AI at F8 2018: Open frameworks and responsible development

Jerome Pesenti

Artificial intelligence has the potential to transform lives, creating new digital experiences while safeguarding our existing online interactions. Balancing rapid and responsible progress requires…

Overview

The article discusses Facebook's advancements in artificial intelligence (AI) showcased at F8 2018, emphasizing open development frameworks like PyTorch 1.0 and ONNX. It highlights the company's commitment to responsible AI development through collaboration with the AI community and ethical considerations.

What You'll Learn

1

How to utilize PyTorch 1.0 for rapid experimentation and deployment of AI models

2

Why ONNX facilitates easier model transfer between AI frameworks

3

How to implement state-of-the-art video understanding with ResNext3D

4

When to apply ethical considerations in AI development

Prerequisites & Requirements

Understanding of AI frameworks and model deployment
Familiarity with PyTorch and ONNX(optional)

Key Questions Answered

What is the significance of PyTorch 1.0 in AI development?

PyTorch 1.0 allows developers to experiment rapidly in a flexible execution mode and transition seamlessly to a graph-based mode for deployment. This flexibility enhances the speed and efficiency of taking AI from research to production.

How does ONNX improve model interoperability?

ONNX, or Open Neural Network Exchange, allows AI engineers to move models between different frameworks without the need for extensive custom engineering. This open format supports various tools, enhancing collaboration and efficiency in AI development.

What advancements were made in computer vision at F8 2018?

Facebook's AI researchers trained a new image recognition model on 3.5 billion photos, achieving 85.4% accuracy on the ImageNet benchmark. This showcases the potential of using large datasets and existing labels for improving AI training.

What ethical considerations does Facebook emphasize in AI development?

Facebook emphasizes the importance of addressing algorithmic bias and promoting diversity in the workforce to enhance the ethical use of AI. They employ internal tools to measure fairness and ensure responsible AI deployment.

Key Statistics & Figures

Image recognition accuracy

85.4%

Achieved by Facebook's new image recognition model trained on 3.5 billion photos.

Number of photos used for training

3.5 billion

Used to train the new image recognition model, showcasing the potential of large datasets.

Technologies & Tools

Some links below are affiliate links. We may earn a commission if you make a purchase.

Framework

Pytorch

Used for rapid experimentation and deployment of AI models.

Framework

Onnx

Facilitates model interoperability between different AI frameworks.

Key Actionable Insights

1
Leverage PyTorch 1.0 for both research and production to enhance your AI projects.
Using PyTorch 1.0 allows for rapid experimentation and efficient deployment, making it a valuable tool for developers looking to streamline their AI workflows.

2
Utilize ONNX to facilitate easier model transitions between different AI frameworks.
By adopting ONNX, developers can save time and resources when moving models across platforms, enhancing collaboration and flexibility in AI development.

3
Incorporate ethical considerations into your AI projects from the outset.
Addressing ethical concerns early in the development process can prevent biases and ensure that AI systems are designed to be fair and responsible.

Common Pitfalls

1

Neglecting ethical considerations in AI development can lead to biased outcomes.

This often occurs when diverse perspectives are not included in the planning and development stages, which can result in systems that do not serve all users fairly.

Related Concepts

Open Source AI Frameworks

Ethical AI Development

Computer Vision Advancements

Natural Language Understanding