中文版本，請看 這裡.
AI Applications in daily life
Artificial intelligence (AI) is developing by leaps and bounds, and AI can now take over many tasks that once had to be done by humans. From simple background noise reduction to sophisticated medical diagnostics, environmental conservation, and urban development, AI can be seen.
Other examples of AI's relevance to our daily lives can be found in areas such as consumer recommendation (recommendation system, including content introduction, orientation, product promotion, etc.), autonomous driving, and even music creation and multilingual simultaneous interpretation.
But what exactly is AI? And why is it one of the skills that our next generation must master?
Applications of AI in different aspects of life. Source: Nvida Talk at IFF2021
Maxine, a real-time conversation interpretation function developed by NVIDIA with AI
3 breakthroughs in AI
According to Dr. Charles Cheung, Senior Data Scientist and Deputy Director of Nvida in Hong Kong, at the Innovate for Future Innovation Competition organized by the Hong Kong Electronic Technology Association. Sharing at the webinar, AI is not new.
It originated from the electronic brain proposed by S. McCulloch and W. Pitts 70 years ago, and it has been improved many times since then, but it has not been successful. It was not until recent human technology made significant breakthroughs in three areas that AI was able to develop rapidly.
1. Increase in the speed of computer computing (computational resources)
2. Increase in data processing amount
3. Optimization in algorithm
The breakthroughs in these areas can eventually train the virtual neural network, so AI can be widely used today.
3 Major AI Areas
One of the foundations of AI is Accelerated Data Science, which uses mathematics and computer computing to imitate the processes that humans perform.
The 3 major AI areas include:
1. Data Analytics
2. Machine Learning
3. Deep Learning
We will discuss each of them in the following.
The 3 major areas of AI; Source: Nvida Talk at IFF2021
GPU: Hardware foundation for AI implementation
The speed and volume of data analysis is the primary requirement for AI implementation. GPUs cannot replace CPUs, but help CPUs to process a large number of repetitive operations, such as determining the attributes or authenticity of images, to achieve machine learning or deep learning.
With the rapid advancement of technology, computation processes that used to take years and months can now be completed in just a few hours. This breakthrough has opened the door for many entrepreneurs to develop new technologies and products through Cloud Computing, for example, for those who were not familiar with AI.
History of AI-related technology development; Source: Nvida Talk at IFF 2022
Machine Learning
On the other hand, machine learning uses mathematical models, such as different formulas and models (e.g. Decision Tree, Random Forest, etc.), to deduce different results for learning purposes. One of the widespread applications is Object Detection).
For example, we give the system a thousand pictures of animals to learn the characteristics of different animals. After training, the system can perform object detection on its own without relying on external programs or human instructions, and without complicated coding. This saves a lot of manpower and resources.
How GPUs work together with CPUs to improve data processing and analysis speed; Source: Nvida Talk at IFF 2022
What is deep learning?
What is the difference between machine learning and deep learning? In short, machine learning takes data (such as a picture of an animal) and analytically extracts its features in different tests to learn the results.
Deep learning, on the other hand, automatically detects and classifies data features, starting with the simplest features (e.g., color) and advancing layer by layer (e.g., facial features) to achieve the best results for object recognition.
Unlike machine learning, these processes are performed automatically in deep learning, eliminating the need for manual instructions. This is what is meant by "deep". Recent studies have shown that deep learning is more effective than manual classification.
Starting from a young age to acquire these necessary future skills
For all AI, the current bottleneck in development is the need for large amounts of data for mapping and learning. However, it is conceivable that in the near future, with the increase in data and the continuous optimization of algorithms, AI will soon be integrated into every detail of our daily life.
We believe our next generation must be equipped and master the technology from a young age in order to keep up with the times and keep innovating to avoid being left behind.
AI application areas and related companies, many of which are familiar to all of us; Source: Nvida Talk at IFF2021
What's Next?
Fortunately, with the help of today's technology, AI is no longer a cold and astringent coding knowledge, even children can learn it easily.
For example, Blueinno offers the following AI courses covering different areas.
1. For basic programming: JavaScript for AI and AI with Python
2. Applications: Human and Object Recognition and Blockchain Fundamentals of Fintech
If adults are interested, they can enroll in AI-related courses at the NVIDIA Deep Learning Institute, or contact its education partner Blueinno for more information.