The Who’s Who of Deep Learning – Part Two

Previously, we featured the contributions of  three distinguished data scientists who helped shape today’s big data landscape – namely Geoffrey Hinton, Andrew Ng, and Yann LeCun. Now we train the spotlight on three accomplished gentlemen who helped further the big data revolution.

Artificial intelligence (AI) has long been used as a means to efficiently accomplish work, taking away much of computational and sometimes labor off human hands. With the fairly recent developments in digital genomics, dashboards and platforms are now acting similar to the human brain, as pegged by Hinton. These personalities are responsible for upgrading AI into what is now coined as Natural Intelligence.

Yoshua Bengio: The Machine Learning Guru

Considered one of the forefathers of deep learning, Yoshua Bengiohas always been a curious fellow. His fascination towards the vast capabilities of computers led him to doing extensive research on data and computer systems. Through an assemblage of algorithms, it can somehow replicate the functions of the human brain. His research would lead him to teaming up with respected data scientists George Hinton and Yann LeCun at the labs of the Canadian Institute for Advanced Research (CIFAR).

This team-up, though largely unnoticed at first, would eventually lead to the development of deep learning. It also resulted in a multitude of awards for their achievements. Knowing the importance of their work, Bengio had a vision, and it revolved around computers recognizing objects, informational trends, and generating insights based on the data with less dependence on human inputs. If Hintonprovided the blueprint for deep learning, it was Bengio and LeCun who would further propagate the idea.

Excited about his concept, Bengio traded slivers of information with  LeCun, gradually building the framework for machine learning. He, along with the CIFAR group, would create algorithms that expand the capacity of artificial intelligence to deduce structure out of labeled and unlabeled data. This would lead to early forms of face recognition in machines, with computers able to assign values to images on their own.

The algorithms Bengio crafted have a striking resemblance to Hinton’s neural networks, but were expanded to make data platforms and applications more intuitive. He was able to successfully train machines to perform calculated deductions based on images and bulks of information fed to them. This would pave the way for machine learning in highly advanced data platforms, which tech companies and numerous industries employ today.

Apart from his work with Hinton and LeCun, Bengio is also the architect of Theano, a now-established open source library that enables users to easily make complex algorithms using the Python programming language, which are run on a graphics processing unit (GPU).

All told, Bengio’s curiosity on artificial intelligence didn’t exactly kill the cat; rather, it taught computers to automatically recognize cats without the need to manually input information.

Jürgen Schmidhuber: Infusing Brain Power with Artificial Intelligence 

Since he was a teenager, Jürgen Schmidhuber had one primary goal in mind, and that was to expand on the capabilities of artificial intelligence until it evolves to a form that has relatively more “brain power” than himself. As early as 1987, he was playing  with the idea of improving AI in a number of ways. He went on to create the very first meta-learning machine, which has the ability to solve a series of complex problems  on its own.

In 1991, Schmidhuber began exploring neural networks. Like Hinton, he produced algorithms that replicated the process in which the human brain generates ideas, sending commands to other parts of the body to follow. This was  no easy feat, but Schmidhuber remained steadfast in reaching his goal. He had research groups stationed in different facilities in Switzerland to put together his central ideas on deep learning and machine learning.

With a lot of  respected researchers and scientists working towards one common goal, Schmidhuber’s vision branched out to a wide variety of applications. The collaborative effort resulted in great improvements on image recognition technology, handwriting recognition, speech recognition, photo caption generation, machine translation, and character recognition.  It also garnered them a slew of awards from the tech world.

Schmidhuber and his team’s contributions are now being used by companies like Google, IBM, Microsoft, and Baidu, just to name a few. Remarkably, Schmidhuber still wasn’t content with his achievements. A more lifelike example would perfectly illustrate that a machine can indeed be smarter than him.

Schmidhuber ventured into robotics where he applied his vast knowledge on “natural intelligence” on robots. It seemed a tad ambitious at first, but he was able to evolve the intelligence of robots, significantly improving and expanding their skills.

His work is far from over. For a man this smart, creating a machine that matches his ability to deduce is quite challenging, but we need not worry. He is smart enough to actually outsmart himself.

Ray Kurzweil: From Naturally Intelligent to Natural Intelligence

The term “genius” and the name “Ray Kurzweil” are closely associated. After all, the English language may not have a more perfect adjective for the man who created the first CCD flatbed scanner, the first print to speech machine (for blind people), the first electronic synthesizer that replicates the sound of various instruments, the first speech recognition machine in the mainstream market, and the first omni-font optical character recognition platform.

On top of these inventions and innovations, Kurzweil is a renowned author, a successful entrepreneur, a scriptwriter for movies, a philosophy geek, an inspirational speaker and an advocate of the futurist movement. A large part of his success, apart from impressive strong entrepreneurial skills, can be attributed to his deep love for science fiction. That love would transcend book covers and bleed into the realm of data analytics.

After earning a degree in Computer Science, Kurzweil established a company called Kurzweil Computer Products Inc.,  where he  studied the application of machine learning algorithms. This helped him in the  creation of the text to speech machine, the scanner and ultimately, the omni-font character recognition system. To further the development of his inventions, other institutions were commissioned to create working models. This gave Kurzweil ample time to focus on selling and marketing his products.

Coming from a musically inclined family, Kurzweil  brought  his creative mind to the field of music. Using his knowledge on technology and a lot of inspiration from a meeting with popular musician Stevie Wonder, he created a souped-up synthesizer that sound can perfectly imitate the sound of actual instruments. He also contributed to the improvement of speech recognition programs for the blind.

Kurzweil’s body of work is nothing short of impressive. At this stage, he  reaped numerous awards. Forbes Magazine called him  “the ultimate thinking machine.” It’s just fitting that the expanded use of “natural intelligence” is propagated by someone filled with true natural intelligence.

The combined efforts of Bengio, Schmidhuber, and Kurzweil have helped numerous tech companies build great  solutions and convenient features that  tremendously benefit individuals and industries worldwide. Their work serves as an inspiration to future scientists and investors. It’s a testament to human ingenuity as well.

Related Posts
The Who’s Who of Deep Learning – Part 1