In various examples, machine learning of encoding parameter values for a network is performed using a video encoder. Feedback associated with streaming video encoded by a video encoder over a network may be applied to an MLM(s). Using such feedback, the MLM(s) may predict a value(s) of an encoding parameter(s). The video encoder may then use the value to encode subsequent video data for the streaming. By using the video encoder in training, the MLM(s) may learn based on actual encoded parameter values of the video encoder. The MLM(s) may be trained via reinforcement learning based on video encoded by the video encoder. A rewards metric(s) may be used to train the MLM(s) using data generated or applied to the physical network in which the MLM(s) is to be deployed and/or a simulation thereof. Penalty metric(s) (e.g., the quantity of dropped frames) may also be used to train the MLM(s).

In various examples, a deep neural network(s) (e.g., a convolutional neural network) may be trained to detect moving and stationary obstacles from RADAR data of a three dimensional (3D) space. In some embodiments, ground truth training data for the neural network(s) may be generated from LIDAR data. More specifically, a scene may be observed with RADAR and LIDAR sensors to collect RADAR data and LIDAR data for a particular time slice. The RADAR data may be used for input training data, and the LIDAR data associated with the same or closest time slice as the RADAR data may be annotated with ground truth labels identifying objects to be detected. The LIDAR labels may be propagated to the RADAR data, and LIDAR labels containing less than some threshold number of RADAR detections may be omitted. The (remaining) LIDAR labels may be used to generate ground truth data.

A sensor transformation attention network (STAN) model including sensors configured to collect input signals, attention modules configured to calculate attention scores of feature vectors corresponding to the input signals, a merge module configured to calculate attention values of the attention scores, and generate a merged transformation vector based on the attention values and the feature vectors, and a task-specific module configured to classify the merged transformation vector is provided.

A method of using a computing device to self-train a machine learning model with an incomplete dataset including original observational data. The method includes receiving a labeled training data, the labeled training data for training a machine learning model. Counterfactual unlabeled training data is received. One or more labels are predicted for the counterfactual unlabeled training data. The machine learning model is trained based upon the labeled training data, the counterfactual unlabeled training data, and the predicted one or more labels for the unlabeled training data. The machine learning model reduces bias in original observational data. An evaluation of the predicted one or more labels is received based on corresponding artificial intelligence explanations provided by an artificial intelligence explainability model.

Various embodiments are generally directed to techniques for prediction based machine learning (ML) models, such as to utilize a ML model to generate predictions based on the output of another ML model. Some embodiments are particularly directed to a secondary ML model that revises predictions generated by a primary ML model based on structured input data. In many embodiments, the secondary ML model may utilize predictions from the primary ML model to learn metadata regarding the structured input data. In many such embodiments, the metadata regarding the structured input data may be used to revise the predictions from the primary ML model. For example, the secondary ML model may utilize a structure of the input data combined with patterns in the predictions from the primary ML model to revise the predictions from the primary ML model.

A character recognition method includes inputting an input image of a document, with the input image including a plurality of characters; selecting the plurality of characters through an object detection module to form at least one character region; separating the plurality of characters in the at least one character region to form a plurality of character boxes; performing calculation to determine a format of a character in each of the plurality of character boxes; recognizing the characters in the at least one character region through an object recognition module to determine a symbol content of the character in each of the plurality of character boxes; and converting the plurality of characters according to the format and symbol content of the character in each of the plurality of character boxes, and outputting corresponding editable characters.

Techniques for predicting punctuation and casing using multimodal fusion are described. An exemplary method includes processing generated text by: tokenizing the generated text into sub-words, and generating a sequence of lexical features for the sub-words using a pre-trained lexical encoder; processing audio of the audio by: generating a sequence of frame level acoustic embeddings using a pre-trained acoustic encoder on the audio, and generating task specific embeddings from the frame level acoustic embeddings; performing multimodal fusion of the sub-word level acoustic embeddings and the sequence of lexical features by: aligning the task specific embeddings to the sequence of lexical features, and combining the sequence of lexical features and aligned acoustic sequence; predicting punctuation and casing from the combined sequence of lexical features and aligned acoustic sequence; concatenating the sub-words of the text, and applying the predicted punctuation and casing; and outputting text having the predicted punctuation and casing.

Embodiments for managing real-time alerts using machine learning are disclosed. For example, a method includes receiving real-time data for one or more parameters of a device for which an alert is to be generated, from one or more sources associated with the device, and selecting a first machine learning model from a plurality of machine learning models based on the received real-time data. The method further includes determining at least one anomaly in the device based on the selected first machine learning model and predicting an impact of the determined at least one anomaly based on a second machine learning model of the plurality of machine learning models. Furthermore, the method includes generating the alert for the device in real-time based on the predicted impact of the determined at least one anomaly and receiving feedback on the generated alert in real-time.

The present disclosure is directed to a computer-implemented method and system for anatomical tree structure analysis. The method includes receiving model inputs for a set of positions in an anatomical tree structure. The method further includes applying, by a processor, a learning network to the model inputs. The learning network comprises a set of encoders and a neural network modeling the anatomical tree structure, wherein each encoder provides features extracted from the model input at a corresponding position. The neural network has a plurality of nodes constructed according to the anatomical tree structure and each node is configured to process the extracted features from one or more of the encoders. The method additionally includes providing an output of the learning network as an analysis result of the anatomical tree structure analysis.

A technology trend prediction method and system are provided. The method comprises acquiring paper data, and further comprises following steps: processing the paper data to generate a candidate technology lexicon; screening the candidate technology lexicon based on mutual information; calculating an independent word forming probability of an OOV word; extracting missed words in a title using a bidirectional long short-term memory network and a conditional random field (BI-LSTM+CRF) model; predicting a technology trend. The technology trend prediction method and system provided analyzes relationship of technology changes in a high-dimensional space, and predicts a development of technology trend based on time by extracting technical features of papers through natural language processing and time sequence algorithms.