Sessions

What and Why and How: Apache Drill 1.0

dbe — Mon, 01 Jun 2015 13:21:12 +0000

Speaker(s):

The 1.0 release of Apache Drill does SQL on Hadoop, but with some big differences.
The biggest difference is that Drill changes SQL from a strongly typed language into a late binding language without losing performance. This allows Drill to process complex structured data in addition to relational data. By dynamically generating code that matches the data types and structures observed in the data, Drill can be both agile as well as very fast. Drill can analyze complex data directly with no ETL steps.

Mon, 06/01/2015 - 17:00 to 17:40

Room:

Stage 1

So you want to be a Data Science consultant (or hire one)? 10 things you should know.

dbe — Fri, 22 May 2015 12:10:47 +0000

Speaker(s):

Radim Řehůřek

More and more people prefer the independent life style of freelancing and increased mobility. At the same, IT workers and machine learning and data experts in particular are in great demand, giving them the luxury of choice. In this talk I'll discuss my 10+ years of industry experience, working first as an independent data mining consultant and later leading a team of such work-for-hire data science experts. I'll share some practical tips for those interested in walking the same path, or for companies wanting to contract such teams.

Mon, 06/01/2015 - 15:40 to 16:10

Room:

Stage 4 / Open Stage

Analyzing and Searching Streams of Social Media at Scale using Spark, Kafka and Elasticsearch

dbe — Mon, 18 May 2015 16:51:16 +0000

Speaker(s):

Markus Lorch

In this session we like to share our experiences from analyzing streams of Twitter data with Apache Spark Streaming in near real-time, leveraging Apache Kafka as a HA messaging backbone plus storing and searching for Tweets in Elasticsearch at a large scale. Key design aspects are short end-end processing delays, sub-second search responses and a highly available system that does not rely on hardware redundancy.

Mon, 06/01/2015 - 17:00 to 17:30

Room:

Stage 4 / Open Stage

New CUDA Concepts and APIs

dbe — Fri, 15 May 2015 14:00:57 +0000

Speaker(s):

Kashif Rasul

In this workshop I will discuss and teach the more advanced CUDA APIs and concept. CUDA is Nvidia's general purpose computing framework for graphic cards and allows you to scale your computations thanks to its multi-processor paradigm. Recent versions of CUDA has a number of new features and APIs which can potentially save developers time as well as provide highly optimized and fast algorithms if used properly. In this workshop we will teach you how to get the most out of these accelerator cards.

Tue, 06/02/2015 - 11:20 to 12:20

Room:

Stage 4 / Open Stage

Prediction @ Criteo: learning at scale on Hadoop

dbe — Mon, 11 May 2015 10:32:10 +0000

Speaker(s):

Olivier Toromanoff

Machine Learning is at the heart of the Criteo platform : Sourcing the inventory and generating the banners for 2 billions daily displays require 15 millions predictions per second. In this talk, we will present the infrastructure that allow us to train and deploy 1700 models a day as well as the tooling we set up to continuously monitor and improve this pipeline.

Mon, 06/01/2015 - 12:10 to 12:40

Room:

Stage 4 / Open Stage

Recommendation at scale

dbe — Wed, 06 May 2015 12:37:39 +0000

Speaker(s):

Simon Dollé

Behavioral retargeting consists of displaying online advertisements that are personalized according to each user’s browsing history. At this point, the selection of the products to display in the banner needs to be fast and accurate. At Criteo, we built a recommender system which is able to choose a dozen of relevant products from over two billion products in a few milliseconds. In this talk, we will expose the problems we faced whilst building this system and how we solved them thanks to a mix of online and offline computations.

Tue, 06/02/2015 - 16:30 to 17:00

Room:

Stage 4 / Open Stage

Automating Cassandra Repairs

dbe — Thu, 30 Apr 2015 08:52:53 +0000

Speaker(s):

Radovan Zvoncek

I'd like to do a short talk about Cassandra Reaper [1]. Cassandra Reaper is a service making it easy to manage repairs in Cassandra. The motivation to start building the Reaper came from our long-term pain we had with runningrepairs. Nobody in the Cassandra community seemed to be looking into that (despite facing similar pain as we), so we went ahead.

We've been building the Reaper as an open-source project from day 1. We already got a few minor contributions, as well as a larger one [2].

Tue, 06/02/2015 - 12:30 to 13:00

Room:

Stage 4 / Open Stage

Using Apache Spark for Graph Computation with Neo4j

dbe — Mon, 27 Apr 2015 08:33:50 +0000

Speaker(s):

Michael Hunger

This talk will introduce Mazerunner, an integration of Apache Spark with Neo4j that can be used to offload expensive global graph compute algorithms to a scalable cluster. Mazerunner uses a Docker based orchestration to set up the different components and reads and writes transactionally from a running Neo4j instance to the Spark cluster using an persistent queue. We will discuss the general architecture, walk through the setup and demo two different graph algorithms PageRank and Betweenness Centrality on the DBPedia
dataset.

Mon, 06/01/2015 - 14:30 to 15:30

Room:

Stage 4 / Open Stage

Kibana 4 - towards a beer analytics engine

dbe — Mon, 27 Apr 2015 08:07:22 +0000

Speaker(s):

Christoffer Vig

(Ab)Using the ELK stack to analyze information from the Norwegian government owned alcohol monopoly, making it help us answer important questions such as: What beer gives the most alcohol for the money? What terms characterize Belgian beers? (aka Belgianness) What beer contains the maximum Belgianness?

Mon, 06/01/2015 - 13:00 to 13:30

Room:

Stage 4 / Open Stage

Understanding What’s Important

dbe — Thu, 23 Apr 2015 15:44:52 +0000

Speaker(s):

Grant Ingersoll

Family is important. Work is important. Data is important. We are constantly inundated with messages telling us what is important, but how do we really determine what is important? Focusing on the work and data sides of this challenge, we’ll explore technologies and approaches that can help us better understand what is important as well as some key challenges to be tackled in this realm.

Tue, 06/02/2015 - 15:20 to 16:20

Room:

Stage 4 / Open Stage