Neural network in Robocode Neural network The human
											Neural network in Robocode
											Neural network • The human brain can be described as a biological neural network—an interconnected web of neurons transmitting elaborate patterns of electrical signals • In machine learning and cognitive science, artificial neural networks (ANNs) are inspired by biological neural networks and are used to estimate functions that can depend on a large number of inputs • are generally presented as systems of interconnected "neurons" which can compute values from inputs, and are capable of machine learning
											Example
											Neural networks 2 • An ANN is typically defined by three types of parameters: Ø The interconnection pattern between the different layers of neurons Ø The learning process for updating the weights of the interconnections Ø The activation function that converts a neuron's weighted input to its output activation. • feed forward, because their graph is a directed acyclic graph • with cycles are commonly called recurrent
											Real-life applications • The tasks artificial neural networks are applied to tend to fall within the following broad categories: Ø Function approximation, or regression analysis, including time series prediction, fitness approximation and modeling. Ø Classification, including pattern and sequence recognition, novelty detection and sequential decision making. Ø Data processing, including filtering, clustering, blind source separation and compression. Ø Robotics, including directing manipulators, prosthesis. Ø Control, including Computer numerical control.
											Resilient Propagation • Resilient Propagation (RPROP) is one of the best general purpose training methods for neural networks • is not the best training method in every case, but in most cases it is • will typically outperform backpropagation by a considerable factor • RPROP has no parameters that must be set • Backpropagation requires that a learning rate and momentum value be specified
											Our outputs
											Our outputs 2
											Our outputs 3
											Our outputs 4
											Problems with NN application • org. encog. ml. data. basic. Basic. MLData • org. encog. ml. data • Deprecated class • Finding a good color palette
											THANK YOU FOR YOUR ATTENTION Peter Babič Jakub Hanák
- Slides: 12