Last year, IBM announced a Data Science Elite team whose only job is to help big enterprise companies push their first AI models into production. Now, more than a year after the program's launch, Rob Thomas, the IBM executive overseeing the AI SWAT team, reports that it has been a "huge success." The team has increased from 30 data scientists to 100, and there are plans to grow significantly next year. "We hire them wherever we can, actually," Thomas said, noting that these data scientists operate all over the world. Companies as diverse as Harley Davidson, Lufthansa, Experian, Sprint, Carrefour, and Siemens used the team for a necessary kickstart on AI projects. And the best part: It's all for free -- or at least there are no contractual obligations to pay.

This was not so much a transformation in terms of connectivity – nearly half the people in the world are not yet internet users, and if you are from a low-income country you probably only have patchy access to 3G if you are lucky. This was a change in the scale and reach of the world's data infrastructures, to a point where no one is truly invisible any more. You may not be able to reach the connected world, but it can certainly reach you. One clear sign of this internationalisation of infrastructure was the expansion of AI-enabled surveillance. A Carnegie Foundation report shows that 47 out of the 65 countries using AI surveillance are doing so with Chinese technology, though US and European firms are also providing substantial amounts.

Technological change and innovation are vitally important, especially for high-tech companies. However, factors influencing their future research and development (R&D) trends are both complicated and various, leading it a quite difficult task to make technology tracing for high-tech companies. To this end, in this paper, we develop a novel data-driven solution, i.e., Deep Technology Forecasting (DTF) framework, to automatically find the most possible technology directions customized to each high-tech company. Specially, DTF consists of three components: Potential Competitor Recognition (PCR), Collaborative Technology Recognition (CTR), and Deep Technology Tracing (DTT) neural network. For one thing, PCR and CTR aim to capture competitive relations among enterprises and collaborative relations among technologies, respectively. For another, DTT is designed for modeling dynamic interactions between companies and technologies with the above relations involved. Finally, we evaluate our DTF framework on real-world patent data, and the experimental results clearly prove that DTF can precisely help to prospect future technology emphasis of companies by exploiting hybrid factors.

Delivering useful hydrological forecasts is critical for urban and agricultural water management, hydropower generation, flood protection and management, drought mitigation and alleviation, and river basin planning and management, among others. In this work, we present and appraise a new methodology for hydrological time series forecasting. This methodology is based on simple combinations. The appraisal is made by using a big dataset consisted of 90-year-long mean annual river flow time series from approximately 600 stations. Covering large parts of North America and Europe, these stations represent various climate and catchment characteristics, and thus can collectively support benchmarking. Five individual forecasting methods and 26 variants of the introduced methodology are applied to each time series. The application is made in one-step ahead forecasting mode. The individual methods are the last-observation benchmark, simple exponential smoothing, complex exponential smoothing, automatic autoregressive fractionally integrated moving average (ARFIMA) and Facebook's Prophet, while the 26 variants are defined by all the possible combinations (per two, three, four or five) of the five afore-mentioned methods. The findings have both practical and theoretical implications. The simple methodology of the study is identified as well-performing in the long run. Our large-scale results are additionally exploited for finding an interpretable relationship between predictive performance and temporal dependence in the river flow time series, and for examining one-year ahead river flow predictability.

There is currently an unprecedented demand for large-scale temporal data analysis due to the explosive growth of data. Dynamic topic modeling has been widely used in social and data sciences with the goal of learning latent topics that emerge, evolve, and fade over time. Previous work on dynamic topic modeling primarily employ the method of nonnegative matrix factorization (NMF), where slices of the data tensor are each factorized into the product of lower-dimensional nonnegative matrices. With this approach, however, information contained in the temporal dimension of the data is often neglected or underutilized. To overcome this issue, we propose instead adopting the method of nonnegative CANDECOMP/PARAPAC (CP) tensor decomposition (NNCPD), where the data tensor is directly decomposed into a minimal sum of outer products of nonnegative vectors, thereby preserving the temporal information. The viability of NNCPD is demonstrated through application to both synthetic and real data, where significantly improved results are obtained compared to those of typical NMF-based methods. The advantages of NNCPD over such approaches are studied and discussed. To the best of our knowledge, this is the first time that NNCPD has been utilized for the purpose of dynamic topic modeling, and our findings will be transformative for both applications and further developments.

Marine buoys aid in the battle against Illegal, Unreported and Unregulated (IUU) fishing by detecting fishing vessels in their vicinity. Marine buoys, however, may be disrupted by natural causes and buoy vandalism. To minimize the effects of buoy disruption on a buoy network, we propose a more robust buoy placement using dropout k-means and dropout k-median. We apply dropout k-means and dropout k-median to determine locations for deploying marine buoys in the Gabonese waters near West Africa. We simulated the passage of ships using historical Automatic Identification System (AIS) data, then compared the ship detection probability of dropout k-means to classic k-means and dropout k-median to classic k-median, taking into account that the current sensor detection radius is 10km. With 5 buoys, the buoy arrangement computed by classic k-means, dropout k-means, classic k-median and dropout k-median have ship detection probabilities of 38%, 45%, 48% and 52%.

We review recent works [1, 2, 3] on analyzing the dynamics of gradient-based algorithms in a prototypical statistical inference problem. Using methods and insights from the physics of glassy systems, these works showed how to understand quantitatively and qualitatively the performance of gradient-based algorithms. Here we review the key results and their interpretation in nontechnical terms accessible to a wide audience of physicists in the context of related works. PACS numbers: 00.00, 20.00, 42.10 Keywords: analysis of algorithms, statistical inference, spin glasses, machine learning.

Personal use of this material is permitted. Abstract --Pre-training a deep neural network on the ImageNet dataset is a common practice for training deep learning models, and generally yields improved performance and faster training times. The technique of pre-training on one task and then retraining on a new one is called transfer learning. We perform three sets of experiments with varying levels of similarity between source and target tasks to investigate the behaviour of different types of knowledge transfer . We transfer both parameters and features and analyse their behaviour . Our results demonstrate that no significant advantage is gained by using a transfer learning approach over a traditional machine learning approach for our character recognition tasks. This suggests that using transfer learning does not necessarily presuppose a better performing model in all cases. Learning to drive a car makes learning to drive a truck easier, and knowing how to speak Spanish makes learning Portuguese easier.

It's easy to get cynical about artificial intelligence (AI). China is using facial recognition against the Uighurs. NYT: 'One Month, 500,000 Face Scans: How China Is Using A.I. to Profile a Minority' Google's participating in the development of autonomous weapons. The Intercept: 'Google Continues Investments in Military and Police AI Technology Through Venture Capital Arm' And facial recognition programmes are still struggling to recognise black faces. But last year I also saw another side.

At Voyance, our ultimate mission is to democratize machine learning for everyone in the data science ecosystem and as such, we are so happy to announce WithData. WithData is a 5 min video interview showcasing and interviewing data analyst, scientist, product managers and engineers helping African organisations make data driven decisions. The data science ecosystem in Nigeria is still very nascent and our goal with this series is to help bring spotlight into it and showcase the ever growing amazing talented individuals working in this industry. This video series will also enable these people talk about their work i.e what goes into building a complete predictive models and so much more. If you would like to be part of the video series, kindly fill this Airtable form https://airtable.com/shrDUiLwTVkyWcj6e with your information and we'll get back to you ASAP.