Codes referred in the blog on SignalProcessing and AI https://dsplog.com

This repository organizes code snippets referred to in the articles on the website https://dsplog.com.

Code Organization

Multiclass classification with Softmax

Read the detailed article : Gradients for multiclass classification with Softmax

• derivative of softmax - implementation of softmax gradient calculation
• gradients for cross entropy loss - computing gradients for cross entropy loss function
• training multi-class classification - complete training pipeline for multi-class classification
• label smoothing implementation - label smoothing technique for improving classification

BPSK Bit Error Rate

Read the detailed article: Bit Error Rate (BER) for BPSK modulation

• bit_error_rate_vs_snr.ipynb: Analysis of BER vs SNR for BPSK
• bpsk_unequal_source_probabilities.ipynb: BPSK with non-uniform source probabilities
• bpsk_with_noise.ipynb: BPSK modulation with AWGN noise

Gradients in Binary Classification

Read the detailed article: Gradients for Binary Classification

• gradients_binary_cross_entropy_loss.ipynb: Computing gradients for binary cross entropy
• sigmoid_and_derivative.ipynb: Sigmoid function and its derivative
• training_loop_binary_classification.ipynb: Complete training loop implementation
• training_probit.ipynb: Training with probit regression

Gradients for Linear Regression

Read the detailed article: Gradients for Linear Regression

• gradients_absolute_function.ipynb: Gradients for absolute error function
• gradients_analytic_vs_finite_difference_vs_pytorch.ipynb: Comparison of gradient computation methods
• linear_regression.ipynb: Basic linear regression implementation
• linear_regression_mean_abs_error.ipynb: Linear regression with MAE loss
• linear_regression_pytorch.ipynb: PyTorch implementation of linear regression

Feel free to explore the notebooks and use the code as a reference for your projects!