From a1324c054c53ea9001a4e751533faef668ba6261 Mon Sep 17 00:00:00 2001 From: patrickjcraig Date: Sat, 12 Oct 2024 18:43:26 -0400 Subject: [PATCH] Preliminary Readme Branch --- README.md | 52 ++++++++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 48 insertions(+), 4 deletions(-) diff --git a/README.md b/README.md index d99c62c..21667d9 100644 --- a/README.md +++ b/README.md @@ -1,5 +1,49 @@ # DL-Project1 -DL-Proj1 -links -[MNIST](https://yann.lecun.com/exdb/mnist/) -[Deep Learning for Classical Japanese Literature](https://arxiv.org/pdf/1812.01718) +## Description +Convolutional neural networks(CNNs) and multi-layer perceptrons are foundational models for baselines in deep learning. They are useful for processing images, automatically determining relevant feature values, and testing novel hyperparameter selection, system blocks, and backpropagation alternatives. + +The purpose of this repository is to test, on the Kuzushiji-MNIST dataset, the effects of different hyperparameter selection on the classification accuracy of these models. + +## Model Architectures +### MLP + +### CNN + +## Hyperparameter Selection + +### MLP + +| Hyperparameter | Value | +|----------------------|-------| +| Learning Rate | 0.01 | +| Batch Size | 64 | +| Number of Epochs | 50 | +| Dropout Rate | 0.5 | +| Optimizer | Adam | +| Weight Initialization| Xavier| + +### CNN + +| Hyperparameter | Value | +|----------------------|-------| +| Learning Rate | 0.001 | +| Batch Size | 128 | +| Number of Epochs | 100 | +| Dropout Rate | 0.25 | +| Optimizer | SGD | +| Weight Initialization| He | + + +## Classification Metrics +| Model | F1-score | Precision | Recall | Accuracy | +|------------------------|----------|-----------|--------|----------| +| CNN | 0.95 | 0.96 | 0.94 | 0.95 | +| Multi-layer Perceptron | 0.92 | 0.93 | 0.91 | 0.92 | + +## Learning Rate +![Learning Rates](path/to/your/image.png) + + +## Relevant Links +- [MNIST](https://yann.lecun.com/exdb/mnist/) +- [Deep Learning for Classical Japanese Literature](https://arxiv.org/pdf/1812.01718)