Conrad CRF engine - The generic liner chain semi-Markov CRF infrastructure and algorithms contained within Conrad. These can be used as a machine learning toolkit for using CRFs in any domain.
Conrad gene caller - The Conrad CRF engine applied to the gene calling. The Conrad gene caller consists of the Conrad engine supplemented with additional data handling utilities, a state model for gene calling, and gene calling specific feature functions. The Conrad gene caller has driven the development of the Conrad engine.
CRF - A Conditional random field. See Wikipedia for more.
Features - Features are the individual predictors which make up a Conrad model. Features use the input data to provide a numerical value for a given segment, state, or transition at each point in an input sequence.
GTF - Gene Transfer Format. The default format that Conrad uses to write out gene predictions. See Wikipedia for more.
Inference - In machine learning, inference is the process of deriving conclusions or making predictions based on existing information. See Wikipedia for more.
Machine Learning - Machine learning is a branch of computer science covering software that uses data to improve it's accuracy at some given task. Conrad uses supervised learning, which is one component of machine learning. See Wikipedia for more.
Model file - A model file is an XML file that configures the Conrad engine for a particular problem. It describes the input data, output data, features, state model, and inference, and optimization. When Conrad is trained takes a model and a set of training data to produce a training file.
Supervised learning - A machine learning technique whereby a system uses a set of training examples to learn how to correctly perform a task. In the case of Conrad Gene Prediction, the training examples are correctly predicted genes. See Wikipedia for more.
Training examples - Training data is a set of input data for a model along with a set of correct "answers". Conrad uses the training data to set the parameters of the model correctly.
Training file - A training file is a binary file produced when Conrad parameterizes a model file on a given set of training examples.