A method and apparatus for training a compound property prediction model, a device, a storage medium and a program product. A implementation of the method comprises: acquiring an unannotated compound data set; pre-training a graph neural network using the unannotated compound data set to obtain a pre-trained graph neural network; acquiring a plurality of annotated compound data sets, each annotated compound data set being annotated with one kind of compound property; and performing multi-task training on the pre-trained graph neural network using the plurality of annotated compound data sets, to obtain a compound property prediction model, the compound property prediction model being used to predict a plurality kinds of properties of a compound.

Machine-learning-aided automatic taxonomy for web data. In an embodiment, a training dataset of annotated features is used to train a model to predict a class in a taxonomy of web-based activities. The features may be derived from a uniform resource locator (URL) of an online resource and associated metadata. During operation, the features may be extracted from the URL and metadata of each activity record in web data. The trained model may be applied to the extracted features for each activity record to predict a class within the taxonomy. The predicted taxonomic class may be stored in association with the URL that was extracted from the activity record to produce a taxonomized URL.

Methods and systems for performing named entity recognition are disclosed. One method includes using a multi-task approach to fine-tune a neural network to perform named entity recognition. A multi-task objective function can include a combination of a triplet loss and a named entity recognition loss. The triplet loss can include the use of supplementary texts. The method further includes using the fine-tuned neural network to identify one or more named entities in a text. Aspects of the disclosure also include integrating named entity recognition with one or more other natural language processing tasks.

According to an embodiment, an information processing device includes processors. The processors receive input of a plurality of pieces of input data obtained during K time periods. K is an integer equal to or greater than two. The processors estimate K first models. Each of the K first models receives input of input data and outputs output data. Each of the K first models is estimated for each period of the K time periods, using a plurality of pieces of input data obtained during the each period. The processors estimate a second model that indicates a relationship between first time parameters related to times of the K time periods, and the K first models. The processors estimate a first model corresponding to a specified second time parameter, based on the estimated second model.

There is provided servers and methods of generating a waveform based on a spectrogram and a noise input. The method includes acquiring a trained flow-based vocoder including invertible blocks, and an untrained feed-forward vocoder including non-invertible blocks, which form a student-teacher network. The method includes executing a training process in the student-teacher network during which the server generates (i) a teacher waveform by the trained flow-based vocoder using a first spectrogram and a first noise input, (ii) a student waveform by the untrained feed-forward vocoder using the first spectrogram and the first noise input, and (iii) a loss value for the given training iteration using the teacher waveform and the student waveform. The server then trains the untrained feed-forward vocoder to generate the waveform. The trained feed-forward vocoder in then used lieu of the trained flow-based vocoder for generating waveforms based on spectrograms and noise inputs.

An apparatus comprises processing circuitry configured to receive a first model and a second model; determine difference information that is representative of a difference between the first model and the second model and/or between the first task and the second task and/or between the first domain and the second domain; and generate a third model using the first model, the second model and the difference information, wherein the generating of the third model comprises training the third model to perform both of the first task and the second task and/or to operate on both the first domain and the second domain.

Hybrid deep learning systems and methods allow for detecting anomalies in objects, such as electrical printed circuit board (PCB) components, based on image data. In one or more embodiments, a hybrid deep learning model comprises a Graph Attention Network (GAT) that uses spatial properties of the PCB components to extract latent semantic information and generate an output set of hidden representations. The GAT treats each of the electrical components as a node and each connection between them as edges in a graph. The hybrid system further comprises a Convolutional Neural Network (CNN) that uses pixel data to obtain its own output set of hidden representations. The hybrid deep learning model concatenates both sets to detect anomalies that may be present on the PCB.

An ear detection method with deep learning pairwise model based on contextual information belongs to the field of biometric recognition technologies, and addresses a problem that an ear location cannot be found in a large scene, especially in a background image containing a whole body. The method includes: performing preprocessing and object labeling on images; modifying an Oquab network to be a local model for four classes through transfer learning and training the local model; training two pairwise models of head and ear as well as body and head based on the local model; and performing joint detection for an ear through the local model, the two pairwise models and body features. The method uses a hierarchical relationship from large to small to establish contextual information, which can reduce the interference of other features and detect the location of the ear more accurately.

A hardware neural network system includes an input buffer for input neurons (Nbin), an output buffer for output neurons (Nbout), and a third buffer for synaptic weights (SB) connected to a Neural Functional Unit (NFU) and a control logic (CP) for performing synapses and neurons computations. The NFU pipelines a computation into stages, the stages including weight blocks (WB), an adder tree, and a non-linearity function.

The present disclosure describes a system, method, and computer program for predicting sentiment labels for audio speech utterances using an audio speech sentiment classifier pretrained with pseudo sentiment labels. A speech sentiment classifier for audio speech (“a speech sentiment classifier”) is pretrained in an unsupervised manner by leveraging a pseudo labeler previously trained to predict sentiments for text. Specifically, a text-trained pseudo labeler is used to autogenerate pseudo sentiment labels for the audio speech utterances using transcriptions of the utterances, and the speech sentiment classifier is trained to predict the pseudo sentiment labels given corresponding embeddings of the audio speech utterances. The speech sentiment classifier is then subsequently fine tuned using a sentiment-annotated dataset of audio speech utterances, which may be significantly smaller than the unannotated dataset used in the unsupervised pretraining phase.