The present invention relates to a method, device, and robot for classifying an outlier during object recognition learning using artificial intelligence. The method or device for classifying an outlier during object recognition learning according to an embodiment of the present invention sets an inlier region and an outlier region through learning using unlabeled data and labeled data.

An apparatus and method for generating learning data for combustion optimization is provided. The apparatus includes a data pre-processor to collect raw data including currently measured real-time data for boiler combustion and previously measured past data for the boiler combustion, and to perform pre-processing for the collected raw data, and a data analyzer to derive learning data from the raw data by analyzing the raw data. An apparatus for combustion optimization includes a management layer to collect currently measured real-time data for boiler combustion, to determine whether to perform combustion optimization, and to determine whether to tune a combustion model and a combustion controller; a data layer to derive learning data from raw data; a model layer to generate the combustion model/controller through the learning data; and an optimal layer to calculate a target value for combustion optimization and to output a control signal according to the calculated target value.

Disclosed are systems and method for training an ensemble of machine learning models with a focus on feature engineering. For example, the training of the models encourages each machine learning model of the ensemble to rely on a different set of input features from the training data samples used to train the machine learning models of the ensemble. However, instead of telling each model explicitly which features to learn, in accordance with the disclosed implementations, ML models of the ensemble may be trained sequentially, with each new model trained to disregard input features learned by previously trained ML models of the ensemble and learn based on other features included in the training data samples.

Systems, methods and products for classifying images according to a visual concept where, in one embodiment, a system includes an object detector and a visual concept classifier, the object detector being configured to detect objects depicted in an image and generate a corresponding object data set identifying the objects and containing information associated with each of the objects, the visual concept classifier being configured to examine the object data set generated by the object detector, detect combinations of the information in the object data set that are high-precision indicators of the designated visual concept being contained in the image, generate a classification for the object data set with respect to the designated visual concept, and associate the classification with the image, wherein the classification identifies the image as either containing the designated visual concept or not containing the designated visual concept.

A processor-implemented method with image analysis includes: receiving a test image; generating a plurality of augmented images by augmenting the test image; determining classification prediction values for the augmented images using a classifier; determining a detection score based on the classification prediction values; and determining whether the test image corresponds to anomaly data based on the detection score and a threshold.

Refined searching based on detected object configurations is provided by training a machine learning model to identify non-naturally occurring object configurations, acquiring images of an initial search area based on scanning it using a camera-equipped autonomous aerial vehicle operating in accordance with an initial automated flight plan defining the initial search area, analyzing the acquired images using the trained machine learning model and identifying that an object configuration is a non-naturally occurring object configuration, then based on identifying the non-naturally occurring object configuration, refining the initial automated flight plan to obtain a modified automated flight plan defining a different search area as compared to the initial search area, and initiating autonomous aerial scanning of the different search area in accordance with the modified automated flight plan.

A system and method for detecting non-compliances using machine learning uses anomaly detection on an input dataset of unlabeled observations to produce output observations with corresponding probability scores of the output observations being anomalous. A portion of the output observations are labeled as being compliant observations based on the corresponding probability scores, which are added to a training dataset of compliant and non-compliant observations to derive an augmented dataset of compliant and non-compliant observations. The augmented dataset of compliant and non-compliant observations is then used to train a machine learning model for non-compliance detection.

Methods and computer program products for training a neural network perform multiple forms of data augmentation on sample waveforms of a training dataset that includes both normal and abnormal samples to generate normal data augmentation samples and abnormal data augmentation samples. The normal data augmentation samples are labeled according to a type of data augmentation that was performed on each respective normal data augmentation sample. The abnormal data augmentation samples are labeled according to a type of data augmentation other than that which was performed on each respective abnormal data augmentation sample. A neural network model is trained to identify a form of data augmentation that has been performed on a waveform using the normal data augmentation samples and the abnormal data augmentation samples.

A knowledge distillation method based on a regression task according to an embodiment of the present disclosure includes training a first deep neural network model that is a teacher model including a first encoder and a first decoder, constructing a second deep neural network model that is a student model including a second encoder and a second decoder according to a preset constraint of a target system and initializing parameters of the second deep neural network model, performing a first knowledge distillation operation on the second encoder of the second deep neural network model based on the trained first deep neural network model, and performing a second knowledge distillation operation on the second encoder and the second decoder of the second deep neural network model based on the trained first deep neural network model.

An approach for altering alter training data and training process associated with a neural network to emulate environmental noise and operational instrument error by using the concepts of shots to sample within a squeezed space model, wherein shots are an uncertainty index that is the average of all shots from a sampling, is disclosed. The approach leverages a squeeze theorem to create a squeezed space model based on the regression of the upper and lower bound associated with the environmental noise and instrument error. The approach calculates an average noise index based on the squeezed space model, wherein the index is used to alter the training data and process.