4. Limited Memory - tanicha/IT115-Wiki-Project GitHub Wiki
written by Kimberly Le
Limited memory consists of machine learning models that take knowledge from previously learned information, facts, stored data, and events. Limited memory can analyze past actions or data to improve its functions. Limited memory consists of technologies such as voice assistants, chatbots, self-driving cars, and many more. AI is continuously growing and developing in order to improve our lives and make it easier. Limited memory is required to create every machine learning models.
- Reinforcement Learning: Predictions through trial and errors
- Long Short-Term Memory (LSTMs): Predications in sequence
- Evolutionary Generative Adversarial Networks (E-GAN): Predications that evolve
- There's a team who continuously trains a model on new data
- AI environment is built in order for models to automatically train and renew
Learning Cycle
- Training Data
- Build
- Model Predictions
- Feedback
- Feedback becomes data
- Repeat Step 1
Self-Driving vehicles, also known as autonomous vehicles, use limited memory technology. Its software analyzes its environment detecting patterns and adapts quickly. It examines its surroundings such as traffic, pedestrians, and other vehicles to determine the appropriate speed or direction to go. According to the site, Sintelly, "In the past, autonomous vehicles without limited memory AI needed up to 100 seconds to react and make decisions based on external factors."