Like every other sphere of our modern lives, artificial intelligence (AI) has also made its way into civil aviation. The AI market, created by giant tech companies that develop software to help pilots and passengers alike, is growing. According to a market report, AI in the aviation sector is worth $152 million today and is expected to rise in value to $2.2 billion in 2024. In the U.S. and Europe, airline giants have been using AI for the last few years. Most recently in April, Air Canada and WestJet, one of Canada's largest airlines, announced that they were investing in AI technologies. WestJet CEO Ed Sims said that they are planning to launch a "virtual concierge service," like Amazon's Alexa or Google Home.

There's no time like the present to toss items in your life that are so 2018 or before. Rocked by data privacy scandals, social media tampering, and #MeToo movement-inspired management oustings, it's easy to get stuck in the storyline that 2018 was a year that the tech industry would like to completely forget. To be clear, those tech company-instigated concerns are all extremely important issues and will likely impact not just the tech world, but our overall society for many years to come. However, there were also quite a few positive and forward-looking developments that came from the tech industry this year as it's struggled through its awkward adolescent stage and into full adulthood. For one, 2018 was the year when the idea of voice computing went mainstream.

If 2019 is the year to sell your house or condo, let an Amazon Alexa help you out a bit. Just because your house was built back in the dark ages, before wireless internet, Bluetooth technology, and smartphones, you can still turn your house into a smart house, with a small monetary investment, and take some of the stress out of selling. Here's what I'd recommend to home sellers. If you're going with Alexa (like I did), then you should consider getting an Echo, Echo Plus, or Echo Dot from Amazon. If you go with Google, there's the Google Home and the Google Home Mini, while Apple offers the HomePod.

These raised expectations are likely due to customer experience leaders such as Amazon and Disney -- companies that raise the stakes for every other business and are masters at using personalization as a brand differentiator. Marketers know how important personalization is to today's consumers, however, delivering personalization at scale remains a challenge for many brands. That's where technology can do the heavy lifting. By using tools that help you automate personalization, make data-driven decisions, and streamline the process of creating dynamic content, you can quickly build personalized, omni-channel experiences at scale.

Looking back at Microsoft's 2018, you could make the argument that the company ended on an all-time high. But our readers buy Microsoft products, not Microsoft stock. From that perspective (and with all due respect to Microsoft's enterprise business, which isn't part of what we cover here at PCWorld), Microsoft's record was conservative and somewhat underwhelming, with a few exceptions. Microsoft added merely a flourish or two to its existing Surface lineup, for instance, and both feature updates to Windows 10 have turned out to be fairly inconsequential. As we've done in recent years, we list the highlights, low points, and yes, "what the hell was that??!" moments, closing with what we think Microsoft needs to work on most for 2019. Everyone loves a killer tech demo, and Microsoft showed off a doozy at its Build conference: a conference room of the future where Cortana could both hear and see, identifying users as they walked in.

A common assumption in recommender systems (RS) is the existence of a best fixed recommendation strategy. Such strategy may be simple and work at the item level (e.g., in multi-armed bandit it is assumed one best fixed arm/item exists) or implement more sophisticated RS (e.g., the objective of A/B testing is to find the best fixed RS and execute it thereafter). We argue that this assumption is rarely verified in practice, as the recommendation process itself may impact the user’s preferences. For instance, a user may get bored by a strategy, while she may gain interest again, if enough time passed since the last time that strategy was used. In this case, a better approach consists in alternating different solutions at the right frequency to fully exploit their potential. In this paper, we first cast the problem as a Markov decision process, where the rewards are a linear function of the recent history of actions, and we show that a policy considering the long-term influence of the recommendations may outperform both fixed-action and contextual greedy policies. We then introduce an extension of the UCRL algorithm ( L IN UCRL ) to effectively balance exploration and exploitation in an unknown environment, and we derive a regret bound that is independent of the number of states. Finally, we empirically validate the model assumptions and the algorithm in a number of realistic scenarios.

A common assumption in recommender systems (RS) is the existence of a best fixed recommendation strategy. Such strategy may be simple and work at the item level (e.g., in multi-armed bandit it is assumed one best fixed arm/item exists) or implement more sophisticated RS (e.g., the objective of A/B testing is to find the best fixed RS and execute it thereafter). We argue that this assumption is rarely verified in practice, as the recommendation process itself may impact the user’s preferences. For instance, a user may get bored by a strategy, while she may gain interest again, if enough time passed since the last time that strategy was used. In this case, a better approach consists in alternating different solutions at the right frequency to fully exploit their potential. In this paper, we first cast the problem as a Markov decision process, where the rewards are a linear function of the recent history of actions, and we show that a policy considering the long-term influence of the recommendations may outperform both fixed-action and contextual greedy policies. We then introduce an extension of the UCRL algorithm ( L IN UCRL ) to effectively balance exploration and exploitation in an unknown environment, and we derive a regret bound that is independent of the number of states. Finally, we empirically validate the model assumptions and the algorithm in a number of realistic scenarios.

The determinantal point process (DPP) is an elegant probabilistic model of repulsion with applications in various machine learning tasks including summarization and search. However, the maximum a posteriori (MAP) inference for DPP which plays an important role in many applications is NP-hard, and even the popular greedy algorithm can still be too computationally expensive to be used in large-scale real-time scenarios. To overcome the computational challenge, in this paper, we propose a novel algorithm to greatly accelerate the greedy MAP inference for DPP. In addition, our algorithm also adapts to scenarios where the repulsion is only required among nearby few items in the result sequence. We apply the proposed algorithm to generate relevant and diverse recommendations. Experimental results show that our proposed algorithm is significantly faster than state-of-the-art competitors, and provides a better relevance-diversity trade-off on several public datasets, which is also confirmed in an online A/B test.

We introduce a game-theoretic approach to the study of recommendation systems with strategic content providers. Such systems should be fair and stable. Showing that traditional approaches fail to satisfy these requirements, we propose the Shapley mediator. We show that the Shapley mediator satisfies the fairness and stability requirements, runs in linear time, and is the only economically efficient mechanism satisfying these properties.

Recommender systems have attracted much attention during the past decade. Many attack detection algorithms have been developed for better recommendations, mostly focusing on shilling attacks, where an attack organizer produces a large number of user profiles by the same strategy to promote or demote an item. This work considers another different attack style: unorganized malicious attacks, where attackers individually utilize a small number of user profiles to attack different items without organizer. This attack style occurs in many real applications, yet relevant study remains open. We formulate the unorganized malicious attacks detection as a matrix completion problem, and propose the Unorganized Malicious Attacks detection (UMA) algorithm, based on the alternating splitting augmented Lagrangian method. We verify, both theoretically and empirically, the effectiveness of the proposed approach.