Technical workshop

Simone Scardapane - Rome ML & Data Science Meetup

Challenge #2

Make news real again!

Your challenge is to build a model which will make it easier and more efficient to identify what really is fake news and what is not (+ everything in between). You model should be able to determine a level of credibility, content authenticity, and limit the viral spread of fake content, including fake images.

Sample dataset: https://www.kaggle.com/mrisdal/fake-news

The dataset contains text and metadata from 244 websites [...] Each website was labeled according to the BS Detector [...] There are (ostensibly) no genuine, reliable, or trustworthy news sources represented in this dataset (so far), so don't trust anything you read.

Another example dataset: http://www.fakenewschallenge.org/

Stance Detection involves estimating the relative perspective (or stance) of two pieces of text relative to a topic, claim or issue. [...] we have chosen the task of estimating the stance of a body text from a news article relative to a headline. Specifically, the body text may agree, disagree, discuss or be unrelated to the headline.

Baseline system: bag-of-words + standard classifier

Text Classification using Neural Networks [machinelearning.co]

Advanced concepts: vector space models

Vector representations of words [TensorFlow]

Advanced concepts: bidirectional recurrent networks

Neural Networks, Types, and Functional Programming [colah.github.io]

Additional topics

• Handling metadata information
• Fake images
• Additional data sources

• Bag of words meets bag of popcorns [Kaggle tutorial]
• Natural language understanding [IBM Bluemix]

Challenge #3

Emotional AI

Your challenge is to build a model (from scratch or on top of an existing model) that is smart enough to recognize various emotions through voice or facial expressions.

Example dataset:

https://www.kaggle.com/c/challenges-in-representation-learning-facial-expression-recognition-challenge/

Convolutional neural networks

Understanding convolutional neural networks for NLP [WildML.com]

Google Cloud ML - Recognizing emotions

Microsoft cognitive services

Google Video Intelligence API

Example dataset 2

http://yugangjiang.info/research/VideoEmotions/index.html

Another dataset:

https://research.google.com/youtube8m/

Good luck to all teams!