Downloaded 23 times
Uploaded byElifTech
Getting started with TensorFlow
TensorFlow is a popular machine learning library developed by Google. It allows highly scalable deep learning across many devices from data centers to phones. TensorFlow represents computations as data flow graphs with variables as tensors. This allows flexible and efficient training and deployment of models for tasks like image recognition, translation, and more.
In this document
Powered by AI
Overview of the presentation by Anatolii Shkurpylo and agenda coverage including what TensorFlow is and its applications.
TensorFlow is a leading ML library developed by Google, released in 2015. Key stats: 1000+ contributors, 72000 stars, 35500 forks.
TensorFlow's scalability allows training on various platforms including data centers and mobile devices.
The core of TensorFlow is in C++, with a Data Flow Graph representation of computations and variables as tensors.
TensorFlow is versatile for voice recognition, sentiment analysis, image recognition, and more across different industries.
TensorFlow is recommended for beginners in ML, with practical tools like tf.estimator and a demo presentation.
Links to resources for learning TensorFlow and thank you message from the presenter, inviting questions.
Getting started with TensorFlow
- 1. Getting started withTensorFlow by Anatolii Shkurpylo, Software Developer at ElifTech
- 2. www.eliftech.com Agenda 1. What isTenforflow? 2. How does it work? 3. Why would you use it? 4. Demo / Tensorboard
- 3.
- 4. www.eliftech.com TensorFlow - themost popular ML library in the World at present. ▪ Developed by Google Brain team for internal Google use. ▪ From November 2015 - released under open source license ▪ > 1000 contributors ▪ ~ 72000 stars ▪ > 35500 forks ▪ ~ 23000 commits
- 5.
- 6. www.eliftech.com Highly Scalable ▪ Trainingon ▪ Data Center ▪ CPU, GPU, TPU ▪ Running on ▪ Mobile phones ▪ IoT devices ▪ ...
- 7.
- 8. www.eliftech.com TensorFlow core Core inC++ ● very low overhead Different front ends ● Python, C++, Go, Java, Haskell, Rust
- 9. www.eliftech.com In TensorFlow allcomputation represented as Data Flow Graph
- 10. www.eliftech.com All variables representedas tensors Dataflow Graph (forward)
- 11.
- 12. www.eliftech.com Why would weuse that?
- 13.
- 14. www.eliftech.com TensorFlow can beused for solving a range of different problems ▪ Voice/Sound recognition ▪ Voice recognition (IoT, Automotive, Security, UX/UI) ▪ Voice Search (Telecoms, Handset) ▪ Sentiment Analysis (CRM) ▪ Flaw Detection (engine noise - Automotive, Aviation) ▪ Text Based Application ▪ Google translate ... ▪ Image Recognition ▪ Time Series ▪ Video Detection
- 15. www.eliftech.com TensorFlow is agood choice to start with ML
- 16.
- 17.
- 18.
- 19.
- 20. www.eliftech.com Links ▪ http://playground.tensorflow.org ▪ https://www.tensorflow.org/get_started ▪https://www.oreilly.com/learning/hello-tensorflow ▪ https://youtu.be/u4alGiomYP4 - I/O 2017 ▪ http://edwardlib.org/tutorials/tensorboard ▪ https://goo.gl/27vWqV
- 21. www.eliftech.com Thank you forattention! Find us at eliftech.com Have a question? Contact us: info@eliftech.com