xv
preface
The most significant event in the recent history of technology is perhaps the explo-
sion in the power of neural networks since 2012. This was when the growth in labeled
datasets, increases in computation power, and innovations in algorithms came
together and reached a critical mass. Since then, deep neural networks have made
previously unachievable tasks achievable and boosted the accuracies in other tasks,
pushing them beyond academic research and into practical applications in domains
such as speech recognition, image labeling, generative models, and recommendation
systems, just to name a few.
It was against this backdrop that our team at Google Brain started developing
TensorFlow.js. When the project started, many regarded “deep learning in JavaScript”
as a novelty, perhaps a gimmick, fun for certain use cases, but not to be pursued with
seriousness. While Python already had several well-established and powerful frame-
works for deep learning, the JavaScript machine-learning landscape remained splin-
tered and incomplete. Of the handful of JavaScript libraries available back then, most
only supported deploying models pretrained in other languages (usually in Python).
For the few that supported building and training models from scratch, the scope of
supported model types was limited. Considering JavaScript’s popular status and its
ubiquity that straddles client and server sides, this was a strange situation.
TensorFlow.js is the first full-fledged industry-quality library for doing neural net-
works in JavaScript. The range of capabilities it provides spans multiple dimensions.
First, it supports a wide range of neural-networks layers, suitable for various data types
ranging from numeric to text, from audio to images. Second, it provides APIs for load-
ing pretrained models for inference, fine-tuning pretrained models, and building and
training models from scratch. Third, it provides both a high-level, Keras-like API for
practitioners who opt to use well-established layer types, and a low-level, TensorFlow-
like API for those who wish to implement more novel algorithms. Finally, it is designed
剩余9页未读，继续阅读
评论0