3 subfields of AI explained simply

How are machine learning, neural networks and deep learning interconnected?

Artificial intelligence is a computer’s version of human-like intelligence. A human can naturally understand speech, use languages, produce creative outputs, reason, and solve problems. A computer can be taught to do it too by giving it massive amounts of data to read, analyze and work with. It learns to mimic human behavior. The term ‘artificial intelligence’ refers to non-human intelligence even though the data the computer is trained with comes, initially, from humans.
Artificial intelligence is a vast concept that can be divided into multiple subfields. The six largest subfields are machine learning, deep learning, neural networks, natural language processing, computer vision and cognitive computing. In this part, the first three subfields are explained simply and with a practical example, to aid comprehension on concepts that may be difficult to grasp!

Machine learning means what it is named after
A computer is fed information and can teach itself to learn from that data. This requires little to no human intervention when the machine is given access to even unlimited sources of data. Therefore, it can make predictions and decisions. GPTs are examples of programs that have access to vast amounts of data from the internet and can perform incredibly well in tasks that normally humans do. Machine learning is also used in autonomous vehicles where the software is constantly making decisions, whether it’s for navigation or recognizing a pedestrian. It is good to understand that machine learning is at the top of the hierarchy of AI subfields, and machine learning, neural networks and deep learning are inherently interconnected.
Neural networks mimic a human brain
The brain has a massive number of neurons that send and receive information to the brain. A computer’s version of this is called a neural network, a set of algorithms, that identify correlations from data. They consist usually of 2-3 layers in which the data is received, analyzed, and put out. For example, Netflix uses neural network to analyze data from its users like what movies have been previously watched or which series never got finished. It then recognizes patterns in user behavior and can give personal recommendations that match user preferences the best. Neural networks are also used in facial, speech and object recognition, anomaly detection in cybersecurity and even medical diagnostics. Essentially, all the above are sets of large amounts of data that neural networks can analyze.
Deep learning allows a computer to learn by example
Like humans do. Deep learning is a form of machine learning where a computer analyses data trying various algorithms and programs until it finds the perfect fit for an output. Deep learning also uses neutral network technology. What differentiates the two, however, is the number of layers that the process consists off; Deep learning can have as much as 150 layers. One key advantage of deep learning is the improvement in results as the amount of data is scaled. The everyday use of a smartphone includes lots of deep learning technology, like the automated categorization of photos or the blurry background -effect that can be activated in a smartphone camera. For example, the data in a camera roll allows the deep learning algorithm to recognize ‘beach’ or ‘nighttime’ in photos and categorize by those labels.

Lack of understanding in massive technological shifts create uncertainty and fear. When new concepts are introduced quickly, a proper deep dive and education help dissolving bias, making new technology feel more familiar to the population. Through practical examples it is easier to understand how artificial intelligence is used and what are the many layers below the concept.
Sanni S
Sources:
.png)



https://soicaulodechuan.mobi/giai-ma-so-hoc cái mình thấy dễ chịu là bài có mục lục đánh số ngay đầu, không phải đọc một hồi mới biết phần cần tìm nằm ở đâu. Mình hay tranh thủ xem vài phút thôi, gặp bài nhiều chữ là hơi ngại kéo xuống dò từng đoạn. Có sẵn các đầu mục như vậy thì nhìn qua cũng nắm được bài đang chia nội dung thế nào, đỡ mất công tìm hơn. Còn mấy nhận định về số thì mình chưa nói tới, ở đây chỉ thấy cách trình bày khá dễ đọc lướt. Một điểm nhỏ nữa mà mình thích là kết quả xổ số miền Bắc hôm qua có hẳn mục riêng. Phần đó có tiêu đề…
Mình xem xổ số miền Bắc kiểu giải trí cho vui, vừa thử chút may mắn vừa có cái để nói chuyện với mấy đứa bạn. Hồi trước mình chọn số khá tùy hứng, khi thì tin vào linh cảm, khi thì bám theo giấc mơ hoặc mấy con số bắt gặp ngoài đường, nên sai hoài cũng bình thường. Dạo này mình đổi cách một chút, đọc thêm vài bài nhận định rồi ghi chú lại những cặp hay xuất hiện, cảm giác đỡ rối hơn. Có lần mình lướt du doan xsmb rồi so lại với thói quen chọn số của mình, mới thấy mình hay bị “dính” mấy số quen mắt nên bỏ qua nhiều hướng khác.…