Description
FBNet is a type of convolutional neural architectures discovered through DNAS neural architecture search. It utilises a basic type of image model block inspired by MobileNetv2 that utilises depthwise convolutions and an inverted residual structure (see components).
Papers Using This Method
Multi-Predict: Few Shot Predictors For Efficient Neural Architecture Search2023-06-04FBNet: Feedback Network for Point Cloud Completion2022-10-08Learning to repair: Repairing model output errors after deployment using a dynamic memory of feedback2022-01-16Learning to Repair: Repairing model output errors after deployment using a dynamic memory of feedback2021-12-16FBNet: Feature Balance Network for Urban-Scene Segmentation2021-11-05HAO: Hardware-aware neural Architecture Optimization for Efficient Inference2021-04-26HW-NAS-Bench:Hardware-Aware Neural Architecture Search Benchmark2021-03-19HW-NAS-Bench: Hardware-Aware Neural Architecture Search Benchmark2021-01-01Parallax Motion Effect Generation Through Instance Segmentation And Depth Estimation2020-10-06Hit-Detector: Hierarchical Trinity Architecture Search for Object Detection2020-03-26Hardware Aware Neural Network Architectures using FbNet2019-06-17FBNet: Hardware-Aware Efficient ConvNet Design via Differentiable Neural Architecture Search2018-12-09