MLUGS

Protokoll

Vorstellungsrunde

• Andreas, Software-Entwickler, AX Semantics
• David, Digital-Analytics-Consultant, diconium
• Lukas, luftdaten.info
• Michael, Software-Entwickler, AX Semantics

Michael - Understanding LSTMs

http://colah.github.io/posts/2015-08-Understanding-LSTMs/

Michael erklärt uns den Blogpost und die Grafiken darin.

Andreas - dotAI Highlights

• https://www.dotai.io/
• https://www.youtube.com/user/dotconferences/videos

• mindmaps von https://twitter.com/SebastienElet/with_replies

• Vadim Markovtsev: Machine learning on source code

  • analyse von sourcecode mit embeddings
  • paper: https://arxiv.org/pdf/1805.11651.pdf
  • https://pbs.twimg.com/media/DehpuQaW0AEhm0g.jpg:large
  • GOOD

• Virginie Mathivet: AI vs Human

  • https://pbs.twimg.com/media/Dehp5rMW4AAYSdP.jpg:large

• Sharada Mohanty: Crowdsourcing Artificial Intelligence for Science

  • https://www.crowdai.org/
  • https://pbs.twimg.com/media/DehvRBfWAAIT7an.jpg:large

• Lightning talks

  • https://pbs.twimg.com/media/DeiCavfW0AAGO8L.jpg:large
  • Isabel Schwende: Deep Learning in the Wild

  • Olivier Wulveryck: Software 2.0, a Babel fish for deep learning?

    • https://github.com/gorgonia/gorgonia

  • Charlotte Ledoux: Logistics and data: towards a decision support tool

    • https://www.quantmetry.com/
    • AI supportet logistics (wenn man daten hat/sammelt)

  • Eliot Andres: Processing 100+ million images per month with deep learning

    • https://ndres.me/
    • kafka
    • tensorflow profiler: https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/profiler/README.md

• Peet Denny: Explainable AI

  • https://www.wealthwizards.com/
  • GOOD

• Jim Dowling: All AI Roads Lead to Distribution

  • https://www.kth.se/profile/jdowling
  • next generation Hadoop Distributed filesystems
  • https://medium.com/@jim_dowling/introducing-hops-hadoop-120c30d02676

• Hagay Lupesko: Model Serving for Deep Learning

  • AWS, https://github.com/awslabs/mxnet-model-server
  • highlight: kann nicht nur mxnet, sondern jegliche models - ersatz für tensorflow-serving auf aws
  • https://pbs.twimg.com/media/DeiVu1mWAAAO-8L.jpg:large
  • GOOD

• Chloé-Agathe Azencott: Using structure to select features in high dimension

  • https://pbs.twimg.com/media/DeibBisWAAEPNiJ.jpg:large
  • GOOD

• Aurélien Géron: Knowledge Graphs & Deep Learning @ YouTube

  • GOOD

• misc

  • swivel + tensorflow
    https://arxiv.org/pdf/1602.02215.pdf
    https://github.com/tensorflow/models/tree/master/research/swivel

  • LIME

    • https://github.com/marcotcr/lime
    • http://www.kdd.org/kdd2016/papers/files/rfp0573-ribeiroA.pdf

Andreas - CoNLL–SIGMORPHON 2018 Shared Task: Universal Morphological Reinflection

• https://sigmorphon.github.io/sharedtasks/2018/
• Starting point: https://arxiv.org/pdf/1707.01355.pdf
• Inspiration: http://www.aclweb.org/anthology/K17-2003

misc

Papers

• https://arxiv.org/pdf/1803.00590.pdf
  website: https://sites.google.com/view/hierarchical-il-rl
• https://arxiv.org/pdf/1806.01946.pdf

slides / blogposts

• ref zu ML auf Code: https://speakerdeck.com/campoy/machine-learning-on-source-code?slide=50
• wie funktioniert FaceID: https://towardsdatascience.com/how-i-implemented-iphone-xs-faceid-using-deep-learning-in-python-d5dbaa128e1d
  Code: https://github.com/normandipalo/faceID_beta/blob/master/faceid_beta.ipynb

Courses

• https://developers.google.com/machine-learning/crash-course/
• https://developers.google.com/machine-learning/practica/image-classification/

next

• erstmal Sommerpause
• Datum: 2018-09-18

• Vorträge

  • Uwe: Lösung von seines Problems unter Verwendung von DTW
  • (weitere Vorschläge sind willkommen!)