The project is divided into three main parts: Generating a noisy sinusoidal signal as input. Training a linear autoencoder to perform PCA-like transformation. Visualizing reconstructed signals and ...

1. autoencoding of 784 (= 28 x 28 black and white pixel) MNIST data 2. autoencoding of 784 (= 28 x 28 black and white pixel) Fast Fourier Transformed (FFT) MNIST data 3. feature spaces consisting of m ...

Abstract: This paper presents a novel structure for data-driven fault detection using multi-layer autoencoder whose hidden layer trajectory is embedded into a linear space. The hidden layer is ...

The linear dimension reduction transformation \(\mathbf{w ... By introducing the dimension reduction layer in the autoencoder structure based on deep learning, we can extend the current DNN framework ...

Soil structure interaction (SSI ... We exploit a data-driven non-linear ROM methodology coupling an autoencoder with long short-term memory (LSTM) neural networks. The ROM is trained to emulate a ...

Abstract: This paper presents a novel structure for data-driven fault detection using multi-layer autoencoder whose hidden layer trajectory is embedded into a linear space. The hidden layer is ...