AI Future meetup presentation about classifying toxic comments with Azure ML Services

1. anvykhod@Microsoft.com
https://www.linkedin.com/in/vykhand/
https://www.kaggle.com/vykhand

3. ▪
▪
▪
▪
▪

4. https://www.kaggle.com/jagangupta/stop-the-s-toxic-
comments-eda
https://www.kaggle.com/vykhand/toxic-fortune-cookie

7. Azure Machine Learning
Studio
(GUI interface, R, Python,
operationalization through webservices,
Azure only)
Azure Machine Learning Services
(code-first, Python (R comming), model
management, runs locally, in docker – Azure
/on-prem, use your own IDE, operationalize
through webservices)
Machine Learning Server
(on-prem & Azure**, license SQL Server,
R/Python, pushdown R/Python to
Hadoop/Spark, SQL Server,
Operationalization module – webservices,
interface R/Python, use your own IDE)
SQL Server Machine Learning
Services
(on-prem & Azure*, license SQL Server,
R/Python inside Stored Procedure,
operationalization SQL, interface SQL,
models stored in table)
ON-PREMISES
A ZURE
DATA SCIENTISTSBUSINESS
POWER USERS

8. Machine Learning & AI Portfolio
When to use what?
What engine(s) do you want
to use?
Deployment target
Which experience do you
want?
Build your own or consume pre-
trained models?
Microsoft
ML & AI
products
Build your
own
Azure Machine Learning
Code first
(On-prem)
ML Server
On-
prem
Hadoop
SQL
Server
(cloud)
AML services (Preview)
SQL
Server
Spark Hadoop Azure
Batch
DSVM Azure
Container
Service
Visual tooling
(cloud)
AML Studio
Consume
Cognitive services, bots
(cloud)
Data Lake
Analytics

10. https://github.com/vykhand/kaggle-jigsaw
Everything is a hyperparameter!

14. https://Aka.ms/dsvm
PyImageSearch review https://www.pyimagesearch.com/2018/03/21/my-review-of-microsofts-deep-learning-
virtual-machine/

15. Demo

16. Demo

17. http://sacred.readthedocs.io
where

21. Demo

22. https://www.kaggle.com/jagangupta/lessons-from-toxic-blending-is-the-new-sexy

23. Understanding GRUs
Understanding LSTMs
Deep learning for NLP best practices
Sentiment analysis for twitter with hierarchical
attention on DSVM

26. https://vykhand.github.io/kaggle-jigsaw/

28. https://buhrmann.github.io/tfidf-analysis.html

29. Glove
Fasttext
Good overview article
https://blog.acolyer.org/2016/04/21/the-amazing-
power-of-word-vectors/
https://www.depends-on-the-definition.com/guide-to-
word-vectors-with-gensim-and-keras/
https://radimrehurek.com/gensim/models/word2vec.html

33. Demo

34. 1st place
3rd place
5th place
34th
27th
Neptune team