news    articles    opinions    tutorials    concepts    |    about    contribute     republish

Tweets from #ICLR2019

May 16, 2019

ICRL, the International Conference on Learning Representations, was held May 6th to 9th 2019 in New Orleans.

Relive the conference through some of the top tweets (#ICLR2019).

Invited talks

Best Papers

Congratulations to the two ICLR 2019 Best Paper winners!

The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks (arXiV)
Jonathan Frankle · Michael Carbin

Abstract - Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of inference without compromising accuracy. However, contemporary experience is that the sparse architectures produced by pruning are difficult to train from the start, which would similarly improve training performance.

We find that a standard pruning technique naturally uncovers subnetworks whose initializations made them capable of training effectively. Based on these results, we articulate the "lottery ticket hypothesis:" dense, randomly-initialized, feed-forward networks contain subnetworks ("winning tickets") that - when trained in isolation - reach test accuracy comparable to the original network in a similar number of iterations. The winning tickets we find have won the initialization lottery: their connections have initial weights that make training particularly effective.
We present an algorithm to identify winning tickets and a series of experiments that support the lottery ticket hypothesis and the importance of these fortuitous initializations. We consistently find winning tickets that are less than 10-20% of the size of several fully-connected and convolutional feed-forward architectures for MNIST and CIFAR10. Above this size, the winning tickets that we find learn faster than the original network and reach higher test accuracy.

Summary in MIT Tech Review.

Ordered Neurons: Integrating Tree Structures into Recurrent Neural Networks (arXiv)
Yikang Shen · Shawn Tan · Alessandro Sordoni · Aaron Courville

Abstract - Natural language is hierarchically structured: smaller units (e.g., phrases) are nested within larger units (e.g., clauses). When a larger constituent ends, all of the smaller constituents that are nested within it must also be closed. While the standard LSTM architecture allows different neurons to track information at different time scales, it does not have an explicit bias towards modeling a hierarchy of constituents. This paper proposes to add such an inductive bias by ordering the neurons; a vector of master input and forget gates ensures that when a given neuron is updated, all the neurons that follow it in the ordering are also updated. Our novel recurrent architecture, ordered neurons LSTM (ON-LSTM), achieves good performance on four different tasks: language modeling, unsupervised parsing, targeted syntactic evaluation, and logical inference.

And a summary tweet from Microsoft with an accessible blog post.

Increasing diversity

Online presentations

38 presentations can be watched here.

As well as debates.

And here are a couple researchers putting their slides online.

Other summaries and highlights from the conference

AI for social good

Also, what is going on here?

Looks like a good PR move for their paper on the wizard of Wikipedia.

That's a wrap! See you all next year!

comments powered by Disqus