• Multilayer perceptron - Feed-forward phase

      Title: Multilayer perceptron - Feed-forward phase

      Topic: FPGA accelerated computing

      Category Level: Problem solving

      Degree weight: 5

      Materials: Alphanumeric LCD, Marvell ARMADA 1500, VHDL, Xilinx ISE, Xilinx Isim

      Augmented Reality Interface: Y

      Remote Lab: N

      Short description: Artificial neural networks (ANN) are computational models inspired by biological neural networks of the brain. ANNs have found applications in many domains, such as signal processing, image analysis, speech recognition, and automation and control systems. One of the most well-known and widely used neural network is the multilayer perceptron (MLP). In this exercise an MLP will be used in a live video application. The MLP will implement an image segmentation function, classifying pixels into one of several groups. The segmentation is performed in real time with result immediately displayed on a monitor. In this exercise you will implement the basic unit of an MLP –the neuron. In the course of implementing the neuron, you will learn how to utilize computationa structures such as multipliers, accumulators and look-up tables in a data-stream context.

    Matrix multiplicationUART Core Implementation on E2LP Base Board platform