machine learning, ai, cs · slides

Paper overview | Do Deep Nets Really Need to be Deep?

Slide 1 of 59Slide 2 of 59Slide 3 of 59Slide 4 of 59Slide 5 of 59Slide 6 of 59Slide 7 of 59Slide 8 of 59Slide 9 of 59Slide 10 of 59Slide 11 of 59Slide 12 of 59Slide 13 of 59Slide 14 of 59Slide 15 of 59Slide 16 of 59Slide 17 of 59Slide 18 of 59Slide 19 of 59Slide 20 of 59Slide 21 of 59Slide 22 of 59Slide 23 of 59Slide 24 of 59Slide 25 of 59Slide 26 of 59Slide 27 of 59Slide 28 of 59Slide 29 of 59Slide 30 of 59Slide 31 of 59Slide 32 of 59Slide 33 of 59Slide 34 of 59Slide 35 of 59Slide 36 of 59Slide 37 of 59Slide 38 of 59Slide 39 of 59Slide 40 of 59Slide 41 of 59Slide 42 of 59Slide 43 of 59Slide 44 of 59Slide 45 of 59Slide 46 of 59Slide 47 of 59Slide 48 of 59Slide 49 of 59Slide 50 of 59Slide 51 of 59Slide 52 of 59Slide 53 of 59Slide 54 of 59Slide 55 of 59Slide 56 of 59Slide 57 of 59Slide 58 of 59Slide 59 of 59
1 / 59

The year 2014 passed under the huge Deep Learning sign. At the last seminar of computational neuroscience I presented a very recent article, which looks into the question of whether depth is a requirement for approximating a function, or, under the right conditions, shallow nets are capable of performing equally well.

Download PDF