Artificial Intelligence: Part 8

Deep Learning

Deep learning is a type of machine learning based on neural networks with many layers. These systems can learn complex patterns in images, text, audio and video. Deep learning powers many modern AI breakthroughs, including speech recognition, image generation, translation and large language models. It often requires large amounts of data and computing power.

Diffusion Model

A diffusion model is a type of generative AI often used to create images. It learns to generate data by starting with noise and gradually refining it into a coherent image or output. Many modern image-generation tools use diffusion techniques. The result can be highly realistic or artistic, depending on the prompt and model.

Digital Twin

A digital twin is a virtual model of a real-world object, system or process. For example, a factory might have a digital twin that simulates machines, energy use and production flow. AI can make digital twins more powerful by predicting problems and testing changes before they happen in the real world. Digital twins are used in engineering, healthcare, transport and smart cities.

Discriminative Model

A discriminative model learns to distinguish between categories. For example, it might decide whether an email is spam or not spam, or whether a tumour image appears benign or malignant. Discriminative models are different from generative models, which create new content. In simple terms, discriminative models classify; generative models produce.

Distillation

AI distillation is training a small, fast AI model to copy the knowledge and behavior of a large, smart AI model.Think of it as a master chef (the large model) teaching an apprentice (the small model) all their secret recipes and techniques so the apprentice can cook the same quality food, just much faster and with fewer kitchen tools.Why It Is UsedSaves space: It shrinks massive AI systems so they can fit on a smartphone or laptop.Saves money: It costs significantly less computer power to run the smaller model.Saves time: The smaller model gives answers much faster than the giant one.

Domain adaptation

Domain adaptation is the process of making an AI model work well in a new but related setting. For example, a model trained on hospital data from one country may not work equally well in another country because equipment, patients or procedures differ. Domain adaptation helps models transfer knowledge while adjusting to local conditions.

Domain-Specific AI

Domain-specific AI is AI designed for a particular field, such as healthcare, law, finance, education, farming or manufacturing. These systems may be more accurate and useful than general tools because they are built around specialist knowledge and workflows. However, domain-specific AI needs expert oversight because mistakes can carry serious consequences.

Drift

Drift occurs when an AI model’s performance changes over time because the real world changes. For example, customer behaviour, fraud tactics, medical practice or language use may evolve. A model that worked well last year may become less reliable today. Monitoring for drift is essential in any AI system used over time.

Dropout

Dropout is a training technique used in neural networks to reduce overfitting. During training, some connections are randomly ignored, forcing the model not to rely too heavily on any one pathway. This helps the model generalise better to new data. It is similar to training a team so that the whole group performs well, rather than depending on one star player.

Dual Use

Dual use means a technology can be used for both helpful and harmful purposes. AI is a classic dual-use technology. The same tool that helps write educational material could also help write phishing emails. The same image-generation system that helps artists could create fake evidence. Responsible AI requires thinking about misuse as well as intended use.

Set The Portugal News as your preferred source for news on Google

Add as a preferred source on Google