5 ms·
One of the professors in my PhD programme made his name by showing that anything you can do with a self-organising map, you can do with a Gaussian process. The
by soVeryTired 9y ago
One of the professors in my PhD programme made his name by showing that anything you can do with a self-organising map, you can do with a Gaussian process. The work essentially killed off self-organising maps as a field of study.
One time we had seminar where a guest speaker had used self-organising maps, and the professor literally fell off his chair.
- 98324298f 9y agoInteresting, can you link the paper please? I'd love to read it
- lkozma 9y agoThey probably referred to this: https://en.wikipedia.org/wiki/Generative_topographic_map https://en.wikipedia.org/wiki/Generative_topographic_map
- nabla9 9y agoHow exactly did theoretical equivalence result kill SOM's as field of study? There is similar connection between deep fully connected neural networks and Gaussian process. Deep Neural Networks as Gaussian Processes https://arxiv.org/abs/1711.00165 https://arxiv.org/abs/1711.00165
- soVeryTired 9y agoIn much the same way the result you linked to slowed research into single-hidden-layer NNs. Another example is when the failure to learn XOR killed off the preceptron and nearly killed neural networks as a whole. Researchers like low-hanging fruit, and some results just make a topic look barren.
- nabla9 9y ago> result you linked to slowed research into single-layer NNs. You are thinking different link.
- soVeryTired 9y agoYep, sorry. I'm thinking of the (much older) results referred to in the abstract of the linked article.
- voxl 9y agoProving two things equivalent does not invalidate the use of one or the other.