🧠 TGNAS Neuron – A C++ Neural Network from Scratch

This project is a hand-crafted implementation of a single-layer linear regression model (a.k.a. a basic neuron) using pure C++ — no external libraries, no frameworks, just raw math and logic.

Inspired by a hands-on session on "Build Your First Neural Network," this model demonstrates how the core mechanics of machine learning (like gradient descent and ReLU activation) work under the hood.

🔢 The model’s logic is based on a quirky mathematical signature derived from the name TGNAS:

T = 20, G = 7, N = 14, A = 1, S = 19
Resulting in a formula: y = 20x - 7x + 14x - 1x + 19x

---

🚀 What It Does

• Implements a simple linear regression model
• Trains using gradient descent
• Uses a ReLU activation function manually coded in C++
• Accepts live user input and makes real-time predictions
• Exits gracefully when you enter the keyword TGNAS

---

📚 Key Concepts Covered

• Linear regression
• Gradient descent
• Loss calculation
• ReLU (Rectified Linear Unit)
• Overfitting avoidance (with learning rate tuning)
• Manual weight & bias updates

---

🔧 How to Run

Prerequisites

• C++ compiler (e.g., g++)
• Terminal or command prompt

Compile & Run

g++ -o tgnasNeuron tgnasAlgo.cpp
./tgnasNeuron

Enter the value to predict : 10
Prediction for 10 is : 450.12

Enter the value to predict : TGNAS
Exiting Cya!!