Qualcomm SVP Alex Katouzian didn't dwell on the chipset for very long after announcing it at the company's Tech Summit in Hawaii -- the big details will apparently drop tomorrow -- but we now have a better sense of what the company wanted to focus on as we barrel into 2019. It's been clear for a while now that 2019 will be the year of 5G, and it's little surprise at this point that the Snapdragon 855 will support "multi-gigabit" data speeds on 5G networks as they light up around the country. The rapid development and rigorous work that went into next year's round of 5G network deployments were a big theme here at Tech Summit day 1, and it's not hard to see how insanely fast data speeds stand to change what we expect from our smartphones. Beyond that, Katouzian also pointed out that the 855 was designed to trounce last-generation chipsets when it comes to AI performance -- we can expect up to 3x performance gains when it comes to these complex computations. It's still relatively early days for software and services fueled by machine learning, but the shift in the industry seems almost palpable at this point. With these AI-focused updates, Qualcomm is clearly gearing up to compete with rivals like Apple as the fundamental nature of our software continues to change.

Deaf children face challenges learning to read. As their parents and teachers often don't know sign language, young ones can't always make the connection between words on the page and their own life experiences. Huawei aims to fix that with its StorySign app for Android. Point your phone at certain children's books and the app will use AI to translate individual words on the page to sign language performed by an avatar (created by Wallace and Gromit's Aardman Animations, no less). This not only helps children read, but can teach parents the sign language they'd need to tell the story later. The app is free on both Google Play and Huawei's own AppGallery, and it doesn't require a Huawei phone.

Qualcomm Ventures, the corporate venture capital arm of the chipmaker, has plans to invest up to $100 million in artificial intelligence. Specifically, Qualcomm says it will provide capital to startups building on-device AI, which is AI that runs on the end device, like a smartphone or vehicle, rather than in the cloud. The fund's leader, Qualcomm investment director Albert Wang (pictured), says on-device AI is the future. "Today's AI processing is very computationally intensive," Wang told TechCrunch. "When you're talking to Alexa, nothing is processed on your device, it gets taken to the cloud and gets scrunched there. There are a few problems with that -- performance deteriorates, it consumes a lot of bandwidth and there are privacy issues. Imagine you have an Alexa that is more private and user-friendly, you ask the questions and can get the answers instantly. It doesn't take the round trip all the way to the cloud."

Micron said in October it plans to invest up to $100 million in startups focusing on artificial intelligence, while Intel's venture capital arm has funneled more than $1 billion in recent years into the technology.

Data collection often involves the partial measurement of a larger system. A common example arises in the process collecting network data: we often obtain network datasets by recording all of the interactions among a small set of core nodes, so that we end up with a measurement of the network consisting of these core nodes together with a potentially much larger set of fringe nodes that have links to the core. Given the ubiquity of this process for assembling network data, it becomes crucial to understand the role of such a core-fringe structure. Here we study how the inclusion of fringe nodes affects the standard task of network link prediction. One might initially think the inclusion of any additional data is useful, and hence that it should be beneficial to include all fringe nodes that are available. However, we find that this is not true; in fact, there is substantial variability in the value of the fringe nodes for prediction. In some datasets, once an algorithm is selected, including any additional data from the fringe can actually hurt prediction performance; in other datasets, including some amount of fringe information is useful before prediction performance saturates or even declines; and in further cases, including the entire fringe leads to the best performance. While such variety might seem surprising, we show that these behaviors are exhibited by simple random graph models.

Cellular network configuration plays a critical role in network performance. In current practice, network configuration depends heavily on field experience of engineers and often remains static for a long period of time. This practice is far from optimal. To address this limitation, online-learning-based approaches have great potentials to automate and optimize network configuration. Learning-based approaches face the challenges of learning a highly complex function for each base station and balancing the fundamental exploration-exploitation tradeoff while minimizing the exploration cost. Fortunately, in cellular networks, base stations (BSs) often have similarities even though they are not identical. To leverage such similarities, we propose kernel-based multi-BS contextual bandit algorithm based on multi-task learning. In the algorithm, we leverage the similarity among different BSs defined by conditional kernel embedding. We present theoretical analysis of the proposed algorithm in terms of regret and multi-task-learning efficiency. We evaluate the effectiveness of our algorithm based on a simulator built by real traces.

The widespread mobile devices facilitated the emergence of many new applications and services. Among them are location-based services (LBS) that provide services based on user's location. Several techniques have been presented to enable LBS even in indoor environments where Global Positioning System (GPS) has low localization accuracy. These methods use some environment measurements (like Channel State Information (CSI) or Received Signal Strength (RSS)) for user localization. In this paper, we will use CSI and a novel deep learning algorithm to design a robust and efficient system for indoor localization. More precisely, we use supervised autoencoder (SAE) to model the environment using the data collected during the training phase. Then, during the testing phase, we use the trained model and estimate the coordinates of the unknown point by checking different possible labels. Unlike the previous fingerprinting approaches, in this work, we do not store the {CSI/RSS} of fingerprints and instead we model the environment only with a single SAE. The performance of the proposed scheme is then evaluated in two indoor environments and compared with that of similar approaches.

Does your mopping robot or your food processor have names? Do you ascribe a character to them? Unless you're a robot yourself, you probably do. The entire history of engineering has been about making the most human-like machines possible. We humans are hardwired to process social interactions, so we subconsciously expect machines to be at least somewhat human.

Huawei's latest phone is something rather special. The Mate 20 Pro packs more cutting-edge technology into it than any other, but unlike most bleeding-edge devices it makes for a great experience. It looks like an iPhone XS mated with a Samsung Galaxy Note 9, with a huge curved 6.39in OLED display and big notch in the top. The screen is great and hides a good pressure-sensitive in-display fingerprint scanner. The curved sides and narrow width makes the massive screen manageable.

Fiber nonlinear interference (NLI) modeling and monitoring are the key building blocks to support elastic optical networks (EONs). In the past, they were normally developed and investigated separately. Moreover, the accuracy of the previously proposed methods still needs to be improved for heterogenous dynamic optical networks. In this paper, we present the application of machine learning (ML) in NLI modeling and monitoring. In particular, we first propose to use ML approaches to calibrate the errors of current fiber nonlinearity models. The Gaussian-noise (GN) model is used as an illustrative example, and significant improvement is demonstrated with the aid of an artificial neural network (ANN). Further, we propose to use ML to combine the modeling and monitoring schemes for a better estimation of NLI variance. Extensive simulations with 1603 links are conducted to evaluate and analyze the performance of various schemes, and the superior performance of the ML-aided combination of modeling and monitoring is demonstrated.