Abstract: This article summarizes some of the features of TensorFlow presented at the Google Cloud Next conference in San Francisco.
1. It’s a powerful machine learning framework
TensorFlow is a machine learning framework based on data flow graph. It is the second generation of Machine learning system of Google Brain, which is often used in various deep learning fields such as perception, language understanding, speech recognition, image recognition and so on. Tensor means n-dimensional arrays, Flow means calculations based on data Flow diagrams.
If you have enough data, and you’re in the deep learning, neural networks, advanced ARTIFICIAL intelligence phase of ARTIFICIAL intelligence, then it might be your best friend. TensorFlow is not a tool, it’s a framework, and if you want to return a regression line through a 20×2 spreadsheet, now you can stop learning and start using it.
TensorFlow is already being used in space to find new planets, prevent blindness by helping doctors screen for diabetic retinopathy, and help save forests by warning of illegal deforestation. AlphaGo and Google Cloud Vision are built on top of TensorFlow, which is what you need to pay attention to. In addition, TensorFlow is open source and can be downloaded for free and used at any time.
Kepler-90i, discovered with the help of TensorFlow, makes Kepler-90 the only known extrasolar system with eight planets orbiting it. No other system has more than eight planets.
- TensorFlow download address: www.tensorflow.org/install/
- TensorFlow initial tutorial: www.datacamp.com/community/t…
2. Methods are optional
If you’ve tried TensorFlow before, you’ve been scared off. It forces you to act more like an academic researcher than a developer, but now there are more options, so get back to it.
TensorFlow eager Execution allows you to interact with the system like a Python programmer: all the real-time coding and debugging is done line by line, rather than the daunting task of writing large chunks of program code in other languages. I’m an academic, but I’ve enjoyed TensorFlow eager execution since the beginning, so I’ll start using it as soon as possible.
3. Support the construction of neural network line by line
Keras + TensorFlow = Convenient construction of neural network
Keras is a deep learning library based on TensorFlow. Written in pure Python, Keras is user-friendly and provides easy and fast prototyping, which is more helpful for some lower versions of TensorFlow. If you like object-oriented thinking and prefer to build one layer of neural networks at a time, you’ll love TensorFlow.keras. In the next few lines of code, we create a coherent neural network with standard bells and whistles that appear to be off-key.
4. It’s not just Python
You’ve probably been complaining about TensorFlow’s obsession with Python by now. Now, TensorFlow is no longer just for Python developers. It now runs in a variety of languages, from R to Swift to JavaScript, as shown below:
You can do everything in the browser
Speaking of JavaScript, you can train and execute models in a browser using tensorflow.js.
Using tensorflow.js to perform real-time human pose assessment in the browser. Open your camera and look at this example.
6. Give micro devices a simple version
TensorFlow Lite enables models to be executed on a variety of devices, on a single or multiple cpus or Gpus on a PC or server, and even on mobile devices and the Internet of Things (IoT). TensorFlow gives you three times more performance. It supports threading, queuing, and asynchronous computation. It makes the best use of available hardware. Now you can start machine learning on your Raspberry Pi computer or mobile phone. In his talk at the conference, Lawrence did the brave thing of sorting images on an Android emulator in front of thousands of people, and it worked.
7. Professional equipment is better
If you’re tired of waiting for the CPU to finish training the neural network with the data it provides, you can now use Cloud TPUs to provide hardware specifically designed for the job. Just a few weeks ago, Google released the 3rd generation TPU (Tensor Processing Unit) on Alpha, an ASIC (integrated circuit chip technology) tailored specifically for machine learning and TensorFlow. TPU is a programmable ai accelerator that provides high throughput for low-precision calculations (e.g. 8-bit) oriented towards use or run models rather than training models.
8. The new data pipeline improves significantly
Are you still using Numpy for data piping? If you want to use TensorFlow, the current TF.Data Namespace allows you to be more expressive with TensorFlow input processing. Tf.data can provide you with a fast, flexible, easy-to-use data pipeline, while also providing synchronization training.
9. You don’t have to start from scratch
Everyone knows that machine learning is not a fun way to start from scratch. When I open the editor, there is only a blank new page and no example code. In this case, you can use the TensorFlow Hub to continue the age-old habit of using other people’s code to help you write your own code and call it your own.
9 Things You Should Know About TensorFlow
Follow public accounts
【 Pegasus Club 】
▼
Weixin.qq.com/r/bThZQajE7… (Qr code automatic recognition)
Pegasus will
AI artificial intelligence/big data/technology management and other personnel learning exchange park
Previous welfare concerns about the pegasus public number, reply to the corresponding keywords package download learning materials; Reply “join the group”, join the Pegasus AI, big data, project manager learning group, and grow together with excellent people!
From beginning to research, the 10 most Readable books in the field of artificial intelligence
RSVP number “2” machine learning & Data Science must-read classic book with resource pack!
Into AI & ML: Learning machine Learning from Basic Statistics (PDF download)
Answer the number “4” to learn about ARTIFICIAL intelligence, 30 books should not be missed (with electronic PDF download)
Answer number “6” AI AI: 54 Industry Blockbuster Reports
Reply number “12”
Small white | Python + + machine learning Matlab neural network theory + practice + + + depth video + courseware + source code, download attached!
Reply number “14” small white | machine learning and deep learning required books + machine learning field video/PPT + large data analysis books recommend!
526 Industry reports + White papers: AI, Artificial intelligence, robotics, smart mobility, smart home, Internet of Things, VR/AR, blockchain, etc. (download)
Reply number “19” 800G ARTIFICIAL intelligence learning materials :AI ebook +Python language introduction + tutorial + machine learning and other limited time free access!
Machine learning: How to go from beginner to Never Giving up? (With benefits)
Respond to digital “24” flash download | 132 g programming data: Python, JAVA, C, C + +, robot programming, PLC, entry to the proficient in ~
Reply number “25” limited resources | 177 g Python/machine learning/TensorFlow video/deep learning algorithm, introduction to cover/intermediate/project each stage!
Reply number “26” introduction to artificial intelligence book list recommended, learn AI please collect well (attached PDF download)
Reply | digital “27” Wu En of Stanford CS230 deep learning course a full range of information release (download)
Reply digital “30” receive | 100 + artificial intelligence study, deep learning, machine learning, big data, algorithms such as data, decisive collection!
FMI Artificial Intelligence and Big Data Summit Guest Speech PPT
Top 10 AI Jianghu Fields
Machine Learning Practical Experience Guide
More than 100 Papers on deep Learning
Top ten Classic Algorithms of Data Mining
6.10 Ele. me & Pegasus Project Management Practice PPT