Qualcomm Incorporated

Google Inc.

Mei Chen

Quoc V Le

Pranav Khaitan

Cong Li

Jay K Adams

Manas Rajendra Joglekar

Brian Clarke Banister

Jong Hyeon Park

Zae Yong Choi

Subramanya Padubidri Nanyam Rao

Jinhua Jiang

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for determining, for each of one or more categorical features, a respective vocabulary of categorical feature values of the categorical feature that should be active during processing of inputs by a machine learning model. In one aspect, a method comprises: generating a batch of output sequences, each output sequence in the batch specifying, for each of the categorical features, a respective vocabulary of categorical feature values of the categorical feature that should be active; for each output sequence in the batch, determining a performance metric of the machine learning model on a machine learning task after the machine learning model has been trained to perform the machine learning task with only the respective vocabulary of categorical feature values of each categorical feature specified by the output sequence being active.

Learning to select vocabularies for categorical features

Methods, systems, and devices are described for mapping, by a UE, uplink transmissions to one or more antennas of the UE. In one aspect, a UE may receive an uplink resource allocation identifying a set of resource blocks (RBs) from a base station. The UE may map subsets of the set of RBs to one or more of a plurality of antennas of the UE and transmit one or more uplink transmissions, using the subsets of the set of resource blocks, on the one or more antennas based at least in part on the mapping. Additionally, in some aspects, the UE may determine channel characteristics or CSI associated with the downlink from the base station and base the mapping of subsets of RBs to one or more antennas of the UE on the determined channel characteristics/CSI.

Transmit antenna diversity scheme

Systems and methods for memory-efficient storage and extraction of maximum power reduction (MPR) values in two-carrier wireless data systems are presented. A wireless broadband device can operate under the HSUPA Category 9 standard, in which two carriers can be employed for data uplinks. Due to power saturation, interference, and other factors, transmission output power is limited to various levels depending on channel configuration. Under previous standards using one carrier, the maximum power reduction (MPR) needed to address those issues could be stored on the device, since the total number of MPR values was limited. With the introduction of dual carriers in HSUPA-9, storing all possible MPR values is no longer feasible. Platforms and techniques are disclosed which allow accurate generation of MPR values in a two-carrier system, utilizing the 2nd, 4th, and 6th moments of the complex signals to derive MPR values without attempting to store all possible carrier combinations.

Systems and methods for memory-efficient storage and extraction of maximum power reduction (MPR) values in two-carrier wireless data systems

Growing community of inventors

Sunnyvale, CA, United States of America

Mei Chen