With more than 4 billion internet users worldwide today and 31 billion connected devices forecasted by 2020, the future of the digital world lies in how people and "things" will interact with each other. The key to this will be the convergence and consolidation of internet of things platforms and devices which will be able to seamlessly exchange data between people, networks, devices and applications. Creating this world, where multiple service and technology layers work harmoniously to create ubiquitous, ultra-connected experiences, is a task that will take years to complete. It requires a robust technology platform, powered by artificial intelligence. Today, we are siloed in how we think about IoT.

Zero Touch & Carrier Automation Congress -- The 3GPP standards group is developing a machine learning function that could allow 5G operators to monitor the status of a network slice or third-party application performance. The network data analytics function (NWDAF) forms a part of the 3GPP's 5G standardization efforts and could become a central point for analytics in the 5G core network, said Serge Manning, a senior technology strategist at Sprint Corp. (NYSE: S). Speaking here in Madrid, Manning said the NWDAF was still in the "early stages" of standardization but could become "an interesting place for innovation." The 3rd Generation Partnership Project (3GPP) froze the specifications for a 5G new radio standard at the end of 2017 and is due to freeze another set of 5G specifications, covering some of the core network and non-radio features, in June this year as part of its "Release 15" update. Manning says that Release 15 considers the network slice selection function (NSSF) and the policy control function (PCF) as potential "consumers" of the NWDAF.

Deep neural networks are often trained in the over-parametrized regime (i.e. with far more parameters than training examples), and understanding why the training converges to solutions that generalize remains an open problem. Several studies have highlighted the fact that the training procedure, i.e. mini-batch Stochastic Gradient Descent (SGD) leads to solutions that have specific properties in the loss landscape. However, even with plain Gradient Descent (GD) the solutions found in the over-parametrized regime are pretty good and this phenomenon is poorly understood. We propose an analysis of this behavior for feedforward networks with a ReLU activation function under the assumption of small initialization and learning rate and uncover a quantization effect: The weight vectors tend to concentrate at a small number of directions determined by the input data. As a consequence, we show that for given input data there are only finitely many, "simple" functions that can be obtained, independent of the network size. This puts these functions in analogy to linear interpolations (for given input data there are finitely many triangulations, which each determine a function by linear interpolation). We ask whether this analogy extends to the generalization properties - while the usual distribution-independent generalization property does not hold, it could be that for e.g. smooth functions with bounded second derivative an approximation property holds which could "explain" generalization of networks (of unbounded size) to unseen inputs.

The research on improving Artificial Intelligence (A.I.) has been ongoing for decades. However, it wasn't until recently that developers were finally able to create smart systems that closely resemble the A.I. capabilities of humans. The main reason for this breakthrough in technology is advancements in Big Data. Recent developments in Big Data have allowed us the capability to organize a very large amount of information into structured components that can be very quickly processed by computers. Another technology that has the potential for rapidly advancing and transforming Artificial Intelligence is the Blockchain.

Today, the amount of data that can be retrieved from communications networks is extremely high and diverse (e.g., data regarding users behavior, traffic traces, network alarms, signal quality indicators, etc.). Advanced mathematical tools are required to extract useful information from this large set of network data. In particular, Machine Learning (ML) is regarded as a promising methodological area to perform network-data analysis and enable, e.g., automatized network self-configuration and fault management. In this survey we classify and describe relevant studies dealing with the applications of ML to optical communications and networking. Optical networks and system are facing an unprecedented growth in terms of complexity due to the introduction of a huge number of adjustable parameters (such as routing configurations, modulation format, symbol rate, coding schemes, etc.), mainly due to the adoption of, among the others, coherent transmission/reception technology, advanced digital signal processing and to the presence of nonlinear effects in optical fiber systems. Although a good number of research papers have appeared in the last years, the application of ML to optical networks is still in its early stage. In this survey we provide an introductory reference for researchers and practitioners interested in this field. To stimulate further work in this area, we conclude the paper proposing new possible research directions.

Considered as a data-driven approach, Fingerprinting Localization Solutions (FPSs) enjoy huge popularity due to their good performance and minimal environment information requirement. This papers addresses applications of artificial intelligence to solve two problems in Received Signal Strength Indicator (RSSI) based FPS, first the cumbersome training database construction and second the extrapolation of fingerprinting algorithm for similar buildings with slight environmental changes. After a concise overview of deep learning design techniques, two main techniques widely used in deep learning are exploited for the above mentioned issues namely data augmentation and transfer learning. We train a multi-layer neural network that learns the mapping from the observations to the locations. A data augmentation method is proposed to increase the training database size based on the structure of RSSI measurements and hence reducing effectively the amount of training data. Then it is shown experimentally how a model trained for a particular building can be transferred to a similar one by fine tuning with significantly smaller training numbers. The paper implicitly discusses the new guidelines to consider about deep learning designs when they are employed in a new application context.

Most Android phones will have to wait until 2019 to duplicate the 3D sensing feature behind Apple s Face ID security, three major parts producers have told Reuters, handicapping Samsung and others on a technology that is set to be worth billions in revenue over the next few years. The development of new features for the estimated 1.5 billion smart phones shipped annually has been at the heart of the battle for global market share over the past decade, with Apple, bolstered by its huge R&D budget, often leading. When the iPhone 5S launched with a fingerprint-sensing home button in September 2013, for example, it took its biggest rival Samsung until just April of the next year to deliver its own in the Galaxy S5, with others following soon after. Most Android phones will have to wait until 2019 to duplicate the 3D sensing feature behind Apple s Face ID security, three major parts producers have told Reuters. The 3D sensing technology is expected to enhance the next generation of phones, enabling accurate facial recognition as well as secure biometrics for payments, gesture sensing, and immersive shopping and gaming experiences.

If a new report is to be believed, Samsung and other Android phone makers would need more time to catch up with Apple's pretty advanced 3D-sensing technology that the Cupertino giant debuted with last year's iPhone X. Tim Cook's company is said to have secured a two-year lead when it comes to this technology, so Apple's rivals won't likely be capable of duplicating Face ID until next year. Reuters reported Tuesday that most Android phone makers will have to wait until 2019 before they could be able to deliver a technology that's up to par with Apple's 3D-sensing, which is powering the Face ID security feature of the iPhone X. This is seen as a setback on the Android manufacturers' end because 3D-sensing technology is expected to be worth billions in revenue in the next few years. Reuters obtained data from parts suppliers, and the publication feels that Huawei and Xiaomi will be among the Android brands that will be capable of matching Apple's 3D-sensing next year. This is because 3D-sensing parts suppliers will be capable of reaching production levels for worldwide adoption by that time.

INTERVIEW: Artificial Intelligence (AI) will be vital to unlocking the "true potential" of IoT, believes Chris Penrose, AT&T's president of IoT solutions (pictured), as he suggested experience will help the operator gain a competitive edge in the segment. Speaking to Mobile World Live, Penrose said it "almost becomes impossible" to use the data generated by connected devices and make it into something actionable, given its volume, without some sort of AI assistance. Using data will ultimately "unlock predictability", he said, and will enable an evolution for the industry from being able to "sense information to being able to predict things". "We will know that a factory floor or a machine might be going wrong ahead of time and can be altered, or for a car battery we will be notified that its going to fail before it does using the power of AI." Penrose also said the operator was confident in the potential of its consumer and enterprise IoT offering, despite increased competition from rival operators in the US. At the start of the year, T-Mobile US launched its Magenta NB-IoT tariff for businesses, while Verizon rolled out a similar offering for its LTE-M network.

It was great catching up with Ericsson last week in San Francisco at the inaugural Mobile World Congress Americas conference. Ericsson is doing incredible work to advance innovation by partnering with operators globally around IoT and 5G deployments, ranging from testing new radio technology, like advanced MIMO, to new core 5G systems for providing network slicing, to applications like Autonomous vehicles. Ericsson's radio access network was also featured at Sprint's booth where the first 2.5 GHz Massive MIMO field tests were conducted using Sprint's spectrum and Ericsson's radios reaching peak speeds of more than 300 Mbps using a single 20 MHz channel! A great new use case for 5G was intelligent video streaming with Verizon for security and smart city applications, with streams coming to a central, video optimized repository in the core of the 5G network. This 5G overlay to an existing 4G network will provide benefits across multiple applications at the edge of the network from video cameras to drones to industrial control endpoints.