Did the Model
TensorFlow & Deep Learning SG
4 September 2018
- Machine Intelligence / Startups / Finance
- Moved from NYC to Singapore in Sep-2013
- 2014 = 'fun' :
- Machine Learning, Deep Learning, NLP
- Robots, drones
- Since 2015 = 'serious' :: NLP + deep learning
- & GDE ML; TF&DL co-organiser
- & Papers...
- & Dev Course...
About Red Dragon AI
- Google Partner : Deep Learning Consulting & Prototyping
- SGInnovate/Govt : Education / Training
- Products :
- Conversational Computing
- Natural Voice Generation - multiple languages
- Knowledgebase interaction & reasoning
whoami = DONE
- What is VQA ?
- VQA models
- "Did the Model Understand the Question?"
- Demo of VQA ...
- ... showing some limitations
- Example network described in original paper
- ~57% accuracy
VQA Modern Network
- VQA performance looks great!
- But is the model doing what we think it is?
- Test by changing the question :
- Remove words
- Add functionless phrases
- Similiar to adversarial attacks
- If only question word allowed is "
- model still more than 50% of original accuracy
- If all questions are
- model still more than 44% of original accuracy
- More generally : Model doesn't care much about nouns
- These phrases can kill performance :
- In not a lot of words, ...
- In not many words, ...
- What is the answer to : ...
- But these phrases do no harm :
- Answer this : ...
- Answer this for me : ...
- Tell me ...
- Black-box models can do cool stuff
- Good idea to test their limitations
- Hope that Explainability theme was useful
* Please add a star... *
- JumpStart module is Module #1 of 5
- Each 'module' will include :
- Individual Projects
- 70%-100% funding via IMDA for SG/PR
- 1-2 Nov : Advanced Computer Vision
- 22-23 Nov : Advanced NLP
- Location : SGInnovate
Deep Learning : Jump-Start Workshop
- Opportunity to do Deep Learning all day
- Work on something cutting-edge
- Location : Singapore
- Status : SG/PR FTW
- Need to coordinate timing...