ch18 Flashcards
Machine learning
systems that learn without being
programmed to learn
Deep learning
machines that think in a way similar to the human brain. They handle huge amounts of data using artificial neural networks/////
* Uses artificial neural network(s)
* … that contain(s) a high number of hidden layers
* … modelled on the human brain.
relation between machine learning and deep learning
Deep learning is a subset of machine learning, which is itself a subset of AI
AI can be split into 3 main categories what are they
- Narrow AI: is when a machine has superior performance to a human when doing one specific task.
- General AI: is when a machine is similar in its performance to a human in any intellectual task.
- Strong AI: is when a machine has superior performance to a human in many tasks
Labelled data
data where we know the target answer and the data object is fully recognised
Unlabelled data
data where objects are undefined and need to be manually recognised
Supervised learning
system which is able to predict future outcomes based on past data. It requires both input and output values to be used in the training process////
* Supervised learning allows data to be collected, or a data output produced, from the previous experience.
* In supervised learning, known input and associated outputs are given // uses sample data with known outputs (in
training) // uses labelled input data.
* Able to predict future outcomes based on past data.
Unsupervised learning
system which is able to identify hidden patterns from input data – the system is not trained on the ‘right’ answer. the algorithm is given input data without explicit instructions on what to do with it. In unsupervised learning, the system tries to learn the patterns and the structure inherent in the data without explicit guidance in the form of labeled outputs or target variables./////
* Unsupervised machine learning helps all kinds of unknown patterns in data to be found.
* Unsupervised learning only requires input data to be given.
* Uses any data // not trained on the right output
Reinforcement learning
system which is given no training – learns on basis of ‘reward and punishment’.
Semi-supervised (active) learning
system that interactively queries source data to reach the desired
result. It uses both labelled and unlabelled data, but mainly unlabelled data on cost grounds.
analysis used in supervised learning
Supervised learning makes use of regression analysis and classification analysis.
Regression analysis
statistical technique used in data analysis to model and examine the relationship between a dependent variable and one or more independent variables. The main goal of regression analysis is to understand how the dependent variable changes as the independent variables vary. This analysis helps in predicting the value of the dependent variable based on the values of the independent variables//// statistical measure used to make
predictions from data by finding learning relationships
between the inputs and outputs.
Data mining
the process of discovering patterns, trends, correlations, or valuable insights from large sets of data. It involves extracting useful and meaningful information from vast and often complex datasets. Data mining is a crucial aspect of artificial intelligence (AI) because it enables machines to learn and make predictions based on historical data.
Classification analysis
type of statistical and machine learning technique used to categorize data into predefined classes or groups based on the characteristics of the input variables. The primary goal of classification is to build a model that can learn from labeled training data and make predictions or decisions about the class labels of new, unseen data.
more about supervised learning
» The system requires both an input and an output to be given to the model
so it can be trained.
» The model uses labelled data, so the desired output for a given input is known.
» Algorithms receive a set of inputs and the correct outputs to permit the learning process.
» Once trained, the model is run using labelled data.
» The results are compared with the expected output; if there are any errors, the model needs further refinement.
» The model is run with unlabelled data to predict the outcome.
Reward and punishment
improvements to a model
based on whether feedback is positive or negative; actions
are optimised to receive an increase in positive feedback.
how does Reinforcement learning learn
on the basis of ‘reward and
punishment’
Web crawler
internet bot that systematically browses the world wide web to update its web page content.
how does deep learning structures algorithms
in layers (input layer, output layer and hidden layer(s))
Artificial neural networks
networks of interconnected nodes based on the interconnections between neurons in the human brain. The system is able to think like a human using these neural networks, and its performance improves with more data.
Back propagation
method used in artificial neural
networks to calculate error gradients so that actual node/neuron weightings can be adjusted to improve the
performance of the model. It involves feeding input data forward through the network to produce an output, calculating the error between the predicted output and the actual target values, and then iteratively adjusting the weights of the network in the reverse direction (backward) to minimize the error. This process is repeated through multiple iterations or epochs until the network learns to make accurate predictions on new, unseen data.
summary of the back propagation process
» The initial outputs from the system are compared to the expected outputs and the system weightings are adjusted to minimise the difference between actual and expected results.
» Calculus is used to find the error gradient in the obtained outputs: the results are fed back into the neural networks and the weightings on each neuron are adjusted (note: this can be used in both supervised and unsupervised networks).
» Once the errors in the output have been eliminated (or reduced to acceptable limits) the neural network is functioning correctly and the model has been successfully set up.
» If the errors are still too large, the weightings are altered – the process continues until satisfactory outputs are produced
two types of back propagation
static and recurrent:
» Static maps static inputs to a static output.
» Mapping is instantaneous in static, but this is not the case with recurrent.
» Training a network/model is more difficult with recurrent than with static.
» With recurrent, activation is fed forward until a fixed value is achieved.
Outline the reasons for using deep learning. (2)
Deep learning makes good use of unstructured data.
* Deep learning outperforms other methods if the data size is large.
* Deep learning systems enable machines to process data with a nonlinear approach
* It can provide a more accurate outcome with higher numbers of hidden layers.
Explain how artificial neural networks enable machine learning. (4)
- Artificial neural networks are intended to replicate the way human brains
work - Weights / values are assigned for each connection between nodes
- The data are input at the input layer and are passed into the system
- They are analysed at each subsequent (hidden) layer where
characteristics are extracted / outputs are calculated - … this process of training / learning is repeated many times to achieve
optimum outputs // reinforcement learning takes place
State the reason for having multiple hidden layers in an artificial neural network. (1)
- Enables deep learning to take place
- Where the problem you are trying to solve has a higher level of
complexity it requires more layers to solve - To enable the neural network to learn and make decisions on its own
- To improve the accuracy of the result.
