that it achieves the right compromise between two extremes. On the one hand, purely time-series models such as autoregressives are adaptive in essence but fail to capture dependence to exogenous variables. On the other hand, machine learning methods allow to learn complex dependence to explanatory variables on a historical data set but fail to forecast non-stationary data accurately. The evaluation period of the competition was the occasion of trial and error and we put the focus on the final forecasting procedure. In particular, it was at the same time that a recent algorithm was designed to adapt the variances of a state-space model and we present the results of the final version only. We discuss day-today predictions nonetheless.

It's no secret that artificial intelligence and technology has been developing quickly in recent times, with applications such as CAPTCHA that prevent bots from accessing sites, thermostats that adapt to our daily schedules or even algorithms that could choose potential vacation destinations for us. But what if machine learning could be used beyond niche or individual contexts? Taking artificial intelligence a step further and implementing it into our cities and infrastructures has the potential for improving operating efficiencies, aiding in sustainability efforts, urban planning and more. Below, we'll be exploring a few of the ways that machine learning can be used for improving our cities and making them smarter overall. Often times, we will hear from various forms of media that we should be aiming to reduce our individual and collective carbon footprints – however, how can cities and organizations accurately calculate their contributions to carbon emissions?

Iron Ox is a leading agriculture tech startup by Brandon Alexander based in Silicon Valley, California. Recently, the firm announced that it had secured about $53 Million in the latest "Series C" funding round. The funding round was led by the new investor, Breakthrough Energy Ventures that strives for innovation in sustainable energy and other technologies that assist in reducing greenhouse gas emissions. Multiple world's top business leaders (about nineteen) supported the Series C funding of Iron Ox with a dedication to achieving net-zero emissions by 2050. Iron Ox began autonomous farming in 2018 that plants produce in proprietary greenhouses created from the base up to reduce agriculture's environmental effect.

Have you ever experienced using a particular year for crop signature analysis and the minute you extend that analysis to a different year, it fails to provide the same insights or you just cannot replicate the results you had derived from the above experiment? I have been working on a machine learning model for a specific Region of Interest, where pixel level annotated data of the crop corn was picked for the year 2019 for specific dates and a model was trained for the same. While doing this exercise, I was presented with a unique problem. While using a particular year for crop signature analysis, the moment I extended the analysis to a different year, the model failed to provide the same insights and I just could not replicate the results I had derived from the above experiment. When the single pixel classifier model was used to predict pixels for the same year, the f1 score for out of sample data was remarkable.

Honda builds much more than cars and trucks -- power equipment, solar cells, industrial robotics, alternative fuel engines and even aircraft are all part of the company's production capacity. On Thursday, Honda announced that it is working to further expand its manufacturing portfolio to include Avatar-style remote telepresence robots and electric VTOLs for inter- and intracity commutes before turning its ambitions to building a fuel-cell driven power generation system for the lunar surface. For its eVTOL, Honda plans to leverage not only the lithium battery technology it's developed for its EV and PHEV vehicles but also a gas turbine hybrid power unit to give the future aircraft enough range to handle regional inter-city flights as well. Honda foresees air taxis as a ubiquitous part of tomorrow's transportation landscape, seamlessly integrating with both autonomous ground vehicles and traditional airliners (though they could soon be flown by robots as well). Obviously, the program is still very much in the early research phase and will likely remain so until at least the second half of this decade.

Haitham Al-Beik is the CEO and Founder of Wings, an emerging startup producing autonomous foodservice businesses designed with proprietary, purpose-built "HiveRobotics" and intuitive end-to-end experiences without human intervention. Could you share the genesis story behind Wings? Wings began as a fundamental research and development lab to address the complexities and frictions of the services industry. The industry has grown more labor intensive, surrounded by non-cohesive, vendor-driven components on non-standard logistics and operations, diminishing the industry's creative potential and resulting in no innovation or increase in entrepreneurship. The barrier to entry has only become higher and more complex for any individual to enter the space regardless of their experience or intended sector.