Alibaba Group Holding Limited

Minghai Qin

Yen-Kuang Chen

Qinggang Zhou

Yuanwei Fang

Shaolin Xie

Tae Meon Bae

Yen-kuang Chen

Yuan Xie

Tianchan Guan

Yuan Gao

Chunsheng Liu

Zhibin Xiao

Xiaoxin Fan

Chao Cheng

Guanlin Wu

Sicheng Li

This application describes methods, systems, and apparatus, including computer programs encoded on computer storage media, for microarchitecture-aware program sampling. An exemplary method includes receiving one or more traces collected from one or more microarchitectures executing a computer program for evaluating hardware configurations; training a machine learning (ML) model with multi-task learning based on the one or more traces as one or more training tasks; generating a plurality of embedded vectors representing the computer program; and updating, based on the trained ML model, the plurality of embedded vectors.

Method and system for microarchitecture-aware program sampling

Disclosed are systems and methods that determine whether instances of data (e.g., forward activations, backward derivatives of activations) that are used to train deep neural networks are to be stored on-chip or off-chip. The disclosed systems and methods are also used to prune the data (discard or delete selected instances of data). A system includes a hierarchical arrangement of on-chip and off-chip memories, and also includes a hierarchical arrangement of data selector devices that are used to decide whether to discard data and where in the system the data is to be discarded.

Hierarchical methods and systems for storing data

Methods and systems are provided for implementing training of learning models, including obtaining a pre-trained weight set for a learning model on a sample dataset and on a first loss function; selecting at least two tasks having heterogeneous features to be computed by a reference model; obtaining a reference dataset for the at least two tasks; designating a second loss function for feature embedding between the heterogeneous features of the at least two tasks; training the learning model on the first loss function and training the reference model on the second loss function, in turn; and updating the weight set based on a feature embedding learned by the learning model and a feature embedding learned by the reference model, in turn. Methods and systems of the present disclosure may alleviate computational overhead incurred by executing the learning model and loading different weight sets at a central network of the model.

Multi-task segmented learning models

The present disclosure relates to a method for scheduling computation resources for generating feature maps for video. The method comprises determining runtime for generating feature maps of a reference picture and a predicted picture, determining available computation resources for generating the feature maps, and allocating, based on the runtime, one or more computation resources among the available computation resources for generating the feature maps such that the feature maps are generated at regular time intervals.

Intelligent computing resources allocation for feature network based on feature propagation

A method and apparatus for characteristic-based video processing include: in response to receiving a region of a picture of a video sequence, determining a characteristic in the region, the region being independent of other regions of the picture for video coding; determining a class associated with the region based on the characteristic, the class being selected from a plurality of classes; and encoding the region using a parameter set associated with the class, the parameter set being selected from a plurality of parameter sets for video coding at different quality levels.

Method and system for characteristic-based video processing

Methods and apparatuses for video processing based on spatial or temporal importance include: in response to receiving picture data of a picture of a video sequence, determining a level of semantic importance for the picture data, the picture data including a portion of the picture; and applying to the picture data a first resolution-enhancement technique associated with the level of semantic importance for increasing resolution of the picture data, wherein the first resolution-enhancement technique is selected from a set of resolution-enhancement techniques having different computational complexity levels.

Growing community of inventors

San Mateo, CA, United States of America

Minghai Qin