Harald ScheidlA simple neural network written in plain C++No Python, no PyTorch, no NumPy, … just 260 lines of plain C++ code to train, evaluate and test a simple binary classifier.Oct 15Oct 15
Harald ScheidlWhy pooling layers in deep learning can cause problemsPooling layers are omnipresent in today’s computer vision deep learning models. They reduce the size of the feature maps from layer to…Jul 21, 2022Jul 21, 2022
Harald ScheidlinTowards Data ScienceAdaHessian: a second order optimizer for deep learningMost of the optimizers used in deep learning are (stochastic) gradient descent methods. They only consider the gradient of the loss…Aug 9, 20211Aug 9, 20211
Harald ScheidlDoes padding matter in deep learning models?The article presents two experiments that show the influence of padding in deep learning models.Mar 24, 20211Mar 24, 20211
Harald ScheidlinTowards Data ScienceWhat a text recognition system actually seesSome insights into the neural network “black box” of a text recognition systemJan 5, 20199Jan 5, 20199
Harald ScheidlinTowards Data ScienceGPU Image Processing using OpenCLImplementation of two image processing methods in less than 120 lines of code using Python and OpenCL.Oct 29, 20182Oct 29, 20182
Harald ScheidlinTowards Data ScienceFAQ: Build a Handwritten Text Recognition System using TensorFlowSep 13, 201822Sep 13, 201822
Harald ScheidlinTowards Data ScienceWord Beam Search: A CTC Decoding AlgorithmImprove text recognition results: avoid spelling mistakes, allow arbitrary numbers and punctuation marks and make use of a word-level…Jul 19, 201813Jul 19, 201813
Harald ScheidlinTowards Data ScienceBeam Search Decoding in CTC-trained Neural NetworksA fast and well-performing algorithm with integrated language model to decode the neural network output in the context of text recognitionJul 10, 20188Jul 10, 20188