In recent years, on-device deep learning has gained attention as a means of developing affordable deep learning applications for mobile devices. However, on-device models are constrained by limited energy and computation resources. In the mean time, a poisoning attack known as sponge poisoning has been developed.This attack involves feeding the model with poisoned examples to increase the energy consumption during inference. As previous work is focusing on server hardware accelerators, in this work, we extend the sponge poisoning attack to an on-device scenario to evaluate the vulnerability of mobile device processors. We present an on-device sponge poisoning attack pipeline to simulate the streaming and consistent inference scenario to bridge the knowledge gap in the on-device setting. Our exclusive experimental analysis with processors and on-device networks shows that sponge poisoning attacks can effectively pollute the modern processor with its built-in accelerator. We analyze the impact of different factors in the sponge poisoning algorithm and highlight the need for improved defense mechanisms to prevent such attacks on on-device deep learning applications.

Last week, Qualcomm announced the Snapdragon 845, which sends AI tasks to the most suitable cores. There's not a lot of difference between the three company's approaches -- it ultimately boils down to the level of access each company offers to developers and how much power each setup consumes. Before we get into that though, let's figure out if an AI chip is really all that different from existing CPUs. A term you'll hear a lot in the industry with reference to AI lately is "heterogeneous computing." It refers to systems that use multiple types of processors, each with specialized functions, to gain performance or save energy.

It's IFA 2019, and that can only mean one thing: Gadgets abound for an entire week straight. LG, as it's apt to do, introduced an incrementally refined phone in the LG G8X (the successor to the LG G8). But the accessory the Seoul company unveiled alongside it is arguably more interesting: an updated version of its snap-on Dual Screen product that improves upon the original in several respects. Pricing has yet to be announced for the G8X and revamped Dual Screen, but we'll keep our eyes peeled for more details. LG anticipates both will ship in Q4.

Tech's biggest players have fully embraced the AI revolution. Apple, Qualcomm and Huawei have made mobile chipsets that are designed to better tackle machine-learning tasks, each with a slightly different approach. Huawei launched its Kirin 970 at IFA this year, calling it the first chipset with a dedicated neural processing unit (NPU). Then, Apple unveiled the A11 Bionic chip, which powers the iPhone 8, 8 Plus and X. The A11 Bionic features a neural engine that the company says is "purpose-built for machine-learning," among other things.

But the new Kryo 485 core includes something unusual: a "prime core." Typically, a Snapdragon chip includes four "performance" cores and four "efficiency" cores, the latter optimized for lower power. The Snapdragon 845 uses four ARM A75 cores at 2.8GHz and four A55 cores running at 1.8GHz. Qualcomm says the Kryo 485 within the Snapdragon 855 is 45 percent more powerful. Here are the speeds of each of the Snapdragon 855's Kryo cores. But there are some interesting differences between the 845 and the 855.

The internet of things has evolved from just connecting and transferring data between devices like sensors, cameras, and thermostats to making these devices smarter with decision-making capabilities. Thanks to machine learning (ML) and artificial intelligence (AI) technologies, that help these connected edge devices perform faster, in smarter ways. Artificial intelligence plays a significant role in helping users analyze myriad of data generated by sensors and act upon them in a manner that is beneficial to users in different ways, such as environmental monitoring, weather analysis, predicting equipment failure in industries, disease prediction, etc. Machine learning and neural networks as parts of the AI technology help detect anomalies and patterns of data generated by sensors and devices, which help in extracting better insights for intelligent decision-making. AI-enabled IoT edge devices also help companies to increase operational efficiency and reduce downtimes, giving a competitive edge to business performance. Let us understand how AI empowers smart and powerful devices on the network edge.

As pop-up stores around the world put the latest OnePlus handset on sale today, and the rest of the world's retail stores put it on sale tomorrow, what does the world's press think of the OnePlus 6? The notch is on show, the dual lens camera is tested, and the all-glass construction help it stand out. is that enough? OnePlus is leaning heavily on "The Speed You Need" as the key marketing phrase, and the specifications bear that out. Much like many high-end handsets, the OnePlus 6 comes with a Qualcomm Snapdragon 845, paired up with 6gB or 8GB of RAM, and internal storage options of 64GB, 128GB, or 256GB. I still think that getting the 8GB RAM model is unnecessary for most customers, and the Pixel 2 XL shows that there is still room for improvement when it comes to smoothness.

When it comes to Android flagships, Samsung is king of the hill. According to a recent IDC report, Samsung is the No. 1 Android brand in the world, with Huawei and Xiaomi close behind. Nowhere on that top-five list is LG, which might seem surprising. LG even had a little bit of help from Google along the way; the G6 was the first non-Pixel phone with Google Assistant, the V20 was the first new phone to ship with Nougat, and the V30S ThinQ was the first with device-specific Google Assistant commands. The G7, then, is a sign that LG isn't giving up.

The first Android phone to feature 3D facial recognition technology will reportedly be released in the third quarter of 2018. It's currently being speculated that the Xiaomi Mi 7 will be the first Android phone to arrive with the feature, which Apple's iPhone X popularized. The reason why it's taking so long for Android phones to adopt 3D-sensing technology is due to the difficulty in integrating devices' hardware and software, according to Digitimes. Apple doesn't suffer from the same issue since the iPhone X's hardware and software are closely tied to each other making it possible for Face ID to work. Three-dimensional-sensing modules developed by Qualcomm, Himax Technologies and Truly Optoelectronics are believed to be the likely candidates that are coming to Android phones.

Samsung's Galaxy S9 is a strange breed of smartphone. With a Snapdragon 845 processor, great camera, and 18:9 screen, it has all the trappings of a fantastic 2018 handset. Yet, at the same time, it's remarkably similar to the Galaxy S8 that it replaces. In fact, it's so much like its predecessor, issues we might normally overlook become all the more obvious this second time around. Samsung's hook with the Galaxy S9 is a "reimagined" camera, but the camera's new features--namely Dual Aperture, Super Slow-mo, and AR Emoji--are equal parts gimmick and catch-up to competing models. And because Samsung is limiting the dual camera and Live Portrait mode to the larger Plus model, the S9 (the phone I'm reviewing here) feels less like a new phone than a mid-cycle refresh.