Type 2 of Functional AI — Limited Memory AI
Limited memory is a type of AI mostly used in the present scenario, which stores the previous data or predictions and derives knowledge from it.
Limited memory AI is the second type based on functionality. It is considered to be one of the most popularly used kinds of AI today. We can see many applications using it. AI systems like ones using deep learning are trained by large volume of data that they store in their memory to use as reference models to solve future problems. It learns from the past and builds experiential knowledge by observing actions or data. Limited memory AI makes use of historical, observational data with pre-programmed information. Depending on this data it gets easy to make predictions and perform complex classification tasks. It extracts knowledge from previously learnt information, facts, stored data with the help of machine learning. Unlike, reactive machines, limited memory can learn from the past by analysing actions given to them. We can say, limited memory has a temporary memory. This AI is applied by virtual voice assistants, chatbots, self-driving cars and many other technologies.
Advantages of Limited Memory over Reactive AI
- It can retain some information from previously observed events or data.
- It can build knowledge using that memory in conjunction with pre-programmed data.
- It can give improved decisions with the help of past data.
The best example under Limited Memory AI is autonomous vehicles. It is used to observe other cars’ speed and direction, helping them “guide the road” and adjust if needed. Self-driving cars use the data collected recently in past to make quick decisions. By understanding and interpreting incoming data, cars can be safe on roads. As like human-reliable cars, they can drive and function adequately. Self-driving cars monitor their environment and consider other traffic participants’ movement also consider the movement of other traffic participants in the line of sight. Before Limited AI, previously needed more time about 100 seconds to react and make decisions based on external factors.
Though with so many features, Limited AI lack in some of the practical applications. We have more advanced theories that are leveraged nowadays, which may be discussed one by one.
