Artificial Intelligence Flashcards
What is meant by artificial intelligence (AI)?
Artificial Intelligence is the creation of intelligent machines that can react like humans do to certain stimuli.
What is meant by data mining?
Data mining is the study of extracting information from large amounts of data. This can also be called data analysis or business intelligence.
What is machine learning?
A subclass of data mining, machine learning attempts to allow a computer to learn from experience using large datasets.
What is deep learning?
A subclass of machine learning, deep learning attempts to allow a computer to learn directly from a dataset with little to no feature engineering.
Which came first - machine learning or deep learning?
Machine learning. Deep learning is a recent discovery/adaptation, allowing us to ‘skip’ the feature selection phase.
What is meant by ‘data’?
Data is some form of unprocessed information.
What is meant by ‘information’ in relation to data?
Information is data that has been processed to be useful to answer ‘who’, ‘what’, ‘when’ and ‘where’ questions. For example, information may be the color of the grass in an image.
What is meant by ‘knowledge’ in relation to data and information?
Knowledge is an application of data and information to answer ‘how’ questions.
Given a picture of a cat - which parts of the image is our data, information and knowledge?
Data is the image itself. Information is features we have picked out from that image, such as the colour of the grass or the location of the cat.
Knowledge may be inferred from that information about the image - i.e. it was taken in a field.
What is meant by ‘strong’ and ‘weak’ AI?
Strong AI exhibits general-purpose thinking, applying knowledge across a wide range of tasks, mimicking human intelligence.
Weak AI is built for a specific task, or a limited range of tasks, that operates within a predefined set of parameters.
What is meant by classification?
Classification is a technique of supervised learning where each possible ‘answer’ to each piece of data is one of a finite set of possible outputs.
What is meant by regression?
Regression is a technique of supervised machine learning where each piece of data’s ‘answer’ may be a range of values - e.g. an integer value representing a house’s value in 10 years.
What is meant by supervised learning?
Supervised learning is an application of machine learning where each row of data contains the ‘answer’, allowing us to compare the predicted and actual outputs.
What is meant by unsupervised learning?
Unsupervised learning is an application of machine learning where each row of data has no ‘answer’ attached. In this case, we typically cluster the data together so that similar outputs are connected.
What is one of the flaws of machine learning, and how can deep learning fix this?
One of the biggest flaws of machine learning is having to find the correct features - information - that is best to represent the model data. This can take a while in large datasets.
Deep learning skips this step, allowing us to perform feature extraction and classification at the same time.
