In contextual linear bandits, the reward function is assumed to be a linear combination of an unknown reward vector and a given embedding of context-arm pairs. In practice, the embedding is often learned at the same time as the reward vector, thus leading to an online representation learning problem. Existing approaches to representation learning in contextual bandits are either very generic (e.g., model-selection techniques or algorithms for learning with arbitrary function classes) or specialized to particular structures (e.g., nested features or representations with certain spectral properties). As a result, the understanding of the cost of representation learning in contextual linear bandit is still limited. In this paper, we take a systematic approach to the problem and provide a comprehensive study through an instance-dependent perspective. We show that representation learning is fundamentally more complex than linear bandits (i.e., learning with a given representation). In particular, learning with a given set of representations is never simpler than learning with the worst realizable representation in the set, while we show cases where it can be arbitrarily harder. We complement this result with an extensive discussion of how it relates to existing literature and we illustrate positive instances where representation learning is as complex as learning with a fixed representation and where sub-logarithmic regret is achievable.

Abhinav Agrawal is the CEO of Rocket, an AI-enhanced recruiting agency as well as Hireflow.ai, The pandemic has inspired fear of the incentive to replace employees with machines. As the co-founder of an AI-powered recruiting organization, I expect that AI won't displace recruiting jobs. In fact, I anticipate that AI will become an indispensable tool for recruiters. Ask any recruiter: What's the most tedious aspect of your job?

Third Space Automation, a small technology startup with offices in London, Helsinki, and Bangalore, has partnered with a few retailers to conduct pilot tests delivering take out and groceries via AI enabled drones.

If you've ever shopped online (*cough* Amazon *cough*), you've probably experienced the "vacuum cleaner effect". You carefully buy one expensive item (e.g. a vacuum cleaner) and then you receive dozens of recommendations for other vacuum cleaners to buy: by email, everywhere on the retailer's website, or sometimes in the ads you see on other websites. In other terms, Amazon is a 1 trillion dollar company that employs hundreds of data scientists and is incapable of understanding that if you bought an expensive appliance, buying another one of the same category in the next weeks is what you're *least* likely to do! But let's think about the problem for a second. Suggesting item that are similar to what you just bought is actually the core feature of recommendation algorithms!

And if you talk to experts in the science of machine learning, you might even learn that they don't really recognize artificial intelligence as a technology but more as a marketing buzzword used to sell machine learning. So, for the sake of simplicity, understand that machine learning is one of the most effective and mature approaches to realizing algorithms that make programs and machines seem "intelligent." Well, we're going to go a lot deeper than that to help you understand how machine learning is the key force shaping the world of artificial intelligence. Machine learning is mostly based on using lots and lots of training data and good algorithms. Though there's a lot of excitement in technology circles about sophisticated algorithms, particularly deep learning, it must be understood that most applications of machine learning are a result of good data.

Choosing a machine learning algorithm is a lot like shoe shopping. If it were, we'd all be wearing thousand-dollar feather-light track shoes. Instead, we consider how we'll be using them. Some shoes are good for standing all day, and some are good for climbing cliffs. And, of course, how they look can trump everything else.