When I started using LLM for work and personal use, I picked up some technical terms, such as “machine learning” and “deep learning”, which are central technology behind these LLMs. I am always interested in learning about the differences between these technologies. Most companies in the industry are now producing their AI tools, which make MLPs necessary to manage and use them.
Before I started learning about MLOs, I tried to understand how I work behind LLM and how they work. In this article, I will share my understanding of machine learning, depicting, and Generative AI with their potential applications.
The table of content
Artificial intelligence (AI)
Artificial intelligence (AI) is a form of technology that allows machines to solve problems that are similar to how people work for it. This helps businesses make widespread decisions by helping to identify photos, produce content, and make data -based predictions. Artificial intelligence includes machine learning, deep learning, and generative AI.
Machine Learning (ML): Foundation
When we give many examples to computers, they learn how to make their decisions or estimates. This is like teaching a child the difference between animals. You show them many pictures of cats and dogs and say things like “this is a cat” and “it’s a dog”. Finally, they learn to make the difference between cats and dogs themselves. The machine learning is similar in which you give a computer a lot of data with examples, and learns how to make predictions about new data.
How does a machine learn work?
Machine learning (ML) is the process of providing computers in data searching for samples in data and makes decisions or predictions without guidance. This process usually has six main steps:
Collecting data: Get many examples, such as thousands of emails, photos, or sales record. The more training data you have, the more your predictions will be.
Data PreparationAt this stage, you clean the data by getting rid of mistakes and adding missing labels.
Choosing algorithm (model): This is equivalent to choosing the right tools for work. Models can find or find samples in data. You can find machine learning models for your data Here.
Phase of Training: After selecting the right model for your cleaned data, you teach it. It is like getting ready for tests.
Diagnosis: Use the test data to evaluate the model’s performance and see if it can make accurate predictions on the data that are not visible.
Appointment: Keep a trained model to work in the real world.
The phase of trainingTeach 10,000 home -selling computer with details such as size (2,000 square feet), number of bedrooms (3), and location (city). Cost:, 000 300,000.
LearnSamples of algorithms are found, such as the fact that large homes cost more and cost more in the city center. More bedrooms make a house higher.
Forecast: Think about a new house with 1,800 square feet, two bedrooms, and a place in the suburbs. It is estimated that what he has learned is based on.
Types of machine learning
Learn under supervision: Give the algorithm labeled and fixed training data to find samples. Sample data tells the algorithm what to do as an output and what to expect. For example, millions of X -ray reports say that someone is healthy or ill, it will need to be tagged. After that, the machine learning programs can use this training data to determine whether a new X -ray disease shows signs.
Non -monitoring learning: Learn from the algorithm who uses unspecified learning from the data that are not labeled. The algorithm will have to look for samples in extraordinary data without any help. For example, search for groups of people on Facebook or Twitter who have similar interests.
Learn: This technique is a type of machine learning in which an agent learns how to communicate with the world around him. The agent earns points to work and loses points to do the wrong job. It aims to get maximum points. For example, cars learn how to drive safely by imitation in imitation. They find revenge for staying in their lane, following traffic rules, and not targeting other cars.
Machine Learning-Examples of the real world
E -mail spam detection
You can show thousands of emails to the computer that say “spam” or “no spam”. It learns samples, such as how “free money” emails are usually spam. Now it can automatically set your inbox.
The identity of the image
Give the computer with millions of images labels that say what is in them. It learns that apples are likely to be round and have trunks. Your phone can now tell what things are in your pictures.
The movie’s recommendations
Netflix monitors the films you’ve seen and ranked. This gives people who like the same thing you do. This suggests films that other people like.
Deep Learning: Adding complexity
Learning deep is a type of artificial intelligence. It helps computers to understand data like humans. Deep learning can indicate complex images, text, sound and other data samples to make correct predictions. It uses artificial nerve networks that act like the human brain. Nerve network is connected nodes that handle information.
How does a deep learning work?
Artificial nerve networks are used to learn from data. These networks consist of the integrated layers of nodes. Each node learns something different about data.
For example, when you show a cat picture to the computer, the picture goes through many steps. The first layer looks for shapes and edges. The second layer puts these shapes together for ears, eyes and whispers. The last layers say things like “this picture looks like a cat.” There may be many mistakes in learning deep when learning, but after each piece of opinion it becomes better and better.
Deep learning-examples of the real world
Tesla Auto Pilot: Views on the streets, acting on eight cameras simultaneously to identify traffic signs and prevent obstacles.
Google’s Deep Mind: 94 % detects more than fifty eye diseases from retina scans with accuracy.
Chat GPT: Helps to solve written, coding, and the problem.
Generative AI: Write new
Generative AI is a sub -set of deep learning that makes new things like stories, photos, music, or codes rather than just watching or sorting things. Generative AI systems learn samples from many training data and then use these samples to create new content.
Examples of real world
Chat boots help organizations provide better customer service by offering product tips and answering questions.
Automatically prepare technical documents from the source code.
Self -born quiz, exercise problems, and explanations
Machine Learning vs Deep Learning Vs
| Quality | Machine Learning (ML) | Deep Learning (DL) | Generative AI (Genai) |
| Applause | All sets of AI where machines learn from data to make predictions or decisions. | Sub -set of AI using artificial nerve networks with multiple layers to create models of complex samples | All sets of deep learning that can create new content (text, images, codes, etc.) like human -created content |
| Data requirements | Small medium datases. | Data in large quantities (structural and non -structured) | Massive datasis for training, different quantities for generation |
| Competition Power | The CPU works on moderate hardware. | GPUS/TPUS is required for training. | Massive GPU/TPU clusters are needed. |
| Use matters | Predictions and ratings. | Recognize complex figures such as speech, photos and language. | New, produce original content. |
| When not use | The data is very complicated/non -structured. Accuracy is important (medical, legal), need to handle photos/audio/video | The dataset is small (<1000 samples), and computational resources are limited. | Copyright/IP restriction |
| The cost of the cost | Low ($ 1k- $ 10k) (standard service) | Medium (K 10K-$ 100K) | High ($ 100K- $ 1M+) |
| Examples of real world | Netflix recommendations, fraud detection, spam filters. | Facial identity, self -driving cars, Siri/Alexa. | The original creative output (text, photos, code, video). |
Conclusion
To summarize this, anyone who wants to learn more about artificial intelligence needs to know the difference between machine learning, deep learning and generative AI.
Machine learning is the basis as it allows the computer to learn and predict data from data. Deep learning takes a step further by using nerve networks to process complex data patterns, which is equivalent to understanding things humans.
Generative AI goes a step further by making new things, which shows how creative AI can be. Since these technologies improve, they open many new opportunities in many fields, such as improving customer service, making medical diagnosis more accurate, and making new materials. To maximize the benefits of AI in your life, be current on new, new developments.