In early August, Sydney, Australia will play host to the International Conference on Machine Learning (ICML) which brings together the best of minds and technologies in the field of artificial intelligence (AI). Since 2012, the conference has been held in locations across the globe including the UK, US, China, France and now its Australia’s turn.
AI and specifically Machine Learning and Deep Learning are creating massive disruptions to so many industries, transport, health, energy, and many, many more, so its a great opportunity to host the conference, bringing together the brightest minds from around the world. Its important Australian businesses are on top of the amazing possibilities afforded by AI, as understanding the latest developments in the field, ensures we remaining globally competitive.
The conference which runs from August 6th to the 11th and is expected to draw up to 3,000 participants, including faculty, researchers and PhD students in machine learning, data science, data mining, AI, statistics, and related fields.
There’s AI instruction sets have exploded in computational efficiency in recent years, largely helped by a transition to perform on the GPU. Naturally NVIDIA will be at the conference and is encouraging participation at ICML by providing scholarships for 52 students from around the world. If you’re attending, head to the NVIDIA booth (Level 2, The Gallery, Booth #4) where you’ll see the company showcase a range of exciting demos, many which are on display in Australia for the first time. If you had any doubt how GPUs can deliver reduce AI processes from years to just weeks and even days, these demos will clear that up.
NVIDIA demos at ICML include:
- 4K style transfer: A deep neural network to extract a specific artistic style from a source painting, and then synthesises this information with the content of a separate video.
- Self-driving auto: The Drive PX2 is an AI car computing platform that enables automakers and tier 1 suppliers to accelerate production of automated and
- AI at the edge: The NVIDIA Jetson TX2 is a credit card-sized platform that puts AI computing to work in devices such as drones, body cams and robots.
- Deepstream: The SDK simplifies development of high performance video analytics applications powered by deep learning.
- NVIDIA GPU Cloud: The GPU-accelerated platform runs everywhere to let data scientists and researchers rapidly build, train, and deploy neural network models to address some
of the most complicated AI challenges.
- NVIDIA Isaac: The AI-based software platform lets developers train virtual robots using detailed and highly realistic test scenarios.
- NVIDIA DGX Station: The world’s first personal supercomputer for leading-edge AI development enables researchers and data scientists to experiment at their desks and
extend their work across DGX Systems and the cloud.
NVIDIA’s Deep Learning Institute (DLI) are offering a one-day hands-on workshop on August 10 from 9am to 5pm with their Open Source library for Machine Learning, TensorFlow.
The video below is a graphic demonstration from researchers from University of California, Berkeley and Siemens that learns how to grip new objects just by studying a database of 3D shapes. It does this using a GTX 1080 GPU and cuDNN with the TensorFlow deep learning framework. To enable the robot to learn the unusual shapes and their possible combinations, the team generated a a mega 6.7 million synthetic point clouds from thousands of 3D models to train their convolutional neural network to recognize robust grasps. Once trained, they gave the robot new objects it’d never seen before and here’s the killer, it was successful at lifting the object 99% of the time. They taught the robot how to pick things up. That’s the power of AI.
NVIDIA is not just attending ICML to show off their wares, their using it as a recruiting opportunity. If you’re a developer and interested in that kind of opportunity, be at the NVIDIA booth from August 7 to 9, 3 to 4pm, and track down a manager to learn more about career opportunities.
Developers can also join the NVIDIA Developer Program to establish a working relationship with NVIDIA engineering, a pretty good career move. Being part of the program gives you access to an extensive range of NVIDIA software and technology updates. In addition, members can contribute to Parallel Forall, a GPU Computing developer blog focused on providing detailed technical information on a variety of massively parallel programming topics.
If you’re a fan of NVIDIA, chances are you’ve heard of the much heralded GeForce GTX USB Drive, be one of the first 50 people to arrive and you’ll score you’re very own. Recently NVIDIA gave 580 of these away to attendees of the PDXLAN conference, which as you can imagine, was met with a great response.
NVIDIA must be feeling generous lately, with news today of NVIDIA CEO Jensen Huang giving away the latest NVIDIA Tesla V100 GPU to 15 participants of their AI Labs Program. Volta, is the seventh-generation GPU architecture, providing a 5x improvement in peak teraflops over its predecessor Pascal. and 15x over the Maxwell architecture, which launched just 2 years ago. This demonstrates how quickly the hardware that powers AI is progressing, surpassing the 4x the improvements that Moore’s law would have predicted.
The Tesla V100 GPU accelerator shatters the 100 teraflops barrier of deep learning performance. The V100 features over 21 billion transistors (seriously that’s so hard to wrap your head around), includes 640 Tensor Cores, delivering 120 teraflops of deep learning performance. It’s all supported by Volta-optimized software, including CUDA, cuDNN and TensorRT, which frameworks and applications can easily tap into to accelerate AI and research.