One day in early December 2012, a secret auction was taking place in a casino hotel in the American ski resort of Lake Tahoe.
Lake Tahoe is located at the junction of California and Nevada. It is the largest alpine lake in North America. It has a sapphire-like lake surface and top-notch snow trails. "The Godfather 2" was filmed here, and Mark Twain once lingered here. However, due to its distance from San Francisco Bay The area is just over 200 miles away, and is often called the “backyard of Silicon Valley.” Big bosses such as Zuckerberg and Larry Ellison also occupy mountains and build mansions here.
The target of the secret auction is DNNresearch, a company that has just been established for one month and has only three employees. It was founded by University of Toronto professor Geoffrey Hinton and two of his students.
The company does not have any tangible products or assets, but the identity of the suitors hints at its weight - the four buyers are Google, Microsoft, DeepMind and Baidu.
Harrah's Hotel holding secret auction, Lake Tahoe, 2012
Hinton, 65, old, thin and suffering from lumbar disc pain, sat on the floor of the hotel's Room 703 and set the rules for the auction - starting at $12 million, with bids going up to at least $1 million.
A few hours later, the bidders pushed the price to US$44 million. Hinton felt a little dizzy, "It felt like we were filming a movie," so he decisively stopped and decided to sell the company to the last bidder - Google. .
Interestingly, one of the sources of this $44 million auction came from Google 6 months ago.
In June 2012, Google Brain, the research department of Google, disclosed the research results of The Cat Neurons project (i.e., "Google Cat"). Simply put, this project is to use algorithms to identify cats in YouTube videos. It was initiated by Andrew Ng, who joined Google from Stanford, and attracted Google legend Jeff Dean to join him. It also obtained a large budget from Google founder Larry Page. .
The Google Cat Project built a neural network, downloaded a large number of videos from YouTube, and did not mark them, allowing the model to observe and learn the characteristics of cats on its own, and then used 16,000 CPUs spread across Google's various data centers for training (internally Rejecting the use of GPU due to excessive complexity and high cost), and finally achieved a recognition accuracy of 74.8%. This number shocked the industry.
Andrew Ng quit the "Google Cat" project before it came to an end and devoted himself to his own Internet education project. Before leaving, he recommended Hinton to the company to take over his job. Faced with the invitation, Hinton said that he would not leave college and was only willing to go to Google to "stay for a summer." Due to the special nature of Google's recruitment rules, Hinton, 64, became the oldest summer intern in Google's history.
Hinton has been fighting at the forefront of artificial intelligence since the 1980s. As a professor, he has many talents (including Ng Enda) and is a master-level figure in the field of deep learning. So when he learned about the technical details of the Google Cat project, he immediately saw the hidden flaw behind the project's success: "They were running the wrong neural network and using the wrong computing power."
Hinton thought he could do the same task better. So after the short "internship period" ended, he immediately jumped into action.
Hinton found two of his students—Ilya Sutskever and Alex Krizhevsky, both Jews born in the Soviet Union. The former was extremely talented in mathematics, and the latter was good at engineering implementation. The three worked closely together to create a new neural network. network, and then immediately participated in the ImageNet Image Recognition Competition (ILSVRC), and finally won the championship with an astonishing 84% recognition accuracy rate.
In October 2012, Hinton's team introduced the champion algorithm AlexNet at the Computer Vision Conference in Florence. Compared to Google Cat, which used 16,000 CPUs, AlexNet only used 4 NVIDIA GPUs. It caused a complete sensation in academia and industry. AlexNet The paper has become one of the most influential papers in the history of computer science. It has been cited more than 120,000 times, while Google Cat was quickly forgotten.
DNNresearch trio
Yu Kai, who had won the first ImageNet competition championship, was extremely excited after reading the paper, "like he was electrocuted." Yu Kai is a deep learning expert born in Jiangxi. He just jumped to Baidu from NEC. He immediately wrote an email to Hinton and expressed the idea of cooperation. Hinton readily agreed and simply packaged himself and two students into a family. The company invited buyers to bid, and so there was the opening scene.
After the auction came down, a bigger competition began: Google pursued the victory and acquired DeepMind in 2014, "all the heroes in the world are in it"; while DeepMind launched AlphaGo in 2016, shocking the world; losing to Google Baidu made up its mind to bet on AI and invested hundreds of billions in ten years. Yu Kai later helped Baidu invite Ng Enda, and he left his job a few years later to found Horizon.
Microsoft seemed slow on the surface, but in the end it won the biggest prize - OpenAI, whose founders include Ilya Sutskever, one of Hinton's two students. Hinton himself will stay at Google until 2023, during which he won the ACM Turing Award. Of course, compared with Google's $44 million (Hinton received 40%), the Turing Award's $1 million prize seems like pocket money.
From Google Cat in June, to the AlexNet paper in October, to the Lake Tahoe auction in December, in almost 6 months, the foreshadowing of the AI wave was almost completely buried - the prosperity of deep learning, GPU and The rise of Nvidia, the dominance of AlphaGo, the birth of Transformer, the emergence of ChatGPT...the grand movement of the silicon-based era has played the first note.
In the 180 days from June to December 2012, the fate of carbon-based humans was forever changed—only a few people realized this.
01
liquid cat
Among these very few people is Stanford University professor Li Feifei.
In 2012, when the results of Hinton's ImageNet competition came out, Li Feifei, who had just given birth to her child, was still on maternity leave, but the error rate of Hinton's team made her realize that history was being rewritten. As the founder of the ImageNet Challenge, she bought the last flight of the day to Florence to personally award the Hinton team [2].
Li Feifei was born in Beijing and grew up in Chengdu. When she was 16, she immigrated to the United States with her parents. She helped in a laundry while finishing her studies at Princeton. In 2009, Li Feifei entered Stanford as an assistant professor. Her research direction is computer vision and machine learning. The goal of this subject is to enable computers to understand the meaning of pictures and images on their own like humans.
For example, when a camera takes a picture of a cat, it just converts the light into pixels through the sensor, without knowing whether the thing in the lens is a cat or a dog. If a camera is compared to a human eye, the problem solved by computer vision is to equip the camera with a human brain.
The traditional way is to abstract things in the real world into mathematical models. For example, abstracting the characteristics of a cat into simple geometric figures can greatly reduce the difficulty of machine recognition.
Image source: Li Feifei’s TED talk
But this idea has very big limitations, because cats are likely to be like this:
In order to enable computers to recognize "liquid cats", a large number of deep learning pioneers such as Jeff Hinton and Yann LeCun began exploring in the 1980s. However, there will always be a bottleneck in computing power or algorithm - good algorithms lack sufficient computing power, and algorithms with small computing power requirements cannot meet the recognition accuracy and cannot be industrialized.
If the "liquid cat" problem cannot be solved, the sexiness of deep learning can only remain at the theoretical level, and industrialized scenarios such as autonomous driving, medical imaging, and precise advertising push are just castles in the air.
To put it simply, the development of deep learning needs to be driven by the troika of algorithms, computing power, and data. The algorithm determines how the computer identifies things; but the algorithm needs enough computing power to drive it; at the same time, the improvement of the algorithm requires Large-scale and high-quality data; the three complement each other and are indispensable.
After 2000, although the computing power bottleneck was gradually eliminated with the rapid advancement of chip processing capabilities, mainstream academic circles still showed little interest in the deep learning route. Li Feifei realized that the bottleneck may not be the accuracy of the algorithm itself, but the lack of high-quality, large-scale data sets.
Li Feifei's inspiration comes from the way three-year-old children understand the world - taking cats as an example, children will meet cats again and again under the guidance of adults, and gradually grasp the meaning of cats. If a child's eyes are treated as a camera, and one eye movement is equivalent to one click of the shutter, then a three-year-old child has taken hundreds of millions of photos.
Apply this method to a computer. Suppose the computer is shown pictures of cats and other animals and writes the correct answer behind each picture. Each time the computer looks at the picture, it checks it against the answer on the back. Then given enough times, it's possible for a computer to grasp the meaning of cat just like a child.
The only problem that needs to be solved is: Where can I find so many pictures with written answers?
Li Feifei came to China in 2016 and announced the establishment of Google AI China Center
This is the opportunity for ImageNet to be born. At that time, even the largest data set, PASCAL, only had four categories with a total of 1,578 images, and Li Feifei's goal was to create a data set containing hundreds of categories with a total of tens of millions of images. It may not sound difficult now, but you have to remember that it was 2006 and the most popular mobile phone in the world was the Nokia 5300.
Relying on the Amazon crowdsourcing platform, Li Feifei's team solved the huge workload of manual annotation. In 2009, the ImageNet data set containing 3.2 million images was born. With the image data set, the algorithm can be trained on this basis to allow the computer to improve its recognition capabilities. But compared to the hundreds of millions of photos of three-year-olds, 3.2 million is still too small.
In order to continue to expand the data set, Li Feifei decided to follow the popular practice in the industry and hold a picture recognition competition. Participants brought their own algorithms to identify the pictures in the data set, and the one with the highest accuracy won. However, the deep learning route was not mainstream at the time. ImageNet could only "affiliate" under the well-known European competition PASCAL at the beginning to barely get the number of participants.
By 2012, the number of images in ImageNet had expanded to 1,000 categories, with a total of 15 million images. Li Feifei spent 6 years making up for the shortcomings in data. However, the best result of ILSVRC has an error rate of 25%, which still does not show enough convincing in terms of algorithm and computing power.
At this time, Mr. Hinton appeared on the stage with AlexNet and two GTX580 graphics cards.
02
convolution
The Hinton team’s champion algorithm, AlexNet, uses an algorithm called Convolutional Neural Networks (CNN). "Neural network" is an extremely high-frequency word in the field of artificial intelligence, and it is also a branch of machine learning. Its name and structure are drawn from the way the human brain operates.
The process of human beings identifying objects is that the pupil first takes in pixels, the cerebral cortex performs preliminary processing through edges and directions, and then the brain makes judgments through continuous abstraction. Therefore, the human brain can identify objects based on some characteristics.
For example, most people can recognize who the person in the picture below is without showing their entire face:
Neural networks actually simulate the recognition mechanism of the human brain. In theory, intelligent computers that can be realized by the human brain can also be realized. Compared with SVM, decision tree, random forest and other methods, only simulating the human brain can process unstructured data like "Liquid Cat" and "Half Trump".
But the problem is that the human brain has about 100 billion neurons, and there are trillions of nodes (that is, synapses) between neurons, forming an extremely complex network. For comparison, the "Google Cat" composed of 16,000 CPUs had a total of 1 billion nodes inside, and this was already the most complex computer system at the time.
This is why even the "father of artificial intelligence" Marvin Minsky is not optimistic about this route. When he published his new book "The Emotion Machine" in 2007, Minsky still expressed his pessimism about neural networks. In order to change the mainstream machine learning community's long-term negative attitude towards artificial neural networks, Hinton simply renamed it deep learning (Deep Learning).
In 2006, Hinton published a paper in Science, proposing the concept of "Deep Belief Neural Network (DBNN)" and providing a multi-layer deep neural network training method, which is considered a major breakthrough in deep learning. . However, Hinton's method requires a large amount of computing power and data, making it difficult to implement in practical applications.
Deep learning requires constant feeding of data to the algorithm. The size of the data sets at that time was too small until the emergence of ImageNet.
In the first two ImageNet competitions, participating teams used other machine learning routes, and the results were quite mediocre. The convolutional neural network AlexNet used by Hinton's team in 2012 was improved from another deep learning pioneer Yann LeCun. The LeNet proposed by him in 1998 allowed the algorithm to extract key features of images, such as Trump's blond.
At the same time, the convolution kernel will slide on the input image, so no matter where the detected object is, the same features can be detected, greatly reducing the amount of calculations.
Based on the classic convolutional neural network structure, AlexNet abandons the previous layer-by-layer unsupervised method and performs supervised learning on the input values, greatly improving the accuracy.
For example, in the picture in the lower right corner of the picture below, AlexNet did not actually identify the correct answer (Madagascar cat), but it listed all small mammals that can climb trees like Madagascar cats, which means that the algorithm can not only identify the object itself , and can also be speculated based on other objects [5].
Image source: AlexNet paper
The industry is excited that AlexNet has 60 million parameters and 650,000 neurons. Complete training of the ImageNet data set requires at least 262 quadrillion floating-point operations. But Hinton's team only used two Nvidia GTX 580 graphics cards during a week of training.
03
GPU
After Hinton's team won the championship, the most embarrassing thing was obviously Google.
It is said that Google has also conducted internal tests on the ImageNet data set, but the recognition accuracy lags far behind Hinton's team. Considering that Google has hardware resources that are unmatched by the industry, as well as the huge data scale of search and YouTube, Google Brain leaders have specially appointed special tasks, and the results are obviously not convincing enough.
Without this huge contrast, deep learning may not have shaken the industry and gained recognition and popularity in a short period of time. The reason why the industry is excited is that Hinton's team only used four GPUs to achieve such good results, so computing power is no longer a bottleneck.
When the algorithm is training, it performs hierarchical operations on the functions and parameters of each layer of the neural network to obtain the output results, and the GPU happens to have very strong parallel computing capabilities. Ng actually proved this in a 2009 paper, but when running "Google Cat" with Jeff Dean, they still used the CPU. Later, Jeff Dean specially ordered equipment worth 2 million US dollars, which still did not include the GPU [6].
Hinton is one of the few people who has realized the huge value of GPU for deep learning very early. However, before AlexNet ranked the list, high-tech companies generally had an unclear attitude towards GPU.
In 2009, Hinton was invited to work as a short-term technical consultant for a speech recognition project at Microsoft. He suggested that the project leader Deng Li buy the top Nvidia GPU and match it with the corresponding server. This idea was supported by Deng Li, but Deng Li’s boss Alex Acero believed that this was purely a waste of money [6]. “GPUs are used for playing games, not for artificial intelligence research.”
Deng Li
Interestingly, Alex Acero later moved to Apple to be responsible for Apple’s speech recognition software Siri.
Microsoft's noncommittal about GPUs obviously made Hinton a little angry. He later suggested in an email that Deng Li buy one set of equipment, while he would buy three sets, and said in a weird way [6]: After all, we are the same university. A Canadian university with deep pockets, not a cash-strapped software seller.
But after the 2012 ImageNet Challenge, all artificial intelligence scholars and technology companies made a 180-degree turn towards GPUs. In 2014, Google's GoogLeNet won the championship with a recognition accuracy of 93%, using NVIDIA GPUs. This year, the number of GPUs used by all participating teams soared to 110.
The reason why this challenge is regarded as the "big bang moment" is that the shortcomings of the troika of deep learning - algorithms, computing power, and data have all been made up, and industrialization is only a matter of time.
At the algorithm level, the paper on AlexNet published by Hinton's team has become one of the most cited papers in the field of computer science. The technical route that originally had a hundred schools of thought contending has become one dominated by deep learning, and almost all computer vision research has turned to neural networks.
In terms of computing power, the super parallel computing capabilities of GPUs and the adaptability of deep learning were quickly recognized by the industry. Nvidia, which started deploying CUDA six years ago, became the biggest winner.
At the data level, ImageNet has become the touchstone of image processing algorithms. With high-quality data sets, algorithm recognition accuracy is improving day by day. In the last challenge in 2017, the recognition accuracy of the champion algorithm reached 97.3%, surpassing humans.
At the end of October 2012, Hinton's student Alex Krizhevsky published a paper at the Computer Vision Conference in Florence, Italy. Then, high-tech companies around the world began to do two things regardless of cost: one was to buy up Nvidia's graphics cards, and the other was to poach all AI researchers from universities.
Lake Tahoe’s $44 million has repriced the world’s deep learning masters.
04
capture the flag
Judging from publicly available information, Yu Kai, who was still at Baidu at the time, was indeed the first person to poach Hinton.
At that time, Yu Kai served as the head of Baidu Multimedia Department at Baidu, which was the predecessor of Baidu Deep Learning Institute (IDL). After receiving Yu Kai's email, Hinton quickly replied that he agreed to cooperate, and by the way, he expressed his wish for Baidu to provide some funds. Yu Kai asked for a specific number, and Hinton said that one million US dollars would be enough - this number was unbelievably low and could only hire two P8s.
Yu Kai asked Robin Li for instructions, and the latter readily agreed. After Yu Kai replied that there was no problem, Hinton, perhaps feeling the hunger of the industry, asked Yu Kai if he would mind asking other companies, such as Google. Yu Kai later recalled[6]:
"I felt a little regretful at the time, guessing I might have answered too quickly and alerted Hinton to a huge opportunity. But, I just graciously said I didn't mind."
In the end, Baidu and Hinton's team missed out. But Yu Kai was not unprepared for this result. Because on the one hand, Hinton has severe lumbar disc health problems and cannot drive or fly, making it difficult to bear the trip across the Pacific to China; on the other hand, Hinton has too many students and friends working at Google, and both parties The origin is too deep, and the other three companies are essentially accompanying the bid.
If the influence of AlexNet is still concentrated in the academic circle, then the secret auction in Lake Tahoe completely shocked the industry - because Google spent US$44 million to buy a less than one-year-old company under the noses of global technology companies. Months, a company with no products, no revenue, only three employees and a few papers.
The most exciting one was obviously Baidu. Although it failed in the auction, Baidu's management witnessed firsthand how Google invested in deep learning at all costs, which prompted Baidu to make up its mind to invest and announced the establishment of deep learning at the annual meeting in January 2013. Institute IDL. In May 2014, Baidu invited Andrew Ng, a key figure in the "Google Cat" project, and in January 2017, it invited Lu Qi, who had left Microsoft.
Google continued its efforts after acquiring Hinton's team, and in 2014, it bought DeepMind, that year's bidding rival, for US$600 million.
At that time, Musk recommended DeepMind, which he invested in, to Google founder Larry Page. In order to take Hinton with him to London to check its quality, the Google team also specially chartered a private jet and modified the seats to solve the problem that Hinton could not The problem of flying [6].
"British player" DeepMind defeated Lee Sedol in the Go game in 2016
Competing with Google for DeepMind is Facebook. When DeepMind fell to Google, Zuckerberg turned to poach Yang Likun, one of the "Big Three" of deep learning. In order to bring Yang Likun under his command, Zuckerberg agreed to many of his demanding requirements, such as setting up the AI laboratory in New York, completely drawing a clear line between the laboratory and the product team, allowing Yang Likun to continue to work at New York University, and so on.
After the 2012 ImageNet Challenge, the field of artificial intelligence faced a very serious problem of "talent supply and demand mismatch":
As industrialization spaces such as recommendation algorithms, image recognition, and autonomous driving are rapidly opened up, the demand for talents has skyrocketed. However, because it has not been favored for a long time, deep learning researchers are only a small circle, and top scholars can be counted on two hands, and the supply is seriously insufficient.
In this case, hungry technology companies can only buy "talent futures": poach professors and wait for them to bring in their own students.
After Yang Likun joined Facebook, six students followed him to join the company. Apple, which is preparing to take a leap in building cars, hired Hinton's student Ruslan Salakhutdinov to serve as Apple's first AI director. Even the hedge fund Citadel joined the competition and poached Deng Li, who worked on speech recognition with Hinton and later participated in secret bidding on behalf of Microsoft.
We can’t be more clear about the history since then: industrial scenarios such as facial recognition, machine translation, and autonomous driving are advancing thousands of miles day by day. GPU orders are drifting like snowflakes to NVIDIA’s headquarters in Santa Clara. The theoretical building of artificial intelligence is also developing day by day. pouring.
In 2017, Google proposed the Transformer model in the paper "Attention is all you need", ushering in today's large model era. A few years later, ChatGPT was born.
The birth of all this can be traced back to the ImageNet Challenge in 2012.
So, in which year did the historical process that promoted the birth of the "Big Bang Moment" in 2012 appear?
The answer is 2006.
05
great
Before 2006, the current situation of deep learning can be summarized by Baron Kelvin's famous saying: The building of deep learning has been basically completed, but there are three small dark clouds floating under the sunny sky.
These three little dark clouds are algorithms, computing power and data.
As mentioned before, deep learning is a theoretically perfect solution because it simulates the mechanism of the human brain. But the problem is that both the data it needs to swallow and the computing power it needs to consume are on a sci-fi scale at the time. From sci-fi to the mainstream academic view of deep learning, scholars with normal brains will not study neural networks. .
But three things happened in 2006 that changed that:
Hinton and his student Salakhutdinov (the one who later went to Apple) published the paper Reducing the dimensionality of data with neural networks in Science, which for the first time proposed an effective solution to the vanishing gradient problem, taking the algorithm level to a new level. Stride.
Salakhutdinov (first from left) and Hinton (middle), 2016
Li Feifei of Stanford University realized that if the data scale makes it difficult to restore the original appearance of the real world, then no matter how good the algorithm is, it will be difficult to achieve the effect of "simulating the human brain" through training. So, she started building the ImageNet data set.
NVIDIA released a new Tesla-based GPU and subsequently launched the CUDA platform. The difficulty for developers to use GPUs to train deep neural networks has been greatly reduced, and the daunting computing power threshold has been cut by a large margin.
The occurrence of these three things blew away the three dark clouds over deep learning, and they converged at the 2012 ImageNet Challenge, completely rewriting the fate of the high-tech industry and even the entire human society.
But in 2006, neither Jeff Hinton, Fei-Fei Li, Jen-Hsun Huang, nor others who promoted the development of deep learning could obviously have predicted the subsequent prosperity of artificial intelligence, let alone the role they would play.
Paper by Hinton and Salakhutdinov
Today, the fourth industrial revolution driven by AI as its core has begun again, and the evolution of artificial intelligence will only get faster and faster. If we can say how much inspiration we can get, perhaps it is nothing more than the following three points:
1. The thickness of the industry determines the height of innovation.
When ChatGPT came out, there were many voices saying "Why is it the United States again?" But if you stretch out the time, you will find that from transistors and integrated circuits to Unix and x86 architecture, and now to machine learning, American academia and industry are almost all leaders.
This is because, although there are endless discussions about the "hollowing out of industries" in the United States, the computer science industry with software as its core has not only never been "exodus" to other economies, but has become increasingly advantageous. To date, almost all of the more than 70 ACM Turing Award winners are Americans.
The reason why Andrew Ng chose Google to cooperate with the "Google Cat" project is largely because only Google has the data and computing power required for algorithm training, and this is based on Google's strong profitability. This is the advantage brought by the thickness of the industry - talents, investment, and innovation capabilities will all move closer to the high ground of the industry.
China also reflects this "thickness advantage" in its own advantageous industries. The most typical one at present is new energy vehicles. On one side, European car companies charter flights to the China Auto Show to learn from the new forces, and on the other side, executives of Japanese car companies frequently switch jobs to BYD - for what? Obviously, I am not just trying to pay social security in Shenzhen.
2. The more cutting-edge the technical field, the greater the importance of talents.
The reason why Google is willing to spend $44 million to buy Hinton's company is because in cutting-edge technology fields such as deep learning, the role of one top scholar is often greater than that of 10,000 fresh graduates majoring in computer vision. If Baidu or Microsoft had succeeded in bidding at that time, the development of artificial intelligence might have been rewritten.
This kind of behavior of "buying the whole company just for you" is actually very common. At the critical stage of Apple's self-development of chips, it bought a small company called PASemi just to poach the chip architecture guru Jim Keller - Apple's A4, AMD's Zen, and Tesla's FSD chips all got Jim Keller's Poverty alleviation through technology.
This is also the biggest advantage brought by industrial competitiveness - the attraction to talents.
None of the "Big Three" of deep learning is American. The name AlexNet comes from Hinton's student Alex Krizhevsky. He was born in Ukraine under the Soviet Union, grew up in Israel, and came to Canada to study. Not to mention the many Chinese faces who are still active in American high-tech companies today.
3. The difficulty of innovation lies in how to face uncertainty.
In addition to the "father of artificial intelligence" Marvin Minsky's opposition to deep learning, another well-known opponent of deep learning is Jitendra Malik of the University of California, Berkeley. Both Hinton and Andrew Ng have been ridiculed by him. Li Feifei also consulted Malik when building ImageNet, and the latter gave her advice: Do something more useful.
Li Feifei’s Ted Talk
It was the disapproval of these industry pioneers that caused deep learning to experience decades of silence. Even when Hinton broke a ray of light in 2006, Yang Likun, another member of the Big Three, was still repeatedly proving to the academic community that "deep learning also has research value."
Yang Likun has been studying neural networks since the 1980s. While at Bell Labs, Yang Likun and his colleagues designed a chip called ANNA to try to solve the computing power problem. Later, AT&T required the research department to "empower business" due to operating pressure. Yang Likun's answer was "I just want to study computer vision. If you can, fire me." In the end, I got the hammer, and I am pleased to mention N+1[6].
Researchers in any cutting-edge technology field must face a problem - what if this thing cannot be made?
Since entering the University of Edinburgh in 1972, Hinton has been fighting on the front lines of deep learning for 50 years. He was 65 years old when the 2012 ImageNet Challenge was held. It is hard to imagine how much self-doubt and denial he had to overcome in the face of various doubts from the academic community over a long period of time.
We now know that Hinton in 2006 had persisted until the last darkness before dawn, but he himself may not know this, let alone the entire academic community and industry. Just like when the iPhone was released in 2007, most people probably had the same reaction as then Microsoft CEO Ballmer:
Currently, the iPhone is still the most expensive phone in the world, and it doesn’t have a keyboard
Those who promote history often fail to guess their own coordinates in the historical process.
The reason why greatness is great is not because of its stunning appearance when it emerges, but because it has to endure a long period of being unknown and not understood in the boundless darkness. It is not until many years later that people can follow these rulers and lament that the stars were shining brightly and geniuses emerged in large numbers.
In one field after another of scientific research, countless scholars have never seen the glimmer of hope throughout their lives. Therefore, from a certain perspective, Hinton and other deep learning promoters are lucky. They have created greatness and indirectly promoted one success after another in the industry.
The capital market will set a fair price for success, and history will record the loneliness and sweat that create greatness.
Babbitt Park is open for cooperation!
Chinese Twitter: https://twitter.com/8BTC_OFFICIAL
English Twitter: https://twitter.com/btcinchina
Discord community: https://discord.gg/defidao
Telegram channel: https://t.me/Mute_8btc
Telegram community: https://t.me/news_8btc
