CATP: Cross-Attention Token Pruning for Accuracy Preserved Multimodal Model Inference.

Ruqi Liao Chuqing Zhao Jin Li Weiqi Feng

Published in: CoRR (2024)

Keyphrases

probabilistic model
bayesian model
computational cost
high accuracy
prior information
multi modal
computational model
mathematical model
theoretical analysis
inference engine
simulation model
statistical model
theoretical framework
neural network
input data
cost function
support vector
genetic algorithm