Bayesian geoacoustic inversion problems are conventionally solved by Markov chain Monte Carlo methods or its variants, which are computationally expensive. This paper extends the classic Bayesian geoacoustic inversion framework using the mixture density network (MDN), which provides a much more efficient way to solve geoacoustic inversion problems in Bayesian inference framework. Some important geoacoustic statistics of Bayesian geoacoustic inversion are derived from the multidimensional posterior probability density (PPD) using the MDN theory. These statistics make it convenient to train the network directly on the whole parameter space and get the multidimensional PPD of model parameters. The network is trained on a simulated dataset of surface-wave dispersion curves with shear-wave velocities as labels. The results show that the network gives reliable predictions and has good generalization performance on unseen data. Once trained, the network can rapidly (within seconds) give a fully probabilistic solution which is comparable to Monte Carlo methods. It provides an promissing approach for real-time inversion.

When the original Microsoft Flight simulator was released almost 40 years ago, it was very much for enthusiasts only. Early home computers could barely cope with drawing cockpit instrument panels, let alone scenery – so what you saw as you fought with the controls was a lot of dials and numbers, usually followed by an on-screen message politely informing you that you had crashed during take-off. This is not the experience you will have with Microsoft Flight Simulator 2020. Developed by French studio Asobo using accurate geographic data culled from Bing Maps, a global cloud computing network, and real-time weather information, this is as much a visual spectacle as it is a simulator. And you will want to see as much as you can, because at 10,000 feet, the world looks spectacular (especially on the Ultra graphical settings, where it's almost photorealistic).

In this tutorial, we're going to show you how to take open source data from the National Oceanic and Atmospheric Administration (NOAA), clean it, and forecast future temperatures using no-code machine learning methods. This particular data comes from the Harmful Algal BloomS Observation System (HABSOS). There are several interesting questions to ask of this data -- namely, what is the relationship between algal blooms and water temperature fluctuations. For this tutorial, we're going to start with a basic question: can we predict what temperatures will be over the next five months? There are a lot of approaches to this; what is shown below is just one approach.

"It's very common," says Jess Hanham casually, when asked how often he finds suspected unexploded bombs. Mr Hanham is a co-founder of Spectrum Offshore, a marine survey firm that does a lot of work in the Thames Estuary. His firm undertakes all sorts of marine surveying, but working on sites for new offshore wind farms has become a big business for him. Work in the Thames Estuary, and other areas that were the targets of bombing in World War 2, are likely to involve picking up signals of unexploded munitions. "You can find a significant amount of contacts that need further investigation and for a wind farm that will be established in the initial pre-engineering survey," he says.

In the era of automation and digital transformation, the shipping industry is undergoing dramatic changes to increase efficiency and safety at the port and on the high seas. From small boats to massive container ships, these seafaring vessels are integral components of the global economy. Maritime companies are developing the next generation of autonomous ships and leveraging artificial intelligence, machine learning, and more to design 21st-century smart ports. That said, inherent within digital transformation is of course the transformative process itself. Historically, some ports have relied on rather low-tech, manual solutions.

"Britons on way to France risk quarantine" is the front page headline in the Times, as it reports that Whitehall officials have placed the country on a list of destinations to be closely monitored. A senior aviation source is quoted saying France is "bubbling" with cases and that travellers should only book trips which can be re-arranged at 24 hours' notice. The Daily Telegraph also reports the close monitoring of France as cases there overtake the numbers for Portugal, which has reduced its infection rate. The paper says about 450,000 Britons are currently holidaying in France, a scale which would make any new restrictions a logistical nightmare. The Guardian leads with an exclusive warning from doctors' leaders that shutting down non-Covid NHS services to deal with any second wave will leave thousands of patients unacceptably "stranded", risking more deaths.

Ideally Open-Domain Question Answering models should exhibit a number of competencies, ranging from simply memorizing questions seen at training time, to answering novel question formulations with answers seen during training, to generalizing to completely novel questions with novel answers. However, single aggregated test set scores do not show the full picture of what capabilities models truly have. In this work, we perform a detailed study of the test sets of three popular open-domain benchmark datasets with respect to these competencies. We find that 60-70% of test-time answers are also present somewhere in the training sets. We also find that 30% of test-set questions have a near-duplicate paraphrase in their corresponding training sets. Using these findings, we evaluate a variety of popular open-domain models to obtain greater insight into what extent they can actually generalize, and what drives their overall performance. We find that all models perform dramatically worse on questions that cannot be memorized from training sets, with a mean absolute performance difference of 63% between repeated and non-repeated data. Finally we show that simple nearest-neighbor models out-perform a BART closed-book QA model, further highlighting the role that training set memorization plays in these benchmarks

This paper describes our winning solution for the ECML-PKDD ChAT Discovery Challenge 2020. We show that whether or not a Twitch user has subscribed to a channel can be well predicted by modeling user activity with boosting trees. We introduce the connection between target-encodings and boosting trees in the context of high cardinality categoricals and find that modeling user activity is more powerful then direct modeling of content when encoded properly and combined with a gradient boosting optimization approach.

All famous machine learning algorithms that comprise both supervised and semi-supervised learning work well only under a common assumption: the training and test data follow the same distribution. When the distribution changes, most statistical models must be reconstructed from newly collected data, which for some applications can be costly or impossible to obtain. Therefore, it has become necessary to develop approaches that reduce the need and the effort to obtain new labeled samples by exploiting data that are available in related areas, and using these further across similar fields. This has given rise to a new machine learning framework known as transfer learning: a learning setting inspired by the capability of a human being to extrapolate knowledge across tasks to learn more efficiently. Despite a large amount of different transfer learning scenarios, the main objective of this survey is to provide an overview of the state-of-the-art theoretical results in a specific, and arguably the most popular, sub-field of transfer learning, called domain adaptation. In this sub-field, the data distribution is assumed to change across the training and the test data, while the learning task remains the same. We provide a first up-to-date description of existing results related to domain adaptation problem that cover learning bounds based on different statistical learning frameworks.

According to the National Academies, a weekly forecast of velocity, vertical structure, and duration of the Loop Current (LC) and its eddies is critical for understanding the oceanography and ecosystem, and for mitigating outcomes of anthropogenic and natural disasters in the Gulf of Mexico (GoM). However, this forecast is a challenging problem since the LC behaviour is dominated by long-range spatial connections across multiple timescales. In this paper, we extend spatiotemporal predictive learning, showing its effectiveness beyond video prediction, to a 4D model, i.e., a novel Physics-informed Tensor-train ConvLSTM (PITT-ConvLSTM) for temporal sequences of 3D geospatial data forecasting. Specifically, we propose 1) a novel 4D higher-order recurrent neural network with empirical orthogonal function analysis to capture the hidden uncorrelated patterns of each hierarchy, 2) a convolutional tensor-train decomposition to capture higher-order space-time correlations, and 3) to incorporate prior physic knowledge that is provided from domain experts by informing the learning in latent space. The advantage of our proposed method is clear: constrained by physical laws, it simultaneously learns good representations for frame dependencies (both short-term and long-term high-level dependency) and inter-hierarchical relations within each time frame. Experiments on geospatial data collected from the GoM demonstrate that PITT-ConvLSTM outperforms the state-of-the-art methods in forecasting the volumetric velocity of the LC and its eddies for a period of over one week.