Every year the technology industry gathers in Las Vegas for the Consumer Electronics Show (CES), an event that often sets the agenda for the coming 12 months. This is what CES 2019 taught us. The first 5G networks are expected to begin rolling out this year, and so the next-generation connectivity technology was being mentioned everywhere at CES. Intel, Qualcomm and Samsung all spoke about harnessing the technology to not just offer faster mobile internet speeds, but also to connect more devices and appliances to each other and be able to handle more data in the process. Experts at the show also commented on the higher capacity of 5G networks being able to support the software needed to power networks of driverless cars and robots. The halls of this year's CES hinted at a world where homes, cars and even entire cities are connected to one another, with people able to use these connections to complete tasks every day.

Major companies are investing in machine learning-powered approaches to improve all aspects of manufacturing. Firms are using this technology to bring down labour costs, reduce product defects, shorten unplanned downtimes, improve transition times, and increase production speed. Artificial intelligence will help drive the fourth industrial revolution – Industry 4.0 – with machine learning and deep learning rapidly becoming mainstream technologies. Recent developments and partnerships have shown how IBM, Microsoft and SAP, among others, are exploring the future of manufacturing. The mission of the Europe-based Korea Institute of Science and Technology (KIST) is contributing towards globalisation of Korean research and innovation by building an open platform where notable South Korean and European institutions and industrial partners can collaborate.

People around the globe are more actively using social media platform such as Twitter, Facebook, and Instagram etc. They share information, opinions, ideas, experiences and other details in the social media. The business communities have become more aware of these developments and they want to use the available information in their favor. One of the ways to understand the people opinions on the product they are using is by collecting tweets related to those products. Then performing the sentiment analysis on the tweets collected on a particular topic.

The open source silicon space has suddenly become more crowded. Shortly before Christmas, Silicon Valley AI startup Wave Computing, which is developing hardware for running deep learning applications in data centers and offices, announced plans to open source its MIPS instruction set architecture, or ISA, under what it's calling the "MIPS Open" program. When the process is completed in the first quarter, participants will have full access, with no licensing fees or royalties, to the most recent versions of the 32-bit and 64-bit MIPS ISA, along with licensing for MIPS's "hundreds of existing worldwide patents." According to Wave, open sourcing the design will open the door for semiconductor companies, developers, and universities to adopt and innovate using MIPS for next-generation system-on-chip (SoC) designs. Related: Is Open Source RISC-V Ready to Take on Intel, AMD, and ARM in the Data Center?

Bank Danamon is one of Indonesia's largest financial institutions, offering corporate and small business banking, consumer banking, trade finance, cash management, treasury and capital markets. Financial service institutions globally are under immense pressure from a new generation of tech-savvy customers to deliver consumer-centric digital services. These new services represent a massive opportunity as well as a massive risk as more consumers connect to products and services digitally. For Bank Danamon, like many other institutions, one of the key challenges in advancing this digital transformation is the ability to gain a single enterprise view of customer data. "Each line of business has their own data mart so there are a lot of data silos," said Billie Setiawan, head, Decision Management Data and Analytics for Bank Danamon Indonesia.

In an interesting turn of events, IIT Kharagpur, this week announced that they were to launch a new course on artificial intelligence and machine learning, specially designed for working professionals and engineering students. The programme, which will be of six months duration, will commence from March 2019 and will be conducted from IIT-Kgp institute units in Kharagpur, Bengaluru and Kolkata and possibly in Hyderabad as well. PP Chakrabarti, director at IIT Kharagpur, told the media on Thursday, "A rigorous AI programme for professionals is the need of the hour. The programme has been designed by IIT Kharagpur faculty in consultation with industry experts." This course will comprise 16 one-credit modules and one capstone project.

Two facts about simple viruses have been known for a long time. Firstly, that genetic economy leads to the use of symmetry, such that virus capsids aremostly icosahedral or helical. Secondly, packaging signals, that is secondary structure features in the viral RNA, are often required for encapsidation inviruses with single-stranded genomes. Examples are the origin of assembly sequence in Tobacco Mosaic virus, the psi element in HIV and the TR sequence in MS2. This is an evolutionary advantage, as it ensures vRNA-specific encapsidation and can increase assembly efficiency through a cooperative role of the RNA, which acts as a nucleation site. More recently, it has been shown that taken together, these two facts suggest that there could be more than one packaging signal, with multiple signalsin fact dispersed throughout the genome. This is because the capsid is symmetric, and the packaging signal mechanism functions via interaction betweenviral RNA and the coat protein (CP). In several cases, this RNA-CP interaction leads to a conformational change in the CP, which only then makes it assembly competent (e.g.

A feature selection algorithm should ideally satisfy four conditions: reliably extract relevant features; be able to identify non-linear feature interactions; scale linearly with the number of features and dimensions; allow the incorporation of known sparsity structure. In this work we propose a novel feature selection algorithm, Gradient Boosted Feature Selection (GBFS), which satisfies all four of these requirements. The algorithm is flexible, scalable, and surprisingly straight-forward to implement as it is based on a modification of Gradient Boosted Trees. We evaluate GBFS on several real world data sets and show that it matches or out-performs other state of the art feature selection algorithms. Yet it scales to larger data set sizes and naturally allows for domain-specific side information.

Generating accurate and reliable sales forecasts is crucial in the E-commerce business. The current state-of-the-art techniques are typically univariate methods, which produce forecasts considering only the historical sales data of a single product. However, in a situation where large quantities of related time series are available, conditioning the forecast of an individual time series on past behaviour of similar, related time series can be beneficial. Given that the product assortment hierarchy in an E-commerce platform contains large numbers of related products, in which the sales demand patterns can be correlated, our attempt is to incorporate this cross-series information in a unified model. We achieve this by globally training a Long Short-Term Memory network (LSTM) that exploits the nonlinear demand relationships available in an E-commerce product assortment hierarchy. Aside from the forecasting engine, we propose a systematic pre-processing framework to overcome the challenges in an E-commerce setting. We also introduce several product grouping strategies to supplement the LSTM learning schemes, in situations where sales patterns in a product portfolio are disparate. We empirically evaluate the proposed forecasting framework on a real-world online marketplace dataset from Walmart. com. Our method achieves competitive results on category level and super-departmental level datasets, outperforming state-of-the-art techniques.

Deep neural networks work well at approximating complicated functions when provided with data and trained by gradient descent methods. At the same time, there is a vast amount of existing functions that programmatically solve different tasks in a precise manner eliminating the need for training. In many cases, it is possible to decompose a task to a series of functions, of which for some we may prefer to use a neural network to learn the functionality, while for others the preferred method would be to use existing black-box functions. We propose a method for end-to-end training of a base neural network that integrates calls to existing black-box functions. We do so by approximating the black-box functionality with a differentiable neural network in a way that drives the base network to comply with the black-box function interface during the end-to-end optimization process. At inference time, we replace the differentiable estimator with its external black-box non-differentiable counterpart such that the base network output matches the input arguments of the black-box function. Using this "Estimate and Replace" paradigm, we train a neural network, end to end, to compute the input to black-box functionality while eliminating the need for intermediate labels. We show that by leveraging the existing precise black-box function during inference, the integrated model generalizes better than a fully differentiable model, and learns more efficiently compared to RL-based methods.