Zalando interview question
- what is a support vector machine? - what's the difference between a Support Vector Machine (SVM) and a linear classifier? - why do I encode visual data and do not feed the SVM with raw information? - what are the convolutional neural networks and how do they work? - how would I treat a picture of a person that wear zalando clothes? - how could I deal with a multiclass classification problem with 1,000,000 different classes? - how could I possibly help zalando to improve their services?
Interview Answer
Anonymous
- I tried to explain how SVM work and what are their difference to a linear classifier - I tried to explain why I encode data and do not use raw information (discriminability) - I tried to give a high-level description of how cNN work - I suggested (obviously) to use an object detector that would discriminate the clothing that the human wears - I didn't know the answer (multi-label features) - I couldn't figure out a new computer vision technology that could improve zalando's online website :'( Overall, I believe that was a misunderstanding gap between me and the interviewer. As hard as I tried to explain and justify my answers there was always an upcoming question that surprised me and undermine my previous explanation. At the end I was too devastated and disappointed to answer the last few questions. Nevertheless it was a helpful useful experience that showed me new things to work on my next interview.