Anne-Marie Kermarrec — Recommenders in practice: Debunking some myths

Abstract: Recommenders are the most prominent way to provide personalization in most applications. Highly popularized by Amazon and Netflix, they are now pivotal in almost all applications out there. While most research on recommenders have focused on improving the quality of the results (aka precision) so far, building an operational end efficient recommender goes far beyond. Recommenders come with many challenges beyond quality. One of the most crucial is their ability to scale to a large number of users and a growing volume of dynamic data to provide real-time recommendations, thus introducing many system challenges. Another challenge is related to privacy awareness: while recommenders rely on the very fact that users give away information about themselves, this potentially raises some privacy concerns. In this talk, I will focus on the challenges associated to building efficient, scalable and privacy-aware recommenders.

Bio: Anne-Marie Kermarrec is a senior researcher at Inria, France where she led a research group on large-scale distributed systems from 2006 to 2015. She is currently the CEO of the Mediego startup that she founded in April 2015. Mediego provides online predictive marketing services that directly leverage her recent research. She is also affiliated with EPFL, Switzerland. Before that, after her PhD thesis at University of Rennes in 1996, she has been with Vrije Universiteit, NL and Microsoft Research Cambridge, UK. Anne-Marie received an ERC grant in 2008 and an ERC proof of Concept in 2013. She received the Montpetit Award from the French Academy of Science in 2011 and the Innovation price in 2017. She is a member of the European Academy since 2013. She was named a 2016 ACM Fellow for contributions to large-scale distributed computing. Her research interests are in large-scale distributed systems and recommenders. She published more than 200 academic papers and received several best papers awards including the Test of time award at ACM/IEEE/ICIP Middleware conference in 2014 for her work on gossip-based peer sampling.

Emin Gün Sirer — Blockchains, The New Wave

Abstract: At once revolutionary and also over-hyped, blockchains represent a dilemma: the underlying technology proposed to date is thoroughly incapable of living up to a fraction of the dream that has been sold to the masses. Current crop of blockchains are unable to scale: if Venezuela switched to Bitcoin today, every adult would be able to go to store once every 36 days. And moderately popular smart contracts, like Cryptokitties, render Ethereum unusable for days. Yet the dream of Byzantine fault tolerant systems that control money flows, without trusted parties, at great scale, remains as compelling as it is unreachable.

This talk will focus on an exciting recent development in blockchain infrastructure, a novel consensus protocol family called Avalanche that combines the best features of the Lamport-Liskov and Nakamoto protocol families that preceded it, to yield a currency with low latency to finality, high throughput, and high degree of decentralization. I will outline the design of new currency applications built on top of the new primitives this new foundation provides, and discuss how the protocol’s inherent operation can help address the governance problems associated with cryptocurrencies. Finally, I will touch upon the next set of challenges that emerge when the consensus protocol is finally made efficient and scalable.

Bio: Emin Gün Sirer is an associate professor at Cornell University, a co-director of the Initiative for Cryptocurrencies and Smart Contracts, and the founder and chief scientist of bloXroute Labs. His research spans operating systems, networking and distributed systems. His current projects involve a novel secure operating system and system infrastructure for high-performance cloud computing applications. He likes building things, especially systems that have some principled reason for why they should work.