DSPT#8 - Detecting Outliers? Cracking Semantics?! Oh boy, you're in for a ride!

"Bracaris"! No, this isn't a Game of Thrones episode, but just as epic: Data Science Portugal (https://www.facebook.com/datascienceportugal) is arriving at Braga!

Bringing you the latest on what's being done on the Data Science sphere of possibilities, this meetup will surely answer a lot of questions regarding topics such as Anomaly Detection and Natural Language Processing.

Be one of the first to attend this event and be prepared to network with other fellow Data Scientists, Engineers, Architects, DataOps... and Unicorns in general! First timers: we. have. beer!

The eighth meetup of Data Science Portugal (https://www.facebook.com/datascienceportugal) is going to take place on Wednesday, 5th April, 2017 around 18h30, at Startup Braga.

=== SCHEDULE ===

The preliminary agenda for the meetup is the following:

• 18:30-19:00: Welcome and Networking.

• 19:00-19:30: Talk 1: "Data Mining Anomaly Detection: Finding 'weirdness'" with João Brandão, Data and Artificial Intelligence Engineer at Robert Bosch.

• 19:45-20:00: Networking / Coffee Break.

• 20:00-20:30: Talk 2: "Have we cracked semantics? A practitioner's exploration into what’s possible" with Daniel Loureiro, Head of Data Science at Followprice.

• 20:45: Closing, hanging out and some beers

• 21:00: Dinner is optional but it might be an excellent opportunity for networking.

Do you want to be a sponsor in future meetups? Please contact us to info@datascienceportugal.com

See you there!

=== TALKS ===

Talk1: Data Mining Anomaly Detection: Finding "weirdness".

As a data nerd one of most enthusiastic things to me is to find things that are not supposed to happen and learn new insights about the business with these unexpected outcomes. One possible way to do it is using anomaly detection techniques that can identify abnormal and unusual behaviors in the data, so on this talk, we will discuss better what it is an anomaly, an outlier, and in which problems these techniques can be applied.

We will also see how some of these techniques cover some common situations like:

• You don’t have much data about these weird events.
• These historical events that you analyzed occurred for different reasons.
• You have unlabeled data- The data don’t follow a statistical distribution
• You have event-based data
• You have a real-time need

It will be an overview on how can statistics and machine learning help us with this type of tasks that are common on domains like Fraud, Intrusion Detection, Process monitoring, etc.

Short Bio:

João Brandão is a Data and Artificial Intelligence Engineer at Robert Bosch in Braga, where multimedia devices are developed and produced for use by the best known car brands. João is working on projects related to Industry 4.0, by creating Data-Driven products on Business Intelligence and Machine Learning that use manufacturing data content to predict future results and detect unusual events.

João previously worked on Business Intelligence at Farfetch for 3 years, while obtaining his MSc Degree in Informatics Engineering at Minho University with a focus on Business Intelligence and Intelligent Systems.

-------------------------------------

Talk2: Have we cracked semantics? A practitioner's exploration into what’s possible.

Words are complex things. Much beyond tokens in a string, they convey meanings and ideas that rely on your past knowledge and experience to be understood. With the arrival of word2vec in 2013, Data Scientists took a huge step in making some of that understanding available to machines.

We’ll start by looking at different word embedding techniques, such as word2vec, and how they capture relations such as king - man + woman = queen. Afterwards, we’ll get to play with them on a Jupyter Notebook and demonstrate how they can be used for search and recommendation in your own projects.

Short Bio:

Daniel Loureiro is currently Head of Data Science at Followprice, where he’s working on automatic product categorisation and non-collaborative recommendation. While finishing his degree in CS, he created his first startup (PepFeed) focused on information retrieval, and recently sold it to Followprice. He’s also an aspiring open-source contributor working on tools to help democratize data and industry-insights.