GPU and Deep learning conference

Today I attended the successful ‘GPU Technology Conference’ event in Tel-Aviv, which is the premier event for AI and Deep-learning in the Israel high-tech industry. More than 1,000 people from the top industry came to hear the latest innovations from a variety of companies, professional teams and the NVIDIA CEO - Jensen Huang about the future of GPU computing and the impact of artificial intelligence and deep learning in the world. The event divided into 3 parts:

• Sessions - covered a broad range of topics and technical depth meetings on various subjects- Cyber security, Self driving cars, Healthcare and more
• Hands-On labs –gain practical, hands-on experience with Neural networks, Deep learning and NVIDIA Drive PX2 card
• Exhibition - meeting leading companies and technology startups.

Some of the more interesting sessions:

• Jensen Huang, the Founder of NVIDIA talks about the artificial intelligence industry and the future of Deep learning. Jensen showed a couple of amazing VR demos and talked about the tremendous progress of GPU in the last year Including entry into new fields such as chemistry and autonomous cars which NVIDIA are working very hard to achieve, including the new dedicated cards- “Drive PX2” and “Pegasus” which were designated for the automotive industry.

• Ilya Bogomolny from UVeye described their ‘Vehicle inspection system’ which scans the underside of moving vehicles to identify threatening objects or unlawful substances, using AI running on accelerated GPU.

• Eli David from Deepinstinct Introduced deep learning-based cybersecurity application that learns to detect any type of cyber threat.
• A couple of ‘Life saving solutions’ startups demonstrated their products, starting from identification system for people who drown in the pool, a system for identifying railway faults and advanced systems for the security of autonomous vehicles, all based on Deep learning and GPU cards
•  The great team from General Motors brought the autonomous vehicle they are developing to demonstrate its capabilities.

Hands-on labs

I personally participated in 4 labs about Deep Learning and the NVIDIA GPU card. Some of the labs were very good, comparing approaches and frameworks as TensorFlow, Caffe, Torch and Theano.  The recommended Labs are:

• Object Detection - A Deep Learning hands-on lab which explores three approaches to identify specific features within an image, and understand the merits of each one. Each approach is measured in relation to three metrics: model training time, model accuracy and speed of detection during deployment.
• Neural Network Deployment with DIGITS and TensorRT - This lab compares different approaches for NN deployment, as well as various optimizations that can be made in the inference process

Feel free to review the labs material and python notebooks on this link

Some screens from the above labs:

If you've read so far, I'd be happy to offer you free codes (that I have left) for NVIDIA's deep learning online labs which you can take. (Each online lab costs ~ $30)

Send me a message if you are interested.

Cheers,

Shmulik Willinger