58 citations · 255 across the 13 of their papers we have counts for
13 papers
Understanding How Encoder-Decoder Architectures Attend
Kyle Aitken, Vinay V Ramasesh, Yuan Cao +1
Encoder-decoder networks with attention have proven to be a powerful way to solve many sequence-to-sequence tasks. In these networks, attention aligns encoder and decoder states an…
The geometry of integration in text classification RNNs
Kyle Aitken, Vinay V. Ramasesh, Ankush Garg +3
Despite the widespread application of recurrent neural networks (RNNs) across a variety of tasks, a unified understanding of how RNNs solve these tasks remains elusive. In particul…
On the asymptotics of wide networks with polynomial activations
Kyle Aitken, Guy Gur-Ari
We consider an existing conjecture addressing the asymptotic behavior of neural networks in the large width limit. The results that follow from this conjecture include tight bounds…
New and Old Fermionic Dualities from 3d Bosonization
Kyle Aitken, Changha Choi, Andreas Karch
We construct novel fermion-fermion dualities in -dimensions using 3d bosonization dualities. This is achieved by relating two-node quiver theories using both the flavor-bounde…
Generalization of QCD Symmetry-Breaking and Flavored Quiver Dualities
Kyle Aitken, Andrew Baumgartner, Changha Choi +1
We extend the recently proposed symmetry breaking scenario of QCD to the so-called "master" d bosonization duality, which has bosonic and fermionic matter on both ends.…
Novel 3d bosonic dualities from bosonization and holography
Kyle Aitken, Andrew Baumgartner, Andreas Karch
We use 3d bosonization dualities to derive new non-supersymmetric dualities between bosonic quiver theories in dimensions. It is shown that such dualities are a natural non-A…