AI is trying to simulate with a computer. Artificial intelligence is a machine that's able to learn, make decisions, and take action even when it encounters a situation it has never come across before. A normal program is made repeating the same steps over and over again. AI is able to learn and solve complex problems itself, even if its facing unseen problems.
Intelligence is the ability to learn, infer and reason. To understand information, solve problems and decision making.
There are several main area's within AI:
- Language
- Vision
- Robotics
- Decisions
Maching Learning
A subfield of AI is Machine Learning or (ML). This enables systems to learn from data and improve the performance on specific tasks without programming. Machine Learning is really good at looking at patterns within data, the more data you can feed it the more accurate in predicting it will become. There are 3 main types:
-
Supervised learning: Algorithm learns from labeled data associated with a know outcome or label, like image classification, spam detection.
-
Unsupervised learning: Algorithm learns from unlabeled data without providing outcome, like customer segmentation, anomaly detection.
-
Reinforcement learning: Algorithm learns through trial and error by interacting with enviroments receiving feedback as rewards or penalties.
ML algorithm can be trained on a dataset of bikes and motors. By analyzing features and patterns in these images the algorithm learns to distinguish between them.
Deep Learning
Deep Learning a subfield of ML uses neural networks to learn and extract features from data. These neural networks identify patterns and representations in large datasets like images, audio and text. Key features are:
- Hierarchical Feature Learning: Deep Learning models learn layered representations, where lower layers detect simple features (e.g., edges, textures) and higher layers capture complex patterns (e.g., shapes, objects).
- End-to-End Learning: Trained end-to-end, they can map raw input directly to desired outputs.
- Scalability: DL models scale well with large datasets.