An apparatus for recognizing an object in an image includes a preprocessing module configured to receive an image including an object and to output a preprocessed image by performing image enhancement processing on the received image to improve a recognition rate of the object included in the received image; and an object recognition module configured to recognize the object included in the image by inputting the preprocessed image to an input layer of an artificial neural network for object recognition.

This machine-learning device is provided with: a detection unit which detects a loss of consistency with a lapse of time in a determination result for unit data, the determination result being output from a determination unit that generates a learning model to be used when performing prescribed determination for one or more pieces of the unit data that form time series data; and a selection unit which selects, on the basis of the result of detection by the detection unit, unit data to be used as teacher data when the determination unit updates the learning model, thereby efficiently raising the accuracy of the learning model when machine learning is performed on the basis of the time series data.

A system may be configured to dynamically update deployed machine learning models. In some aspects, the system may receive sampled video information, generate first object detection information based on a cloud model and the sampled video information, and generate second object detection information based on a first edge model and the sampled video information. Further, the system may select, based on the first object detection information and the second object detection information, a plurality of training images from the sampled video information, detect motion information corresponding to motion of one or more detected objects within the plurality of training images, generate a plurality of annotated images based at least in part on the first object detection information and the motion information, and generate a second edge model based upon training the first edge model using the plurality of annotated images.

The information processing device performs distillation learning of a student model using unknown data which a teacher model has not learned. The label distribution determination unit outputs an arbitrary label for the unknown data. The data generation unit outputs new generated data using an arbitrary label and unknown data as inputs. The distillation learning part performs distillation learning of the student model using the teacher model and using the generated data as an input.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium that validates labels associated with sensor measurements of a scene in an environment. One of the methods includes receiving data representing a sensor measurement of a scene in an environment generated by one or more sensors. The sensor measurement can be associated with one or more labels, and each label can identify a portion of the sensor measurement that has been classified as measuring an object in the environment. For each of the labels, a determination can be made as to whether the label satisfies each of the validation criteria. Each validation criterion can measure whether one or more characteristics of the label are consistent with one or more characteristics of real-world objects in the environment. In response to determining that a particular label of the one or more labels does not satisfy one or more of the validation criteria, a notification can be generated indicating that the particular label is not a valid label for any real-world object in the scene of the environment.

Various implementations disclosed herein include devices, systems, and methods that involve federated learning techniques that utilize locally-determined ground truth data that may be used in addition to, or in the alternative to, user-provided ground truth data. Some implementations provide an improved federated learning technique that creates ground truth data on the user device using a second prediction technique that differs from a first prediction technique/model that is being trained. The second prediction technique may be better but may be less suited for real time, general use than the first prediction technique.

Teacher and student models output inference results for training data. A loss calculation unit calculates a total loss using at least one of (1) a loss obtained by multiplying a difference between a true value and a student model output by a weight increasing as a confidence of the teacher model output is lower, (2) a loss obtained by multiplying a difference between the true value and the student model output by a weight increasing as a difference between the true value and the teacher model output is greater, and (3) a loss obtained by multiplying a difference between the teacher and student model outputs by weights increasing as the difference between the teacher and student model outputs is greater and increasing as the difference between the true value and the teacher model output is smaller. An update part updates parameters of the student model based on the total loss.

A data generation apparatus (2) has: an obtaining unit (21) that obtains real data (D_real); a fake data generating unit (22) that generates fake data (D_fake) that imitates the real data; and a mix data generating unit (23) that generates mix data (D_mix) by mixing the real data and the fake data at a desired mix ratio (a), the mix data generating unit changes the mix ratio that is used to generate a data element of the mix data based on a position of the data element in the mix data.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training a policy neural network. In one aspect, a method for training a policy neural network configured to receive a scene data input and to generate a policy output to be followed by a target agent comprises: maintaining a set of training data, the set of training data comprising (i) training scene inputs and (ii) respective target policy outputs; at each training iteration: generating additional training scene inputs; generating a respective target policy output for each additional training scene input using a trained expert policy neural network that has been trained to receive an expert scene data input comprising (i) data characterizing the current scene and (ii) data characterizing a future state of the target agent; updating the set of training data; and training the policy neural network on the updated set of training data.

In a creation method of a trained model, a reconstructed image (60) obtained by reconstructing three-dimensional X-ray image data (80) is generated. A projection image (61) is generated from a three-dimensional model of an image element (50) by a simulation. The projection image is superimposed on the reconstructed image to generate a superimposed image (67). A trained model (40) is created by performing machine learning using the superimposed image, and the reconstructed image or the projection image.