• On Understanding Types, Data Abstraction, and Polymorphism
    Pour la rentrée de PWL Paris, Nicolas Rolland propose de nous intéresser à "On Understanding Types, Data Abstraction, and Polymorphism" de Luca Cardelli. http://lucacardelli.name/Papers/OnUnderstanding.A4.pdf Le programmeur est inevitablement confronté a 4 challenges dans sa pratique: les bugs, l'abstraction, l'interfacage avec des API, la maintenabilité. Dans ces quatre domaines, les types s'annoncent comme de veritables sesames, comme en temoigne leur utilisation sans cesse plus etendue chez des grands noms comme Facebook. Cet article, toujours aussi moderne malgré son âge, explore ce que sont les types et comment ils peuvent structurer notre pratique logicielle au delà de l'Agile, du TDD, du scrum etc.. Nicolas Rolland : Avec plusieurs casquettes, dont celle de programmeur, principalement fonctionnel (haskell, ocaml, etc..), j'aime apporter des réponses théoriques puissantes aux problèmes pratiques. Dans la revolution digitale qui n'en est qu'à ses balbutiements, il est important de réfléchir au delà de sa pratique afin d'anticiper les changements à venir, et c'est ce que je vous propose de faire ensemble.

    Criteo

    32 Rue Blanche · Paris

    9 commentaires
  • Paul Feyerabend's Against Method
    It's with great pleasure to announce that we'll have Tomas Petricek (http://tomasp.net/), PhD student at University of Cambridge, functional programer, and F# enthusiast, presenting on the 1975 book -- Against Method: Outline of an Anarchist Theory of Knowledge (https://en.wikipedia.org/wiki/Against_Method) by Paul Feyerabend (https://en.wikipedia.org/wiki/Paul_Feyerabend). This will be a little different! There's no open-access PDF version of the book around, but you can find at bookstores, Amazon (http://www.amazon.com/Against-Method-Paul-Feyerabend/dp/1844674428), or on the Internet... or so I've heard. Intro How is computer science research done? What we take for granted and what we question? And how do theories in computer science tell us something about the real world? Those are some of the questions that may inspire computer scientist like me (and you!) to look into philosophy of science. I’ll present the work of one of the more extreme (and interesting!) philosophers of science, Paul Feyerabend. In “Against Method”, Feyerabend looks at the history of science and finds that there is no fixed scientific methodology and the only methodology that can encompass the rich history is ‘anything goes’. We see (not only computer) science as a perfect methodology for building correct knowledge, but is this really the case? To quote Feyerabend: Science is much more 'sloppy' and 'irrational' than its methodological image. I’ll be mostly talking about Paul Feyerabend’s “Against Method”, but as a computer scientist myself, I’ll insert a number of examples based on my experience with theoretical programming language research. I hope to convince you that looking at philosophy of science is very much worthwhile if we want to better understand what we do and how we do it as computer scientists! Bio Tomas Petricek (http://tomasp.net/) (@tomaspetricek (https://twitter.com/tomaspetricek)) recently submitted his PhD thesis at University of Cambridge on context-aware programming languages. He is active in the functional programming community, contributes to various F# open-source libraries, writes and talks about F# and offers trainings and consulting at fsharpWorks (http://www.fsharpworks.com/). When he isn’t writing equations or code, he enjoys reading philosophy of science books and occasionally publishes an essay (http://tomasp.net/blog/tag/philosophy/) combining philosophy and computer science.

    Talent.io

    13 rue d'Uzès, 75002 · Paris

    5 commentaires
  • A Gentle Introduction to RTB Mathematics
    About the talk In this talk we will discuss the paper Optimal Real Time Bidding Strategies which provides a mathematical model of RTB auctions with the aim of fully solving the media-buying problem on a given audience: maximizing a KPI under a limited budget. As we will show, this can be achieved by using stochastic control techniques. Once the model is presented, we will discuss some interesting consequences both from the scientific point of view (e.g. bandits, prediction etc.) and from the industry point of view. https://arxiv.org/abs/1511.08409 About the speaker Engineer from Ecole Polytechnique and PhD in Mathematics (stochastic control theory, stochastic approximation, algorithmic trading), Joaquin Fernandez-Tapia is an active industrial researcher in the topic of mathematical optimization, control and online learning of low-latency auction systems.

    Criteo

    32 Rue Blanche · Paris

  • Neural methods in information retrieval with Maarten de Rijke
    Papers We Love Paris & Criteo Research are honored to have Dr. Maarten de Rijke, one of the world’s leading experts in information retrieval to inaugurate our Distinguished Lecture Series. About the talk Several communities have enthusiastically embraced and innovated neural algorithms in recent years. Examples include computer vision, machine translation, natural language processing. The information retrieval community is probably one of the next communities to jump on this train. In the talk Martin will discuss this ongoing trend, highlighting recent work on neural methods for text similarity, ad hoc retrieval, and predicting and modeling information interaction behavior (such as clicks and dwell time). The talk is based on joint work with Alexey Borisov, Evangelos Kanoulas, Tom Kenter, Ilya Markov, Pavel Serdyukov, Christophe Van Gysel. About the speaker Official bio and photo can be found here: https://staff.fnwi.uva.nl/m.derijke/bio/ .

    Criteo

    32 Rue Blanche · Paris

    3 commentaires
  • Natural Language Processing (almost) From Scratch
    Natural Language Processing (almost) From Scratch Link to the paper: https://arxiv.org/abs/1103.0398 Authors: Ronan Collobert, Jason Weston, Léon Bottou, Michael Karlen, Koray Kavukcuoglu, Pavel Kuksa Authors own abstract We propose a unified neural network architecture and learning algorithm that can be applied to various natural language processing tasks including: part-of-speech tagging, chunking, named entity recognition, and semantic role labeling. This versatility is achieved by trying to avoid task-specific engineering and therefore disregarding a lot of prior knowledge. Instead of exploiting man-made input features carefully optimized for each task, our system learns internal representations on the basis of vast amounts of mostly unlabeled training data. This work is then used as a basis for building a freely available tagging system with good performance and minimal computational requirements. In other, fewer words In this paper, the authors showed that, using neural networks, you could tackle very different NLP tasks in a more general manner than the existing techniques. Since this paper's publication, a lot of work and interesting results have emerged from this approach. About Yann Schwartz (@abolibibelot (https://twitter.com/abolibibelot)) My day job and interests revolve around distributed systems that stream buckets of bytes, most recently to perform anomaly and fraud detection at a French Ad Retargeting Company. My other interests often revolve around streams of characters, from books to NLP. My greatest achievement - and failure - was taking a sabbatical to write a novel on a remote island, and ending up making a zombie movie (http://www.imdb.com/title/tt4503126/?ref_=nm_flmg_dr_1) instead.

    Ce Meetup est passé

    Criteo

    32 Rue Blanche · Paris

    2 commentaires
  • Implementing location independent invocation, Christopher Meiklejohn
    We are very honored to welcome Christopher (@cmeik (https://twitter.com/cmeik)) again! Abstract A brief overview is presented of work on building a highly distributed office application based on mobile objects. The authors explain the techniques used to find the target of an invocation and describe how the technique is implemented. Location-independent invocation (LII) is presented as a conceptual service that is independent of any particular application, operating system, or programming language. LII completely removes remote call processing buildings from the view of application programmer. It is shown how LII can be implemented without language or system support in any environment that provides reliable interprocess communication. The indications for LII are studied, i.e. under what circumstances the proposed abstractions are beneficial. A description is given of an application domain in which LII is useful and the core services that support it. An object-finding algorithm is described. The relationship of LII to earlier work on object-finding and location independence is included Paper Distributed Computing Systems, 1989., 9th International Conference on Speaker Christopher Meiklejohn is a researcher, Ph.D. candidate and creator of LASP (http://lasp-lang.org (https://t.co/SvIEg5Sxvt)). Host Thanks to Xebia for hosting this session!

    Xebia

    156 boulevard Haussmann, 7e étage · Paris

    9 commentaires
  • Out of the tar pit - part II
    After explaining the basic concepts outlined in the paper "Out of the tar pit" by Ben Moseley and Peter Marks (full paper can be found here (http://shaffner.us/cs/papers/tarpit.pdf)), Romeu Moura (@malk_zameth (https://twitter.com/malk_zameth)) comes back for the second part of his talk: now that complexity is well defined, let us see how it can be addressed! Romeu will quickly recap what was said in the first talk, therefore all attendees are welcome! About the speaker Romeu Moura, consultant at Arolla (http://www.arolla.fr/) Writer of tales – on art, puzzles & crafts – to friends & compilers. Likes dialectics, paradigm jumping, serendipity, endless conversations.

    Arolla

    25 rue du Louvre · Paris

    2 commentaires
  • Out of the tar pit
    Time to revisit a classic! Romeu Moura (@malk_zameth (https://twitter.com/malk_zameth)) will talk about "Out of the tar pit" by Ben Moseley and Peter Marks. Full paper can be found here (http://shaffner.us/cs/papers/tarpit.pdf). About the speaker Romeu Moura, consultant at Arolla (http://www.arolla.fr/) Writer of tales – on art, puzzles & crafts – to friends & compilers. Likes dialectics, paradigm jumping, serendipity, endless conversations.

    Arolla

    25 rue du Louvre · Paris

    3 commentaires
  • Christopher Meiklejohn on "Logic and Lattices for distributed programming”
    Hello, we have a special summer session on distributed systems. Christopher Meiklejohn, Senior Engineer at Basho, will join us to talk about Neil Conway's "Logic and Lattices for distributed programming". More later. Huge thanks to Criteo for hosting us the second time.

    Criteo

    32 Rue Blanche · Paris

    5 commentaires
  • Session #1 - Harmful GOTOs, Premature Optimizations, and Programming Myths...
    Harmful GOTOs, Premature Optimizations, and Programming Myths Are the Root of All Evil Over the years our industry has accumulated folklore that shapes what we are as an industry. We repeat maxims like "Premature optimisation is the root of all evil" but perhaps ignoring the context on which that sentence was said. This ignorance has led us to creating myths that sometimes do more harm than good. In this talk I will review some of those myths, see their origin and at the same time make a historical review of what was going on in our industry in the '70s, when those myths appeared. As the old saying goes: Those who cannot remember the past are condemned to repeat it About the speaker Alvaro Videla works as a developer for RabbitMQ/Pivotal. Before moving to Europe he used to work in Shanghai where he helped building one of Germany biggest dating websites. He co-authored the book "RabbitMQ in Action" for Manning Publishing. Some of his open source projects can be found here: http://github.com/videlalvaro. Apart from code related stuff he likes traveling with his wife, listening/playing music and reading books. You can find him on Twitter as @old_sound (https://twitter.com/old_sound). About the host At Criteo (http://www.criteolabs.com), personalized performance advertising is what we do. And it’s what we do best. Known for the quality of our digital advertising technologies, developed by the 250+ world-class engineers working for our R&D in Paris, Grenoble and Palo Alto and having one of the biggest Hadoop clusters in Europe, our technology takes an algorithmic approach to determining what user we show an ad to, when, and for what products. We serve over 7500 advertisers worldwide and have direct relationships with over 10000 publishers. Criteo helps advertisers sell more.

    Ce Meetup est passé

    Ajouter un lieu

    17 commentaires