A fully convolutional spatio-temporal patch mixer, STLight, matches or beats RNN-based baselines on predictive learning benchmarks at a fraction of the parameter and FLOP budget.
Anticipating many futures: Online human motion predic- tion and generation for human-robot interaction
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
STLight: a Fully Convolutional Approach for Efficient Predictive Learning by Spatio-Temporal joint Processing
A fully convolutional spatio-temporal patch mixer, STLight, matches or beats RNN-based baselines on predictive learning benchmarks at a fraction of the parameter and FLOP budget.