Study On Deep Neural Network's Stability And Application In Video Object Detection

Recent approaches interpret deep neural works(DNNs)as dynamical systems,drawing the connection between stability in forward propagation and generalization of DNNs.In this paper,we take a step further to be the first to reinforce this stability of DNNs without changing their original struc-ture and verify the impact of the reinforced stability on the network represen-tation from various aspects.More specifically,we reinforce stability by mod-eling attractor dynamics of a DNN and propose relu-max attractor network(RMAN),a light-weight module readily to be deployed on state-of-the-art ResNet-like networks.RMAN is only needed during training so as to mod-ify a ResNet's attractor dynamics by minimizing an energy function together with the loss of the original learning task.Through intensive experiments,we show that RMAN-modified attractor dynamics bring a more structured representation space to ResNet and its variants,and more importantly im-prove the generalization ability of ResNet-like networks in supervised tasks due to reinforced stability.Besides studying the property of deep neural networks,this paper fo-cuses on exploring deep neural network in video object detection.Video object detection is more challenging than image object detection because of the deteriorated frame quality.To enhance the feature representation,state-of-the-art methods propagate temporal information into the deterio-rated frame by aligning and aggregating entire feature maps from multiple nearby frames.However,restricted by feature map' s low storage-efficiency and vulnerable content-address allocation,long-term temporal information is not fully stressed by these methods.In this work,we propose the first object guided external memory network for online video object detection.Storage-efficiency is handled by object guided hard-attention to selectively store valuable features,and long-term information is protected when stored in an addressable external data matrix.A set of read/write operations are de-signed to accurately propagate/allocate and delete multi-level memory fea-ture under object guidance.We evaluate our method on the ImageNet VID dataset and achieve state-of-the-art performance as well as good speed-ac-curacy tradeoff.Furthermore,by visualizing the external memory,we show the detailed object-level reasoning process across frames.