An AI model needs to be trained in order to learn and perform future actions. There are several ways for this to happen, and in this section, we will discuss three of them. Image
Expert Systems
Expert systems are designed to mimic the decision-making and problem-solving abilities of human experts in a specific knowledge domain. These systems use rules and knowledge gathered from experts to make informed decisions and provide solutions. These systems are developed for specific tasks and do not learn anything new on their own. They can best be compared to what we call flowcharts. If one action occurs, go right; if another action occurs, go left.
Although expert systems have their limitations and can depend on the accuracy and completeness of the acquired expertise, they are also useful tools. They are usually the safest systems and will deliver the result they are programmed to achieve.
Machine Learning (ML)
Machine learning is a branch of artificial intelligence (AI) that focuses on developing algorithms and methods that enable computers to learn from data. Machine learning gives computers the ability to analyze large amounts of data, identify patterns, and make decisions or predictions based on those patterns. Instead of being programmed with specific instructions to perform a particular task, machine learning systems learn from examples or experiences. This is done by training models with large datasets and improving performance based on the feedback the model receives. There are different training approaches for such AI models, but the common aspect is that they receive positive points when they perform a desired action and negative points when they perform an undesired action. Machine learning is used today in various areas such as autonomous driving, speech recognition, recommendation systems, medical diagnosis, and many others.
Deep Learning (DL)
Deep learning is a subfield of machine learning that focuses on building and training artificial neural networks. These neural networks are inspired by the structure and function of the human brain and are particularly effective in handling complex and abstract problems. In deep learning, data is fed into a network with multiple layers, and each layer processes the information at different levels. The initial layers of the network learn to identify simple features, such as lines or edges, while the later layers learn to combine everything to find advanced patterns. The ability of neural networks to work in this way allows them to adapt to new data and work independently. One of the key characteristics of deep learning is the ability to automate learning. This means that the model can learn to perform tasks directly from raw data without the need for human feedback. This is a great solution for tasks like image and speech recognition, natural language processing, translation, autonomous vehicle technology, and many other complex challenges. Deep learning requires large amounts of labeled training data and significant computational power.
Tasks Questions about the text: 3.1) What limits expert systems? 3.2) What can be good about expert systems? 3.3) What is the difference between expert systems and machine learning? 3.4) How does an AI model that uses machine learning improve? 3.5) What are the neural networks supposed to mimic? 3.6) What is one of the key characteristics of deep learning?
Vocabulary Bank 3.7)
Term | Explanation |
Flowchart | |
Neural Networks | |
Automate | |
Computational Power | |
Training Data | |