Deep learning in mobile and wireless networking: A survey
The rapid uptake of mobile devices and the rising popularity of mobile applications and
services pose unprecedented demands on mobile and wireless networking infrastructure …
services pose unprecedented demands on mobile and wireless networking infrastructure …
Deep learning on mobile and embedded devices: State-of-the-art, challenges, and future directions
Recent years have witnessed an exponential increase in the use of mobile and embedded
devices. With the great success of deep learning in many fields, there is an emerging trend …
devices. With the great success of deep learning in many fields, there is an emerging trend …
Streaming end-to-end speech recognition for mobile devices
End-to-end (E2E) models, which directly predict output character sequences given input
speech, are good candidates for on-device speech recognition. E2E models, however …
speech, are good candidates for on-device speech recognition. E2E models, however …
Hello edge: Keyword spotting on microcontrollers
Keyword spotting (KWS) is a critical component for enabling speech based user interactions
on smart devices. It requires real-time response and high accuracy for good user …
on smart devices. It requires real-time response and high accuracy for good user …
Large-scale visual speech recognition
This work presents a scalable solution to open-vocabulary visual speech recognition. To
achieve this, we constructed the largest existing visual speech recognition dataset …
achieve this, we constructed the largest existing visual speech recognition dataset …
[PDF][PDF] Shallow-Fusion End-to-End Contextual Biasing.
Contextual biasing to a specific domain, including a user's song names, app names and
contact names, is an important component of any production-level automatic speech …
contact names, is an important component of any production-level automatic speech …
Two-pass end-to-end speech recognition
The requirements for many applications of state-of-the-art speech recognition systems
include not only low word error rate (WER) but also low latency. Specifically, for many use …
include not only low word error rate (WER) but also low latency. Specifically, for many use …
Deep context: end-to-end contextual speech recognition
In automatic speech recognition (ASR) what a user says depends on the particular context
she is in. Typically, this context is represented as a set of word n-grams. In this work, we …
she is in. Typically, this context is represented as a set of word n-grams. In this work, we …
Federated evaluation and tuning for on-device personalization: System design & applications
We describe the design of our federated task processing system. Originally, the system was
created to support two specific federated tasks: evaluation and tuning of on-device ML …
created to support two specific federated tasks: evaluation and tuning of on-device ML …
Towards individuated reading experiences: Different fonts increase reading speed for different individuals
In our age of ubiquitous digital displays, adults often read in short, opportunistic interludes.
In this context of Interlude Reading, we consider if manipulating font choice can improve …
In this context of Interlude Reading, we consider if manipulating font choice can improve …