Chapter 1865 Granddaughter


Even though Zhou Zhi had thought that Mai Mingdong might give a very nonsensical answer, he never expected it to be like this and couldn't help but froze there.

"Ah what? First tell me what a no-op sled is?"

"Oh," Zhou Zhi quickly explained the principles of this system attack method to Mai Mingdong. Read it again.

Mai Mingdong's first reaction after hearing this was: "Shouldn't this be a research topic of your Clover System anti-virus department? Do you think I will work in this direction?"

There is nothing wrong with this. The projects that Mai Mingdong is responsible for are all national-level major projects, focusing on applications. This kind of work of plugging rat holes will not be looked at at all.

Zhou Zhi said with some embarrassment: "I found your old name when I was checking information. I thought it was weird at the time, so I called you. But the article on Yizhi.com is indeed your signed article."

"I'll check it out online later..." Mai Mingdong replied, and then Zhou Zhi heard the crackling sound of the keyboard over there, and then heard Mai Mingdong's surprised voice: "Huh? Still... It’s really weird...”

“You keep reminiscing?” Zhou Zhi asked: “You think I didn’t lie to you?”

“Hi! It's Xiaomiao who is playing a prank!" Mai Mingdong said cheerfully: "This little girl is too naughty. I'll deal with her when I get back!"

"Mr. Mai, this is Xiaomiao you are talking about. …”

"My granddaughter! She just came back from Berkeley. She was Cai Shaotang's student before. Do you know Cai Shaotang, right?"

AXA Information Industry Intelligence Department has established files for Chinese people who have made achievements in the global industry, Zhou Zhi Of course I know this big shot.

However, this person did not come from China, but was born in the Philippines to a Chinese family. After receiving a bachelor's degree in electrical engineering from the Mapua Institute of Technology in the Philippines in 1959, he went to the United States to study, and successively studied in Massachusetts. Polytechnic Institute and the University of Illinois at Urbana-Champaign, where he received his master's and doctoral degrees. He later taught at Purdue University and joined the University of California, Berkeley in 1971 as a professor in the Department of Electrical Engineering and Computer Science. He was also a foreign academician of the European Academy of Sciences and the Hungarian Academy of Sciences.

In the same year, he proposed the memristor theory.

Memristor is a passive circuit component related to magnetic flux and charge. It is considered to be the fourth basic circuit component after resistance, capacitance, and inductance. It plays an important role in information storage, logic operations, and neuromorphology. The fields of computing and nonlinear circuits have very important application prospects.

The proposal and realization of this concept brought fundamental changes to traditional circuit theory. On this basis, Cai Shaotang even proposed theories such as Chua's circuit and cellular neural network.

Chuai's circuit is a simple nonlinear electronic circuit design, but it can exhibit standard chaos theory behavior. In 1983, Cai Shaotang published the circuit while he was a visiting scholar at Waseda University in Japan.

This circuit is extremely easy to make, thus making it a ubiquitous real-world example of the existence of chaotic systems, leading some scholars to declare it a "model of a chaotic system."

The cellular neural network, referred to as CNN, is even more incredible.

Humanity has also gone through many detours in the neural network model, from the earliest nerve cell model, to the neural network model, to the exploration of the three-layer structure of the perceptron, and finally set off a wave of research in the 1960s. The climax of text recognition, image recognition, and voice recognition. However, soon, because the model at that time was too simple, the perceptron had huge limitations in both principle and function. It was not until Minsky and others from the Massachusetts Institute of Technology pointed out after research that the perceptron under the existing mechanism could not be used at all. It is impossible to identify linearly inseparable patterns, and even simple puzzles cannot be solved.

This research result directly put a damper on the research enthusiasm of perceptrons.

However, in the theoretical field, people have never stopped analyzing models of nonlinear chaotic states, and ideas and ideas such as "learning matrices" and "quasi-neurons" continue to appear.

This theory finally achieved great breakthroughs in the 1980s, such as "fully interconnected artificial neural network", "simulated annealing" methodology, "cognitive process microstructure theory", "back propagation learning algorithm" New methods such as "error correction" and "adaptive resonance theory" began to appear, and successfully proved that the nonlinear perception problems, complex pattern recognition problems, adaptive characteristics problems, and nonlinear system optimization problems that had troubled people before were completely possible. Solved through neural network theory.

On the basis of these achievements, Cai Shaotang proposed a method of circuit theoretical design and hardware implementation, namely the cellular neural network model, through his own research.

This is a locally interconnected, dual-value output signal nonlinear analog processor, which has the characteristics of continuous real-time, high-speed parallel computing, and is suitable for the implementation of very large-scale integrated circuits.

Unlike biological neurons, the connections between CNN cell neurons are mainly controlled by weight templates. Different templates have different nonlinear characteristics, and memristors with memory characteristics can be It is applied to the functional connection points between neurons to simulate the brain cell neuron network, realize the simulation and simplification of the information processing mechanism, and realize functions such as logical operations and image processing.

This research result directly opens the door for humans to apply artificial intelligence to many fields such as biomedicine, image processing, automatic control, pattern recognition, signal processing, secure communications, etc. Decades later Emerging technologies such as big data and blockchain are also closely related to it.

Although this technology represents the future development direction, it is actually a bit too advanced. Currently, it is basically still undergoing laboratory research, and there are not many problems that can really be used to solve it.

There is only one place in China that can provide such research, and that is the graph database used by Zhou Zhi in the digital library against all objections.

There are also application scenarios for radical recognition, pattern recognition, oracle bone fusion, chaotic super search, etc. for practice.

The advantage of graph databases lies in their powerful functions.

The current mainstream traditional relational database requires strict data normalization when designing, dividing the data into different tables and deleting duplicate data. This normalization ensures strong consistency of the data and improves the data quality. Row-by-row access can be quickly achieved only after huge restrictions are placed on the relationship.

But when complex relationships are formed between data, and cross-table correlation queries increase to the point where strong constraints become unbearable, problems arise.

Although complex queries can be performed by correlating different attributes that exist in different tables, the overhead becomes exponential. In programmer terms, the system is overwhelmed by huge data correlations. I was "suffocated" to death.

Graph databases do not have this problem. Although their data relationships are also mapped to data structures, their special organizational structure and network analysis functions make them the opposite of traditional relational databases. The higher the correlation, the more complex the data volume. The larger the data set, the faster the query speed, especially suitable for object-oriented applications.

At the same time, the graph database can be more naturally extended to big data application scenarios, because the construction of the graph database is not subject to the strong consistency constraints of the table structure and can be more flexible, so it is more suitable for managing temporary or changing data.

As someone who has traveled through time, Zhou Zhi certainly knows what the future trends will be, and also knows the significance of winning at the starting line. (End of this chapter)

Previous Details Next