Apache Spark + AI Munich - Car Classification + PySpark #Dec19

Nimmst du teil?

RSVPs schließen am 4.12. um 23:30.

151 Plätze frei

Teilen:
Bild des Veranstaltungsortes

Details

We are delighted to announce our car-focused event this December in Munich.

This event will be hosted by Datainsights together with two great speakers from the community:

* Dr. Evan Eames (Datainsights)

* Jannis Bergbrede (Inovex)

Talk 1: Car Classification Using a Deep Convolutional Neural Network
(This talk assumes a basic-to-intermediate understanding of Neural Networks)

Talk 2: A Case for Isolated Virtual Environments with PySpark

TALKS:

Title:
Car Classification Using a Deep Convolutional Neural Network Abstract

Abstract:
In this talk, we will first go through a number of real Use Case examples, involving Convolutional Neural Networks (CNN), and then walk through the development, training, and eventual deployment of a heavy-duty full-scale CNN, in this case used to accomplish car model classification. Finally, we will discuss how this CNN can be easily reworked to accomplish a wide variety of other computer vision tasks.

Bio:
Dr. Evan Eames completed his PhD in Computational Astrophysics, in which he worked with early-universe full-numerical simulations. He now designs Deep Learning applications, with a focus on industry application. In his free time he's also tinkering with some interesting new ML architectures.
You can see some of his projects on his Github: https://github.com/EvanEames

Title:
A Case for Isolated Virtual Environments with PySpark

Abstract:
When deploying and using PySpark applications in production, it is often necessary to use Python libraries like pandas, numpy or custom packages on the worker nodes of a spark cluster. Since the cluster is often shared among many users, managing global packages and their versions quickly becomes a hopeless endeavor. Based on practical experience at Germany's largest car market, I will motivate the use of isolated virtual environments for individual jobs. Further, I will talk about best practices on how to build and distribute these environments, leveraging conda requirement definitions and spark-submit functionality.

Bio:
Jannis Bergbrede completed his Master in Business Informatics at the University of Mannheim. Now he develops big data applications at inovex while focussing on building spark data pipelines and bringing Machine Learning use cases to production. He loves the outdoors and enjoys going on hikes.

HOUSEKEEPING:

• 2 talks (each ca. 40 min incl. discussion)

• Networking, food & drinks

• Language: English

• There will be photos taken

• A list of registered users will be provided to the host

• Please bring your ID