When's the last time you chirped, "Hey Google" (or Siri for that matter), and asked your phone for a recommendation for good sushi in the area, or perhaps asked what time sunset would be? Most folks these days perform these tasks on a regular basis on their phones, but you may not have realized there were multiple AI (Artificial Intelligence) engines involved in quickly delivering the results for your request. In these examples, AI neural network models were used to process natural language recognition, and then also inferred what you were looking for, to deliver relevant search results from internet databases around the globe, but also targeting the most appropriate results based on your location and a number of other factors as well. These are just a couple of examples but, in short, AI or machine learning processing is a big requirement of smartphone experiences these days, from recommendation engines to translation, computational photography and more. As such, benchmarking tools are now becoming more prevalent, in an effort to measure mobile platform performance. MLPerf is one such tool that nicely covers the gamut of AI workloads, and today Qualcomm is highlighting some fairly impressive results in a recent major update to the MLCommons database.

Due to the growing privacy concerns, decentralization emerges rapidly in personalized services, especially recommendation. Also, recent studies have shown that centralized models are vulnerable to poisoning attacks, compromising their integrity. In the context of recommender systems, a typical goal of such poisoning attacks is to promote the adversary's target items by interfering with the training dataset and/or process. Hence, a common practice is to subsume recommender systems under the decentralized federated learning paradigm, which enables all user devices to collaboratively learn a global recommender while retaining all the sensitive data locally. Without exposing the full knowledge of the recommender and entire dataset to end-users, such federated recommendation is widely regarded `safe' towards poisoning attacks. In this paper, we present a systematic approach to backdooring federated recommender systems for targeted item promotion. The core tactic is to take advantage of the inherent popularity bias that commonly exists in data-driven recommenders. As popular items are more likely to appear in the recommendation list, our innovatively designed attack model enables the target item to have the characteristics of popular items in the embedding space. Then, by uploading carefully crafted gradients via a small number of malicious users during the model update, we can effectively increase the exposure rate of a target (unpopular) item in the resulted federated recommender. Evaluations on two real-world datasets show that 1) our attack model significantly boosts the exposure rate of the target item in a stealthy way, without harming the accuracy of the poisoned recommender; and 2) existing defenses are not effective enough, highlighting the need for new defenses against our local model poisoning attacks to federated recommender systems.

This paper presents a neural network recommender system algorithm for assigning vehicles to routes based on energy and cost criteria. In this work, we applied this new approach to efficiently identify the most cost-effective medium and heavy duty truck (MDHDT) powertrain technology, from a total cost of ownership (TCO) perspective, for given trips. We employ a machine learning based approach to efficiently estimate the energy consumption of various candidate vehicles over given routes, defined as sequences of links (road segments), with little information known about internal dynamics, i.e using high level macroscopic route information. A complete recommendation logic is then developed to allow for real-time optimum assignment for each route, subject to the operational constraints of the fleet. We show how this framework can be used to (1) efficiently provide a single trip recommendation with a top-$k$ vehicles star ranking system, and (2) engage in more general assignment problems where $n$ vehicles need to be deployed over $m \leq n$ trips. This new assignment system has been deployed and integrated into the POLARIS Transportation System Simulation Tool for use in research conducted by the Department of Energy's Systems and Modeling for Accelerated Research in Transportation (SMART) Mobility Consortium

YouTube relies heavily on AI to deliver content. The newest YouTube algorithms put a great deal of value on the average time that a person views any video, gives it a like or dislike, and comments. Similarly, the recommender system is one of the most powerful use cases of ML that is encountered by every one of us many times a day. Collaborative Filtering: This is a type where we tend to build collaborations between various users and items(videos). Matrix Factorization: It tries to dissolve both user and item vectors together thus decomposing them and providing YouTube with better comparison metrics.

Being one of the recurring themes of science fiction, we all have at least a vague idea of what Artificial Intelligence (AI) is. In countless novels and movies, often in the guise of an ominous and threatening machine, other times instead portrayed as a technology so futuristic as to seem almost unattainable, AI has risen to the role of a mysterious entity that mostly belongs to fiction. At least, this was the scenario a few years ago. More recently, in fact, we all came to know AI by witnessing its expansion and pervasiveness to a point where it constitutes a substantial part of our daily interactions, either via smart assistants (e.g., Amazon's Alexa or Apple's Siri, just to name a few), or even by the AI-powered features of our smartphones (e.g., facial and fingerprint screen unlocking, or photo enrichment), not to mention self-driving cars. However, despite being already deployed in millions of products and services around the world, AI is still something obscure.

A TikTok video showing a woman's shock at how much voice data Amazon has collected on her has received millions of views and been liked hundreds of thousands of times. "I requested all the data Amazon has on me and here's what I found", she said in a video. The woman has two Amazon Echo Dot speakers and another Echo device to control her smart home lightbulbs. "When I downloaded the ZIP file these are all the folders it came with," she said. The audio files reveal thousands of short voice clips, which she describes as "so scary".

It seems that the world is gradually turning everything into data, and we are storing data at lightspeed. How are banks capitalising on new sources of data and using it to develop new services? Our watches now track our sleep, our movement, our heart rate and much more. With it they can "coach" us with training plans that are specific to our individual capabilities. At the top end of smartwatches, you have the ability to pay for things, listen to music, get guided GPS navigation and more.

Artificial agents that support people in their daily activities (e.g., virtual coaches and personal assistants) are increasingly prevalent. Since many daily activities are social in nature, support agents should understand a user's social situation to offer comprehensive support. However, there are no systematic approaches for developing support agents that are social situation aware. We identify key requirements for a support agent to be social situation aware and propose steps to realize those requirements. These steps are presented through a conceptual architecture that centers around two key ideas: (1) conceptualizing social situation awareness as an instantiation of `general' situation awareness, and (2) using situation taxonomies as the key element of such instantiation. This enables support agents to represent a user's social situation, comprehend its meaning, and assess its impact on the user's behavior. We discuss empirical results supporting that the proposed approach can be effective and illustrate how the architecture can be used in support agents through a use case.

In addition to the new Pixel 6 and Pixel 6 Pro, Google also released more details about new capabilities that its Tensor chip enables. One of them is a much more intelligent way of handling those calls to businesses that sometimes have you waiting hours on end just to speak to a representative. Now, the Pixel will show you the current and projected wait times before you even place a call so you can call when it works for you. Additionally, when you do call and encounter an endless list of options (like, "Press 1 for branch location and hours" if you're calling a bank), you don't need to remember all of them carefully. Instead, Google will listen to them for you and show the automated menu options on the screen for you to tap.

Setting up the Nest Cam is quick and easy in the Google Home app. Setting up the camera took me about five minutes. Plug it into a power outlet, open the Google Home app, and follow the in-app instructions. Drywall anchors, screws, and a base with a built-in mounting bracket are also included in the box. The Google Home app is where you can view and manage notifications, as well as toggle video recordings settings.