Learning Safe Policies with Expert Guidance

We propose a framework for ensuring safe behavior of a reinforcement learning agent when the reward function may be difficult to specify. In order to do this, we rely on the existence of demonstrations from expert policies, and we provide a theoretical framework for the agent to optimize in the space of rewards consistent with its existing knowledge. We propose two methods to solve the resulting optimization: an exact ellipsoid-based method and a method in the spirit of the "follow-the-perturbed-leader" algorithm. Our experiments demonstrate the behavior of our algorithm in both discrete and continuous problems. The trained agent safely avoids states with potential negative effects while imitating the behavior of the expert in the other states.

Automatic 3D liver location and segmentation via convolutional neural networks and graph cut

Purpose Segmentation of the liver from abdominal computed tomography (CT) image is an essential step in some computer assisted clinical interventions, such as surgery planning for living donor liver transplant (LDLT), radiotherapy and volume measurement. In this work, we develop a deep learning algorithm with graph cut refinement to automatically segment liver in CT scans. Methods The proposed method consists of two main steps: (i) simultaneously liver detection and probabilistic segmentation using 3D convolutional neural networks (CNNs); (ii) accuracy refinement of initial segmentation with graph cut and the previously learned probability map. Results The proposed approach was validated on forty CT volumes taken from two public databases MICCAI-Sliver07 and 3Dircadb. For the MICCAI-Sliver07 test set, the calculated mean ratios of volumetric overlap error (VOE), relative volume difference (RVD), average symmetric surface distance (ASD), root mean square symmetric surface distance (RMSD) and maximum symmetric surface distance (MSD) are 5.9%, 2.7%, 0.91%, 1.88 mm, and 18.94 mm, respectively. In the case of 20 3Dircadb data, the calculated mean ratios of VOE, RVD, ASD, RMSD and MSD are 9.36%, 0.97%, 1.89%, 4.15 mm and 33.14 mm, respectively. Conclusion The proposed method is fully automatic without any user interaction. Quantitative results reveal that the proposed approach is efficient and accurate for hepatic volume estimation in a clinical setup. The high correlation between the automatic and manual references shows that the proposed method can be good enough to replace the time-consuming and non-reproducible manual segmentation method.

Flip-Rotate-Pooling Convolution and Split Dropout on Convolution Neural Networks for Image Classification

This paper presents a new version of Dropout called Split Dropout (sDropout) and rotational convolution techniques to improve CNNs' performance on image classification. The widely used standard Dropout has advantage of preventing deep neural networks from overfitting by randomly dropping units during training. Our sDropout randomly splits the data into two subsets and keeps both rather than discards one subset. We also introduce two rotational convolution techniques, i.e. rotate-pooling convolution (RPC) and flip-rotate-pooling convolution (FRPC) to boost CNNs' performance on the robustness for rotation transformation. These two techniques encode rotation invariance into the network without adding extra parameters. Experimental evaluations on ImageNet2012 classification task demonstrate that sDropout not only enhances the performance but also converges faster. Additionally, RPC and FRPC make CNNs more robust for rotation transformations. Overall, FRPC together with sDropout bring $1.18\%$ (model of Zeiler and Fergus~\cite{zeiler2013visualizing}, 10-view, top-1) accuracy increase in ImageNet 2012 classification task compared to the original network.