Microsoft Technology Licensing, LLC

Xiaozhi Yu

Ji Li

Amit Srivastava

Youjun Liu

Han Hu

Adit Krishnan

Qi Dai

Yixuan Wei

Gregory Alexander DePaul

A data processing system implements a dynamic knowledge distillation process including dividing training data into a plurality of batches of samples and distilling a student model from a teacher model using an iterative knowledge distillation. The process includes instantiating an instance of the teacher model and the student model in a memory of the data processing system and obtaining a respective batch of training data from the plurality of batches of samples in the memory. The process includes training the teacher and student models using each of the samples in the respective batch of the training data, evaluating the performance of the student model compared with the performance of the teacher model, and providing feedback to student model to adjust the behavior of the student model based on the performance of the student model.

Scalable knowledge distillation techniques for machine learning

A textual user input is received and a plurality of different text-to-content models are run on the textual user input. A selection system attempts to identify a suggested content item, based upon the outputs of the text-to-content models. The selection system first attempts to generate a completed suggestion based on outputs from a single text-to-content model. It then attempts to mix the outputs of the text-to-content models to obtain a completed content suggestion.

Machine learning model-based content processing framework

Growing community of inventors

San Jose, CA, United States of America

Xiaozhi Yu