# Articles by Yoshiyuki Kobayashi

## Tutorial. Using original loss functions

Posted by Yoshiyuki Kobayashi

Neural Network Console provides basic loss functions such as SquaredError, BinaryCrossEntropy, and CategoricalCrossEntropy, as layers. However, depending on the problem, there are many cases in which you need to optimize using original loss functions. This tutorial explains how to define your own loss functions that are not available in Neural Network Console and use them in training.   1. Which value is handled as loss? Before explaining how to define loss functions, let’s review how loss functions are handled on…

## Tutorial. Expressing complex networks concisely using the unit function

Posted by Yoshiyuki Kobayashi

From version 1.20 of the Neural Network Console Windows version, it has become possible to use the unit function to concisely express complex neural networks, such as those with nested structures. This tutorial describes how to use the unit function by taking the basic convolutional neural network LeNet as an example. LeNet   1. What you can do with the unit function In the LeNet shown above, the structure Convolution→MaxPooling→ReLU appears twice consecutively immediately after the Input layer. If you…

## Tutorial. Network that uses several types of input data

Posted by Yoshiyuki Kobayashi

This tutorial describes how to handle neural networks that use several types of data as inputs. This method, for example, can be used to perform classification based on multiple images or based on image and vector inputs.   1. Performing classification based on multiple images First, the method of estimating y based on three images x, x2, and x3 will be explained. 1.1 Preparing a dataset for handling multiple input image data The Neural Network Console’s dataset CSV supports the…

## Tutorial. Two methods of using neural networks trained on Neural Network Console using Neural Network Libraries

Posted by Yoshiyuki Kobayashi

Neural networks that have been trained on Neural Network Console can be executed only using the open source Neural Network Libraries (without using Neural Network Console). This tutorial explains two methods of executing inference on neural networks that have been trained on Neural Network Console. One method uses the command line interface of Neural Network Libraries. The other method uses the Python API.   1. Method of executing inference using the command line interface Inference using the command line interface…

## Tutorial. Profiling the training processing time

Posted by Yoshiyuki Kobayashi

This tutorial describes the profiling function, which measures in detail the processing time (wall clock time) needed to perform training and classification on neural networks that have been designed. In addition to measuring the processing time of each layer and each forward/backward calculation during training, using this function allows you to confirm the execution order of processes in each layer during training and the implementation used to process each layer. Profiling preparation First, using Neural Network Console as usual, we…

## Tutorial. Analyzing partial output of trained neural networks

Posted by Yoshiyuki Kobayashi

This tutorial explains how to view what kind of data is being output in the middle of a trained neural network. We will use the LeNet sample project that contains a trained neural network. The sample project uses the MNIST dataset. Loading a trained neural network with weight To analyze using a trained neural network, right-click the training result on the TRAINING tab to open a shortcut menu, and click Open in EDIT Tab with Weight. This will load the…

## Tutorial. Using vectors or matrices as input to the neural network

Posted by Yoshiyuki Kobayashi

In addition to images, Neural Network Console also supports vector and matrix inputs. This tutorial explains how to input vector and matrix data. How to input vectors (entering directly in a dataset CSV file) You can directly enter vectors in dataset CSV files of Neural Network Console. For example, to use a five-dimensional vector, create five rows, x__0 to x__4 (variable name x, double underscores, vector element index 0 to 4), as shown below, and enter the element values in…

## Tutorial. Estimating continuous values based on input image

Posted by Yoshiyuki Kobayashi

This tutorial explains the training method of a neural network that calculates some continuous values based on input images. The following figure illustrates a neural network that outputs continuous values based on images. Illustration of a neural network that outputs continuous values based on input images A problem that estimates function y=f(x), which determines continuous values y from input data x, is called regression in the field of machine learning and statistics. For example, this neural network can be used…

## Dataset management window

Posted by Yoshiyuki Kobayashi

On the dataset management window, you can create or add datasets and preview datasets you used in the past.   Using the dataset management window   1 Creating an image classification dataset based on images grouped in different folders for each image category Have the image files grouped in folders ready. In the window that appears after the application starts, select DATASET to open the dataset management window. Click Create Dataset. Or, press Ctrl+N on the keyboard. An image classification…