The ability to understand high-dimensional data, and to distill that knowledge into useful representations in an unsupervised manner, remains a key challenge in deep learning. One approach to solving these challenges is through disentangled representations, models that capture the independent features of a given scene in such a way that if one feature changes, the others remain unaffected.
State of the art in the representation disentanglement currently remains unestablished. There are two major reasons for that: (1) we did not settle on the formal definition of representation disentanglement yet, (2) as a consequence of that we did not yet agree on the best metric.
At the upcoming seminar, we will discuss the definition, evaluation metrics and state of the art candidates to representation disentanglement.
Speaker: Rauf Kurbanov.
Presentation language: Russian.
Date and time: September 10th, 8:00-9:30 pm.
Location: Times, room 204.
Videos from previous seminars are available at http://bit.ly/MLJBSeminars