AI Paper Atlas

Research map / Learning theory and algorithms / Kernels and regression theory

Kernel Multimodal Continuous Attention

Alexander Moreno, Zhenke Wu, Supriya Nagesh, Walter H. Dempsey, James Matthew Rehg

NeurIPS 2022 · Learning theory and algorithms › Kernels and regression theory · approach: Kernel regression under shift · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace Gaussian/unimodal continuous attention with kernel exponential family attentioncontinuous softmax attention · Synthetic time warpingaccuracy: 77.72 -> 85.71Improved
replace truncated-parabola continuous sparsemax with kernel deformed exponential attentioncontinuous sparsemax · Synthetic time warpingaccuracy: 77.96 -> 88.1Improved
use kernel sparsemax instead of continuous sparsemaxcontinuous sparsemax · FordAaccuracy: 90.95 -> 92.61Improved
use kernel softmax instead of Gaussian-based continuous softmaxcontinuous softmax · FordAaccuracy: 74.88 -> 92.44Improved
use kernel softmax instead of Gaussian mixture attentionGaussian mixture · FordAaccuracy: 69.56 -> 92.44Improved
kernel attention compared with LSTM FCNkernel sparsemax · FordAaccuracy: 93.41 vs 92.61Hurt
use kernel softmax instead of multi-head unimodal continuous softmax (100 heads)continuous softmax · uWave irregular samplingaccuracy: 67.78 -> 93.85Improved
use kernel softmax instead of Gaussian mixture attentionGaussian mixture · uWave irregular samplingaccuracy: 81.13 -> 94.26Improved
increase basis functions N from 64 to 128kernel softmax · uWave irregular samplingaccuracy: 93.85 -> 94.26No clear effect
increase basis functions N from 64 to 128kernel sparsemax · uWave irregular samplingaccuracy: 92.32 -> 92.15No clear effect
use kernel softmax instead of continuous softmaxcontinuous softmax · IMDBmean accuracy: 90.32 -> 90.94Improved
use kernel sparsemax instead of continuous sparsemaxcontinuous sparsemax · IMDBmean accuracy: 89.95 -> 90.39Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy