Next Meetup

The Papers We Love!!
Hello Data Hackers. For our next meetup we will try out something new and host a “Papers we love” session. You may ask yourself what is that supposed to mean? It’s easy: the speakers chose a scientific paper they particularly liked and will present them to you. Let me tell you that the chosen papers are quite something, whether because of their impact, or their ingenious application of Data Science. ____________________________________________________ Christian Hammerschmidt, PhD Chris works on problems over discrete, tabular data without reliably labels, using recent advances from machine learning such as probabilistic programming and deep learning to deliver solutions with performance beyond the possibilities of traditional learning approaches when building user profiles and classifiers from weakly supervised data. If you have challenging problems between data science and applied machine learning, he’s your person to approach. He is a research associate in the SEDAN group of University of Luxembourg’s SNT. Paper he loves: Generative Adversarial Networks by Ian Goodfellow et al. https://arxiv.org/abs/1406.2661 Why he chose this paper: I chose this paper because it is the starting point of a wave by now over 300 novel papers within the short span of 3-4 years), giving us new tools to make use of data in unsupervised fashion. It has been hailed as “one of the most important developments in deep learning” by Yann LeCun, one of the deep learning field’s fathers. While I will focus on this paper, I will also spend some time to contextualize it and the impact it had so far. ____________________________________________________ Eric Falk, PhD Eric is a research associate at the SnT. He is working on data management topics, mostly in the realm of large databases and streaming data architectures. The business field he is foremost working in is mobile networking. Before joining the SnT in December 2014, Eric work in the field of software engineering in Luxemburgish financial industry. Paper he loves: The Case for Learned Index Structures https://arxiv.org/pdf/1712.01208.pdf Why he chose this paper: Well I am passionate about data management, therefore a lot of my work evolves around data access patterns and alike. That means I work a lot with data structures to optimize the access times. I liked the idea to use machine learning, to build models of database index structures. I think it shows an impressive and out of the box usage of machine learning. ____________________________________________________ Redouane Sassioui, M.Sc. Redouane Sassioui is a research associate at the Interdisciplinary Center for Security, Reliability and Trust (SnT). He received his M.Sc. degree from INRS, Montreal, Canada, in 2015. His research interests are in the areas of software engineering, machine learning and AI. He has worked on various data science projects in different areas including telecommunication, finance, real time marketing and avionics. Paper he loves: Playing Atari with Deep Reinforcement Learning https://arxiv.org/pdf/1312.5602.pdf Why he chose this paper: I believe that reinforcement learning (RL) will play an important role in the future of AI. I chose this paper because it was an important contribution to the filed of RL. It showed that RL with deep learning can surpass human performance in some Atari games. I will introduce reinforcement learning, explain the contribution of this paper and talk about the challenges we are facing in RL.

SnT - Luxembourg University

29 Avenue John F. Kennedy · Luxembourg

    Past Meetups (50)