Big data and the future of the self-driving car

Each year, car manufacturers get closer to successfully developing a fully autonomous vehicle. Over the last several years, major tech companies have paired up with car manufacturers to develop the advanced technology that will one day allow the majority of vehicles on the road to be autonomous. Of the five levels of automation, companies like Ford and Tesla are hovering around level three, which offers several autonomous driving functions but still requires a person to be attentive behind the wheel.

However, car manufacturers are expected to release fully automatic vehicles to the public within the next decade. These vehicles are expected to have a large number of safety and environmental benefits. Self-driving technology has come a long way over the last few years, as the growth of big data in technology industries has helped provide car manufacturers with the programming data needed to get closer to fully automating cars. Big data is helping to install enough information and deep learning in autonomous cars to make them safer for all drivers.

History of self-driving cars

The first major automation in cars was cruise control, which was patented in 1950 and is used by most drivers to keep their speed steady during long drives nowadays. Most modern cars already have several automated functions, like proximity warnings and steering adjustment, which have been tried and tested, and proven to be valuable features for safe driving. These technologies use sensors to alert the driver when they are coming too close to something that may be out of the driver’s view or something that the driver may simply not have noticed.

The fewer functions drivers have to worry about and pay attention to, the more they’re able to focus on the road in front of them and stay alert to dangerous circumstances that could occur at any moment. Human error causes 90 percent of all crashes on the roads, which is one of the main reasons so many industries support the development of autonomous vehicles. However, even when a driver is completely attentive, circumstances that are out of their control could cause them to go off the road or crash into other vehicles. Car manufacturers are still working on the programming for autonomous driving in weather that is less than ideal.

Big data’s role in autonomous vehicle development

Although these technologies provided small steps toward automation, they remained milestones away from a fully automated vehicle. However, over the last decade, with the large range of advancements that have been made in technology and the newfound use of big data, tech companies have discovered the necessary programming for fully automating vehicles. Autonomous vehicles rely entirely on the data they receive through GPS, radar and sensor technology, and the information they process through cameras.

The information cars receive through these sources provides them with the data needed to make safe driving decisions. Although car manufacturers are still using stores of big data to work out the kinks of the thousands of scenarios an autonomous car could find itself in, it’s only a matter of time before self-driving cars transform the automotive industry by making up the majority of cars on the road. As the price of the advanced radars for these vehicles goes down, self-driving cars should become more accessible to the public, which will increase the safety of roads around the world.

Big data is changing industries worldwide, and deep learning is contributing to the progress towards fully autonomous vehicles. Although it will still be several decades before the mass adoption of self-driving cars, the change will slowly but surely come. In only a few decades, we’ll likely be living in a time where cars are a safer form of transportation, and accidents are tragedies that are few and far between.

Source: Insidebigdata

The ability to speed up the training for deep learning networks used for AI through chunking

At the International Conference on Learning Representations on May 6, IBM Research shared a look around how chunk-based accumulation can speed the training for deep learning networks used for artificial intelligence (AI). 

The company first shared the concept and its vast potential at last year’s NeurIPS conference, when it demonstrated the ability to train deep learning models with 8-bit precision while fully preserving model accuracy across all major AI data set categories: image, speech and text. The result? This technique could accelerate training time for deep neural networks by two to four times over today’s 16-bit systems.

In IBM Research’s new paper, titled 'Accumulation Bit-Width Scaling For Ultralow Precision Training of Deep Networks', researchers explain in greater depth exactly how the concept of chunk-based accumulation works to lower the precision of accumulation from 32-bits down to 16-bits. 'Chunking' takes the product and divides it into smaller groups of accumulation and then adds the result of each of these smaller groups together, leading to a significantly more accurate result than that of normal accumulation. This allows researchers to study new networks and improve the overall efficiency of deep learning hardware.

Although this approach was previously considered infeasible to further reduce precision for training, IBM expects this 8-bit training platform to become a widely adopted industry standard in the coming years.

Author: Daniel Gutierrez

Source: Insidebigdata