Abstract: A learning system deploys, to one or more client devices, modules to be deployed in a learning environment of a respective client node. The learning environment of a respective client node may include modules for the client device (or client node) to collaborate with the central learning system and other client nodes via a distributed learning (e.g., federated learning, split learning) framework. In one embodiment, the learning system deploys an interoperable distributed learning environment for training a neural network encoder which can be used in heterogenous datasets to transform the heterogenous datasets across different institutions or entities into a common latent feature space. After training, the learning system receives data instances including a set of features and labels from different client nodes and trains a task neural network model configured to receive features in the latent space and generate an estimated label from the received data instances.

Abstract: The present invention relates to a quantum multi-agent meta reinforcement learning apparatus, which receives at least one observation value from different single-hop offloading environments, and the apparatus includes: a state encoding unit for calculating an angle along each axis by encoding the at least one observation value, and converting the angle along each axis into a quantum state; a quantum circuit unit for learning the angle along each axis, and overlapping the learned base layer using a controlled X (CX) gate; and a measurement unit for learning the overlapped base layer and measuring an axis parameter. Through the apparatus, the non-stationarity characteristic and credit-assignment problem of the conventional multi-agent reinforcement learning can be solved.

Abstract: Provided is a surveillance system employing a plurality of unmanned aerial vehicles (UAVs), the surveillance system showing improved surveillance performance while optimizing common energy consumption for computing of all the UAVs and also providing a stable visual monitoring service using autonomous mobility of the plurality of UAVs regardless of movement of an object to be monitored and action uncertainty of an adjacent UAV.

Abstract: A processor-implemented method with virtual object rendering includes: determining a plurality of predictive trajectories of a first object according to a Gaussian random path based on a high-level model that is trained by hierarchical reinforcement learning; determining direction information of a second object according to subgoals corresponding to the predictive trajectories based on a low-level model that is trained by hierarchical reinforcement learning; determining direction information of the second object according to a subgoal corresponding to one of the predictive trajectories based on an actual trajectory of the first object; and rendering the second object, which is a virtual object, based on the determined direction information.

Abstract: Provided are a drone taxi system based on multi-agent reinforcement learning and a drone taxi operation method using the same. The drone taxi system includes a plurality of drone taxies configured to receive call information including departure point information and destination information from passenger terminals present within a certain range and a control server configured to receive call information of passengers from each drone taxi, select a candidate passenger depending on whether a passenger is present, generate travel route information of each drone taxi from drone state information of the plurality of drone taxies through multi-agent reinforcement learning, and transmit the travel route information to the drone taxi.

Abstract: An object recognition method using queue-based model selection and optical flow in an autonomous driving environment includes preprocessing data through a dense flow in a matrix form by calculating an optical flow of images captured consecutively in time by a sensor for an autonomous vehicle, generating a confidence mask by generating a vectorized confidence threshold representing a probability that there is a moving object for each cell of the preprocessed matrix, determining whether there is a moving object on the images by mapping the images captured consecutively in time to the confidence mask, and selecting an object recognition model using a tradeoff constant between object recognition accuracy and queue stability in each time unit. Accordingly, it is possible to improve the performance of object recognition in an autonomous driving environment by applying the optical flow to the confidence threshold of the object recognition system.