Knowledgefeed | Vienna Data Science Group

Knowledgefeed vol. 29: Efficient ML for mobile devices & Segmentation in Surveys

Thu, 12 Dec 2019 19:00:00 +0000

Dear Community,

We hope you had a great summer! We are very happy to announce that the VDSG is back from summer break, too, and our next meetup is happening in two weeks in A1!

This time we are having two talks: Segmentation in surveys and efficient machine learning for mobile devices.

Jelena Milosevic: “Efficient machine learning for mobile devices” (30-35 mins)

Increased amount of data allows for better training and more accurate machine learning systems. Big part of generated data today is coming from embedded and mobile devices, whose number is constantly on the rise. In order to fully profit from the collected information, artificial intelligence should come to these devices. However, this is currently difficult to achieve, mostly due to the computational demands of machine learning systems being too high for constrained computational resources of mobile devices. One of the reasons for this is that, when designing machine learning methods, most people only focus on accuracy, without taking into account constrained computational resources of developed solutions.

Many applications rely on efficient machine learning. Some examples are: vision and image processing, autonomous driving, and malware detection. In order to facilitate novel applications in these domains, it is of utmost importance to provide machine learning solutions that are not just accurate, but also suitable for constrained environments. In my talk I will discuss how we can design and develop such solutions suitable to be used in real-time, on-device, and at the same time customizable with respect to application requirements (accuracy, inference time, and power consumption).

Jelena Milosevic is passionate about machine learning and cybersecurity. She is currently a postdoctoral researcher at the Institute of Telecommunications, TU Wien, where she designs and develops machine-learning-based methods for detection of cyber attacks. She obtained her PhD in 2017 from Faculty of Informatics, University of Lugano, Switzerland, where her main focus was on the malware detection systems suitable for runtime usage on resource constrained systems and based on machine learning methods of low complexity. Previously, Jelena was an intern at IBM Cyber Security Center of Excellence in Beer Sheva, Israel, where she worked on the time-series analysis for anomaly detection and in Movidius an Intel Company in Dublin, Ireland, where she worked on the design and development of deep learning methods suitable for embedded environments. https://www.linkedin.com/in/milosevicjelena/

Marcin Kosiński: “Segmentation in Surveys using NMF” (30-35mins)

Working with high dimensional data? Often facing the need to group observations? This presentation is for you.

Segmentation should be balanced and distinctive, the discovered over- and under-indexed features within segments should create a meaningful story, and, ideally, the amount of differentiative factors that drives segmentation should be small.

The last requirement often becomes a bottleneck in a survey where respondents are asked an enormous amount of questions. One solution is the nonnegative matrix factorization that, in one attempt, segments respondents and their features! The concept of the NMF decomposition and applications in R will be presented with the explanation of diagnostic plots.

Marcin has a master degree in Mathematical Statistics and Data Analysis specialty. Challenges seeker and devoted R language enthusiast. In the past, keen on the field of large-scale online learning and various approaches to personalized news article recommendation. Community events host: organizer of Why R? conferences whyr.pl. Interested in R packages development and survival analysis models. Currently explores and improves methods for quantitative marketing analyses and global surveys at Gradient Metrics. https://www.linkedin.com/in/marcin-kosi%C5%84ski-81435aab/

Stay tuned and RSVP on our Meetup: https://www.meetup.com/Vienna-Data-Science-Group-Meetup/events/264362828/

Slides are available here.

You can find slides for this talk here.

Knowledgefeed vol. 30: How to get into Kaggle?

Thu, 12 Dec 2019 19:00:00 +0000

How to get into Kaggle? by Philipp Singer and Dmitry Gordeev from Vienna Data Science Group

Our next meetup is dedicated to Kaggle community. Two seasoned Kagglers share their experiences and tricks with you about how to start with Kaggle and how to be successful in Kaggle Competitions.

Doors open at 18:00. Hope to see you there!

Kaggle is one of the largest online communities for data scientists specifically known for their competitions where participants aim to solve data science challenges. Kaggle has a long history of varying types of competitions from different areas such as medicine, finance, scientific research, or sports focusing on different types of data and prediction problems such as tabular data, time series, NLP, or computer vision. In this meetup, Philipp and Dmitry who are one of the top contenders on Kaggle and usually compete together under the name “The Zoo”, will give an introduction and an overview of their journey. In detail, they will cover the following topics:

– What is Kaggle? – What types of competitions do exist on Kaggle? – Overview of competitions “The Zoo” participated – Short overview of utilized methods and main principles – Where and how can you start to compete?

SPEAKERS

Philipp Singer currently works as a data scientist in UNIQA Insurance Group. Previously, he was working as a post-doctoral researcher at the Data Science Lab of GESIS. Philipp obtained a PhD in computer science with honors at the Technical University of Graz where he also finished his Master studies in Software Development and Business Management. During his career he has obtained several awards including multiple winning and top placements on Kaggle as well as several scientific honors such as a best paper award at the renowned World Wide Web Conference. Kaggle profile: https://www.kaggle.com/philippsinger

Dmitry Gordeev currently works as a data scientist in UNIQA Insurance Group. He holds a master degree in data mining and has in-depth experience of machine learning applications in financial institutes, focused on risk management and credit scoring. He has a track record of multiple winning and top placements in kaggle competitions, currently possessing a title of kaggle competitions grandmaster. Kaggle profile: https://www.kaggle.com/dott1718

AGENDA

18:00-18:30 get together 18:30-18:40 intro speech VDSG 18:40-19:40 talks 19:40-NA socializing

For Location and RSVP follow the link: https://www.meetup.com/Vienna-Data-Science-Group-Meetup/events/266382385/

Knowledgefeed vol. 27: Deep Learning for Predictive Quality & Predictive Maintenance

Tue, 23 Apr 2019 19:00:00 +0000

Get ready for our 27th Knowledgefeed featuring characteristics of Industrial AI and how state of the art deep learning methods can be applied to solve complex problems and bring more value to companies.

Talk (duration 60 – 90 mins): Predictive Maintenance, Predictive Quality & Visual Inspection

By Simon Stiebellehner -Head of AI, craftworks & Daniel Ressi -Data Scientist, craftworks

Artificial Intelligence plays a major role in Industry 4.0 and more industrial companies than ever are starting to utilize their data to gain value and insights. The industrial domain offers very promising opportunities but this potential also comes with very specific requirements and challenges. This talk gives insights into the characteristics of Industrial AI and how state of the art deep learning methods can be applied to solve complex problems and bring more value to companies. Based on real use cases, three common areas of Industrial AI and the applied modelling approaches will be presented:

Predictive Maintenance: Can faults of machines be predicted in advance?
Visual Inspection: Can computer vision automatically assess the quality of products?
Predictive Quality: Can product defects be predicted in advance and prevented in future?

Simon is a Head of AI at craftworks and lecturerin statistics and digital marketing at WU Wien and FH Wien. After having completed his Bachelorin Information Systems, he gained diverse industry experience,ranging from Microsoft to global players of the consulting industry. Subsequently, Simon obtained his Masters degree from University College London (UCL), specializing in Machine Learning and Data Science. Afterwards, he was a doctoral candidate and research associate, conducting research at the intersection of Neural Probabilistic Language Models and Recommendation Systems in a Real-Time Bidding context.

LinkedIn: https://www.linkedin.com/in/simonstiebellehner/ Website: https://craftworks.at

Daniel is a Data Scientist at craftwork and develops customized deep learning solutions for industrial clients. His background is in Biomedical Engineering, where he focused his research on Recurrent Neural Networks for Brain Machine Interfaces (BSc, TU Graz) and for Computational Neuroscience (MSc, Imperial College London). Craftworks develops award-winning artificial intelligence solutions for industrial enterprises. Their customers range from the automotive to the paperindustry and everything in between.

LinkedIn: https://at.linkedin.com/in/daniel-ressi Website: https://craftworks.at

As always: grab the opportunity to ask our lecturers questions, discuss your ideas and of course enjoy the company of some interesting folks!

Please do not hesitate to present your own projects, ideas or thoughts. We are more than gladly sharing the stage with you! Please refer to our blog-post / Discussion entry for further details:

https://viennadatasciencegroup.at/2016/09/06/power-to-the-people/

Looking forward to meeting you at the Knowledgefeed!

Stay tuned and RSVP on our Meetup:

https://www.meetup.com/Vienna-Data-Science-Group-Meetup/events/258669227/

Knowledgefeed vol. 28: (Big) Data (Science) for Security

Tue, 23 Apr 2019 19:00:00 +0000

Dear Community,

Get ready for our 28th Knowledgefeed featuring security research question that motivated an ambitious data collection and analysis approach.

While a common opinion is that a collection of “big data” only leads to security and privacy problems – and it often does – however the analysis of large amounts of IT systems behavioral data also enables new experimental approaches to improve IT security and protect us from the cesspool of malware that the internet is.

The TARGET research project at St.Pölten UAS started with an initial IT security research question that motivated an ambitious data collection and analysis approach, and that then spawned subsequent challenges in data collection, encoding, storage, processing, experimental algorithm implementation, up to production deployment.

This talk reflects on the challenges encountered and experiences made.

Talk (duration 45- 60 mins)

By Martin Pirker who is a Senior Researcher at the Institute of IT Security Research, St. Pölten University of Applied Sciences. His current work focus is the Josef Ressel Center for Unified Threat Intelligence on Targeted Attacks (TARGET), and all kinds of weird problems that arise when IT (security) meets privacy and big data.

As always: grab the opportunity to ask our lecturers questions, discuss your ideas and of course enjoy the company of some interesting folks!

Website: https://isf.fhstp.ac.at/en https://research.fhstp.ac.at/en/projects/josef-ressel-center-for-unified-threat-intelligence-on-targeted-attacks-target https://isf.fhstp.ac.at/en/forschende/martin-pirker

VDSG Knowledgefeed vol. 24

Sat, 30 Jun 2018 19:00:00 +0000

Download Krista’s presentation here , an Christian’s presentation here .

VDSG – Knowledgefeed vol. 25: Managing spatio-temporal data using GeoMesa

Sun, 17 Jun 2018 19:00:00 +0000

Dear Community,

get ready for our 25th Knowledgefeed on Monday, June 25 – the last one before our summer hiatus:

Managing massive amounts of spatio-temporal data using GeoMesa

Description: With recent improvements in tracking and communications technology, we live in an age where movement data is collected on a big scale. This data harbors enormous potential for decision makers but we face the challenge of extracting relevant information from these data sets. This talk introduces GeoMesa, an open source solution for storing, analyzing, and visualizing massive spatio-temporal datasets on top of common big data tools, such as Accumulo, HBase, or Cassandra.

Duration of talk: 30 min

Anita Graser is spatial data scientist, open source GIS advocate, and author with a background in information technology and geographic information systems. She is working with the Center for Mobility Systems at the Austrian Institute of Technology in Vienna.

As always: grab the opportunity to ask our lecturers questions, discuss your ideas and of course enjoy the company of some interesting folks!

Please do not hesitate to present your own projects, ideas or thoughts. We are more than gladly sharing the stage with you! Please refer to our blog-post / Discussion entry for further details:

Stay tuned and RSVP on our Meetup:

https://www.meetup.com/Vienna-Data-Science-Group-Meetup/events/251584992/