In this paper, a neural architecture search (NAS) framework is proposed for 3D medical image segmentation, to automatically optimize a neural architecture from a large design space. Our NAS framework searches the structure of each layer including neural connectivities and operation types in both of the encoder and decoder. Since optimizing over a large discrete architecture space is difficult due to high-resolution 3D medical images, a novel stochastic sampling algorithm based on a continuous relaxation is also proposed for scalable gradient based optimization. On the 3D medical image segmentation tasks with a benchmark dataset, an automatically designed architecture by the proposed NAS framework outperforms the human-designed 3D U-Net, and moreover this optimized architecture is well suited to be transferred for different tasks.

No abstract

No abstract

Abstract. Electronic Sports (e-Sports) is computer and video games played as competitive sports over the Internet and a local area network, and are provided to users through a TV relay broadcasting system and the Internet due to tremendous popularity of the e-Sports recently. The main drawback of the previous eSports relay is to broadcast the e-Sports selected by the provider without viewers' intention, and thus the viewers can not watch desired information. Accordingly, this paper proposes a message-based semi-interactive relay broadcasting system called e-Sports Live over the Internet. The proposed system captures all regions of playing a game, and transfers the regions to a client's viewer. The client can watch desired information by selecting a slice of all regions. However, transferring continually all regions to the client over the Internet leads to a high-traffic due to high-capacity of data. To reduce the traffic, the system transfers all messages occurred in the game such as movements of characters instead of transferring continually all areas, based on resources repetitively used in game such as the whole map and game characters, and the resources are repetitively transferred to the client's viewer in an initial stage. Consequently, the proposed system can reduce the traffic over the Internet by transferring only message, and can perform a semi-interaction by watching desire information, as the client's viewer receives all areas occurred in the game based on resources repetitively used in the game and messages. Now, our system can not apply to TV relay broadcasting, as it does not have buffers for resources, but may be utilized for a variety of fields of interactive TV.

No abstract