Authored by Dr. Victor Fang, Founder & CEO of AnChain.AI
Today brought groundbreaking news that stirred the entire AI community!
“ACM named Yoshua Bengio, Geoffrey Hinton, and Yann LeCun recipients of the 2018 ACM A.M. Turing Award for conceptual and engineering breakthroughs that have made deep neural networks a critical component of computing. “
As an AI researcher for the past 15 years and a hardcore supporter for Deep Learning / NeuralNets since ANN / perceptron era (yeah I’m that old), I have mixed feelings when I heard about this great news. The 3 professors that motivated me to start my CS PhD study specializing in AI / ML truly deserve this belated award.
I still vividly remember how shocked I was by Prof. Hinton’s preliminary results on DBN on various domains like computer vision and NLP in NIPS 2010 Vancouver.
Coincidentally, on this same date my startup officially launched our new website, which we’d been planning for months, with an updated mission statement:
AnChain.AI: Award Winning AI Products will Secure and Grow Your Blockchain Business.
This coincidence motivated me to spend a bit time to write a little article, to share my thoughts with the AI and Blockchain communities that I belong to.
What are the implications of these awards to the significance of AI, blockchain, and other emerging computational technologies?
Turing Award in Review — What Else Other Than AI?
This is NOT the first time the Turing Award has been won by Artificial Intelligence researchers. In fact, in 2011 it was awarded to Prof. Judea Pearl on his Bayesian Network contribution.
It usually takes decades for the ACM Turing Award committee to reach consensus on the Laureates.
Why blockchain? Turing Award 2012 was awarded to Prof. Silvio Micali, one of the world’s leading cryptographers, who founded Algorand, the blockchain consensus and startup.
My story with AI and Machine Learning
I was shocked when I first exposed to the idea of Machine Learning, a branch of AI, back in 2004.
I was an EE undergrad taking an image processing course that semester. One summer afternoon on my way to the gym I dropped by a senior friend’s dorm room, where he was busy working on a fingerprint recognition project. I looked at his code and understood most of the parts that involved Gaussian convolution, ridge extraction, etc. I’ve always been partial to mathematics.
Yet, one strange chunk of code aroused my curiosity. It was way different than image processing: it read in hundreds of images labeled by person ID, fed them into a magic function call ANN (Artificial Neural Network), and then the strange chunk of code spat out a data structure that could be used to recognize fingerprint images.
“This is mind blowing!” I called out.
I immediately saw the technological implications of this new software paradigm. Outside of coding up all the image filters and adjusting all the parameters to make it work, the new magic function actually “learned” from the data we fed into it.
Machine Learning. This is the concept that stuck into my head on that summer afternoon.
This is the practical route to real Artificial Intelligence. We build a machine that can learn from the data we feed it to solve real world problems.
What followed that enlightening summer afternoon was quite straightforward. My life had a new, clear goal. Cliche, but true!
I applied for the summer internship at one of the most prestigious research labs, National Lab of Pattern Recognition, and luckily got accepted out of hundreds of applicants. I was fortunate to work with an IEEE Fellow, Prof. Stan Li, who wrote the book Markov Random Field, and who demoed facial recognition to Bill Gates while he was at Microsoft Research.
Back in 2005 we experienced the AI winter, very similar to the current crypto winter. At that time, SVM, Random Forest, GBM, Bayesian Network Boosting, and even ANN were the hot topics in AI research communities. However, those algorithms sucked. They suffered from lots of problems, primarily low accuracy and high false positives.
None of those algorithms could meet real world applications. The best ML model I built at this time was a video object detection system classifying humans / bikes / cars that achieved 99% accuracy, but still failed on the real world deployment; the false positive rate was too high.
During that time I made up my mind to pursue my CS PhD specializing in AI, seeking for better alternatives than those algorithms that I had bad experiences with.
Why I Predict AI Researchers Will Win more Turing Awards
If we look at the last 7 years of Turing Awards, most of them are related to computer architecture. (Yeah Alan Turing’s machine is about architecture 😜)
Why are we building faster, bigger machines?
Simply because we can solve more problems with these machines!
1. With faster and bigger machines, data is generated at a more rapid rate.
2. With more data, machine learning and AI can be then trained more efficiently.
3. With faster CPU , more scalable architecture like cloud, these AI can then have a bigger impact on our world. Think about Alexa, Google, Facebook. We are all enjoying AI today without even realizing it.
Three key components to AI advancement:
1. Data is the fuel to AI research. Without millions, billions of annotated datapoints, there is NO way the 3 deep learning god fathers could have won the Turing Award.
2. Algorithm is the soul. The new frontiers such as Reinforcement Learning, AlphaStar by DeepMind that won StarCraft, AlphaGo that defeated world’s best Go players, etc. These mathematics are being perfected and disrupted by the AI research communities.
3. Hardware is the engine. Without GPU and Cloud, it’s literally impossible for small research teams to make progress in AI. With thousands of cores packaged into our GPU, thanks to Nvidia, even a small university lab can enjoy the technology that was only available for state level institutions as recently as 20 years ago.
Today’s AI innovation is right on the edge. This is why I predict the next few years will see more Turing Awards to the AI research community.
Why Blockchain is Relevant Here?
Blockchain represents the next wave of technology architecture. It’s more than just a distributed ledger, more than just a smart contract, more than just a cryptocurrency.
Blockchain is a disrupting technology that will change the world.
As mentioned, Turing Awards have already been awarded to a few researchers in cryptography and architecture, such as Prof. Silvio Macali(2012) and Prof. Andrew Chi-Chih Yao(2000). These two are now devoting their research toblockchain.
“The Future Has Arrived — It’s Just Not Evenly Distributed Yet” — William Gibson
My definition of Blockchain:
“Blockchain is a decentralized super computer with trust built on top of mathematics. “ — Victor Fang, 2019
The 3 elements of AI advancement that I summarized above fit perfectly into blockchain.
The new frontier for AI innovations is on Blockchain:
- Data: The distributed ledger implies all transactions need to be recorded for future verification. With certain level of permissions, privacy settings, these data are theoretically accessible to the trusted parties.
- Algorithm: There will be plethora algorithms designed for blockchain. For example, AnChain.AI has developed a few AI algorithms that detected Blockchain APT hackers hacking smart contracts on Ethereum that stole $4 million USD. This AI product is featured by MIT Technology Review, and it’s powering the blockchain ecosystem to secure and grow the businesses of Protocols, DApps, Exchanges, and Enterprises.
- Hardware: Blockchain is decentralized. The ecosystems have miners, senders, receivers, oracles, etc. Trusted execution environment (TEE) , Privacy Preserving Computation (PCC), these strong cryptographically and CPU intense computations impose new requirements for new hardware. I expect new hardware edge computing may be developed to enable secure and privacy preserving computing for blockchain infrastructure.
I am proud that the AnChain.AI team has been helping the blockchain ecosystem. For example, our AI powered smart contract sandbox leveraged machine learning to help developers to write secure smart contracts, and it’s FREE for developers.
Congrats on the Turing Award, laureates! I will always hold a special appreciation for the 3 AI professors that inspired and motivated me to start my career in AI. This path has led me to found AnChain.AI in 2018 and to pioneer this new blockchain technology frontier.
The AI powered blockchain era has only just begun. Join our mission at AnChain.AI . 🚀🚀🚀
A blockchain data analytics firm providing intelligence, indicators, and investigative resources for clients to enhance their security, risk, and compliance strategies.
Feel free to reach out to us directly at: email@example.com
With extensive experience in cybersecurity, artificial intelligence, cloud computing, and big data AnChain is continuously securing top-tier crypto exchanges, protocols, investors, custodians, and enterprise with our Blockchain Ecosystem Intelligence.