Past Recording
Share
Star()
Proving the Lottery Ticket Hypothesis: Pruning is All You Need
Thursday May 7 2020 14:00 GMT
Please to join the live chat.
Proving the Lottery Ticket Hypothesis: Pruning is All You Need
Why This Is Interesting

The lottery ticket hypothesis (Frankle and Carbin, 2018), states that a randomly-initialized network contains a small subnetwork such that, when trained in isolation, can compete with the performance of the original network. We prove an even stronger hypothesis (as was also conjectured in Ramanujan et al., 2019), showing that for every bounded distribution and every target network with bounded weights, a sufficiently over-parameterized neural network with random weights contains a subnetwork with roughly the same accuracy as the target network, without any further training.

Time of Recording: Thursday May 7 2020 14:00 GMT