Building a Neural Network from Scratch: The Complete Implementation

Building a Neural Network from Scratch

Introduction

The best way to understand neural networks is to build one from scratch — no TensorFlow, no PyTorch, just NumPy and math. In this chapter, you'll implement a complete 3-layer neural network for handwritten digit recognition, including forward pass, backpropagation, and training loop. By the end, you'll truly understand what happens inside deep learning.

Architecture

Input (784) → Hidden 1 (128, ReLU) → Hidden 2 (64, ReLU) → Output (10, Softmax)

import numpy as np

class NeuralNetwork: def __init__(self, layer_sizes): self.layers = len(layer_sizes) - 1 self.params = {} # He initialization for i in range(self.layers): self.params[f'W{i+1}'] = np.random.randn(layer_sizes[i+1], layer_sizes[i]) * np.sqrt(2.0 / layer_sizes[i]) self.params[f'b{i+1}'] = np.zeros((layer_sizes[i+1], 1)) def relu(self, Z): return np.maximum(0, Z) def relu_derivative(self, Z): return (Z > 0).astype(float) def softmax(self, Z): exp_Z = np.exp(Z - np.max(Z, axis=0, keepdims=True)) return exp_Z / exp_Z.sum(axis=0, keepdims=True) def forward(self, X): cache = {'A0': X} A = X for i in range(1, self.layers): Z = self.params[f'W{i}'] @ A + self.params[f'b{i}'] A = self.relu(Z) cache[f'Z{i}'] = Z cache[f'A{i}'] = A # Output layer (softmax) Z = self.params[f'W{self.layers}'] @ A + self.params[f'b{self.layers}'] A = self.softmax(Z) cache[f'Z{self.layers}'] = Z cache[f'A{self.layers}'] = A return A, cache def compute_loss(self, Y_pred, Y_true): m = Y_true.shape[1] return -np.sum(Y_true * np.log(Y_pred + 1e-8)) / m def backward(self, Y, cache): m = Y.shape[1] grads = {} # Output layer dZ = cache[f'A{self.layers}'] - Y grads[f'dW{self.layers}'] = (1/m) * dZ @ cache[f'A{self.layers-1}'].T grads[f'db{self.layers}'] = (1/m) * np.sum(dZ, axis=1, keepdims=True) # Hidden layers for i in range(self.layers - 1, 0, -1): dA = self.params[f'W{i+1}'].T @ dZ dZ = dA * self.relu_derivative(cache[f'Z{i}']) grads[f'dW{i}'] = (1/m) * dZ @ cache[f'A{i-1}'].T grads[f'db{i}'] = (1/m) * np.sum(dZ, axis=1, keepdims=True) return grads def update(self, grads, lr=0.01): for i in range(1, self.layers + 1): self.params[f'W{i}'] -= lr * grads[f'dW{i}'] self.params[f'b{i}'] -= lr * grads[f'db{i}'] def train(self, X, Y, epochs=100, lr=0.01, batch_size=64): m = X.shape[1] losses = [] for epoch in range(epochs): indices = np.random.permutation(m) for start in range(0, m, batch_size): end = min(start + batch_size, m) X_batch = X[:, indices[start:end]] Y_batch = Y[:, indices[start:end]] Y_pred, cache = self.forward(X_batch) grads = self.backward(Y_batch, cache) self.update(grads, lr) Y_pred, _ = self.forward(X) loss = self.compute_loss(Y_pred, Y) losses.append(loss) if epoch % 10 == 0: acc = np.mean(np.argmax(Y_pred, axis=0) == np.argmax(Y, axis=0)) print(f"Epoch {epoch}: Loss={loss:.4f}, Acc={acc:.4f}") return losses

# Usage with MNIST-like digits
from sklearn.datasets import load_digits
digits = load_digits()
X = digits.data.T / 16.0  # Normalize to [0,1]
Y = np.eye(10)[digits.target].T  # One-hot encode

nn = NeuralNetwork([64, 128, 64, 10])
losses = nn.train(X, Y, epochs=100, lr=0.1)
print(f"Final accuracy: {np.mean(np.argmax(nn.forward(X)[0], axis=0) == digits.target):.4f}")

What You Just Built

This is a genuine neural network — the same mathematics used in GPT-4 and DALL-E, just smaller. The difference between this and ChatGPT is scale (billions of parameters vs hundreds) and architecture (transformers vs MLP), but the core learning mechanism is identical.

Practice Problems

1. Add L2 regularization to the network and compare overfitting behavior.

2. Add dropout to the hidden layers and compare training curves.

3. Port this network to PyTorch and verify you get the same results.

Key Takeaways

• A neural network = forward pass + loss computation + backpropagation + weight update
• He initialization prevents vanishing/exploding gradients at the start
• Mini-batch SGD is more efficient than full-batch gradient descent
• Softmax + cross-entropy is the standard output for multi-class classification
• Building from scratch gives deep understanding that frameworks can't provide

Engineering Perspective: Building a Neural Network from Scratch: The Complete Implementation

When you sit for a technical interview at any top company — whether it is Google, Microsoft, Amazon, or an Indian unicorn like Zerodha, Razorpay, or Meesho — they are not just testing whether you know the definition of building a neural network from scratch: the complete implementation. They are testing whether you can APPLY these concepts to solve novel problems, whether you understand the TRADEOFFS involved, and whether you can reason about system behaviour at scale.

This chapter approaches building a neural network from scratch: the complete implementation with that depth. We will examine not just what it is, but why it works the way it does, what alternatives exist and when to choose each one, and how real systems use these ideas in production. ISRO's mission control systems, India's UPI payment network handling 10 billion transactions per month, Aadhaar's biometric authentication serving 1.4 billion identities — all rely on the principles we discuss here.

Transformer Architecture: The Engine Behind GPT and Modern AI

The Transformer architecture, introduced in the landmark 2017 paper "Attention Is All You Need," revolutionised NLP and eventually all of deep learning. Here is the core mechanism:

# Self-Attention Mechanism (simplified)
import numpy as np

def self_attention(Q, K, V, d_k): """ Q (Query): What am I looking for? K (Key): What do I contain? V (Value): What do I actually provide? d_k: Dimension of keys (for scaling) """ # Step 1: Compute attention scores scores = np.matmul(Q, K.T) / np.sqrt(d_k) # Step 2: Softmax to get probabilities attention_weights = softmax(scores) # Step 3: Weighted sum of values output = np.matmul(attention_weights, V) return output

# Multi-Head Attention: Run multiple attention heads in parallel
# Each head learns different relationships:
# Head 1: syntactic relationships (subject-verb agreement)
# Head 2: semantic relationships (word meanings)
# Head 3: positional relationships (word order)
# Head 4: coreference (pronoun → noun it refers to)

The key insight of self-attention is that every token can attend to every other token simultaneously (unlike RNNs which process sequentially). This parallelism enables efficient GPU training. The computational complexity is O(n²·d) where n is sequence length and d is dimension, which is why context windows are a major engineering challenge.

State-of-the-art developments include: sparse attention (reducing O(n²) to O(n·√n)), mixture of experts (MoE — activating only a subset of parameters per input), retrieval-augmented generation (RAG — grounding responses in external documents), and constitutional AI (alignment through principles rather than RLHF alone). Indian researchers at institutions like IIT Bombay, IISc Bangalore, and Microsoft Research India are actively contributing to these frontiers.

India's Scale Challenges: Engineering for 1.4 Billion

Building technology for India presents unique engineering challenges that make it one of the most interesting markets in the world. UPI handles 10 billion transactions per month — more than all credit card transactions in the US combined. Aadhaar authenticates 100 million identities daily. Jio's network serves 400 million subscribers across 22 telecom circles. Hotstar streamed IPL to 50 million concurrent viewers — a world record. Each of these systems must handle India's diversity: 22 official languages, 28 states with different regulations, massive urban-rural connectivity gaps, and price-sensitive users expecting everything to work on ₹7,000 smartphones over patchy 4G connections. This is why Indian engineers are globally respected — if you can build systems that work in India, they will work anywhere.

Advanced Algorithms: Dynamic Programming and Graph Theory

Dynamic Programming (DP) solves complex problems by breaking them into overlapping subproblems. This is a favourite in competitive programming and interviews:

# Longest Common Subsequence — classic DP problem
# Used in: diff tools, DNA sequence alignment, version control

def lcs(s1, s2): m, n = len(s1), len(s2) dp = [[0] * (n + 1) for _ in range(m + 1)] for i in range(1, m + 1): for j in range(1, n + 1): if s1[i-1] == s2[j-1]: dp[i][j] = dp[i-1][j-1] + 1 else: dp[i][j] = max(dp[i-1][j], dp[i][j-1]) return dp[m][n]

# Dijkstra's Shortest Path — used by Google Maps!
import heapq

def dijkstra(graph, start): dist = {node: float('inf') for node in graph} dist[start] = 0 pq = [(0, start)]  # (distance, node) while pq: d, u = heapq.heappop(pq) if d > dist[u]: continue for v, weight in graph[u]: if dist[u] + weight < dist[v]: dist[v] = dist[u] + weight heapq.heappush(pq, (dist[v], v)) return dist

# Real use: Google Maps finding shortest route from
# Connaught Place to India Gate, considering traffic weights

Dijkstra's algorithm is how mapping applications find optimal routes. When you ask Google Maps to navigate from Mumbai to Pune, it models the road network as a weighted graph (intersections are nodes, roads are edges, travel time is weight) and runs a variant of Dijkstra's algorithm. Indian highways, city roads, and even railway networks can all be modelled this way. IRCTC's route optimisation for trains across 13,000+ stations uses graph algorithms at its core.

Research Frontiers and Open Problems in Building a Neural Network from Scratch: The Complete Implementation

Beyond production engineering, building a neural network from scratch: the complete implementation connects to active research frontiers where fundamental questions remain open. These are problems where your generation of computer scientists will make breakthroughs.

Quantum computing threatens to upend many of our assumptions. Shor's algorithm can factor large numbers efficiently on a quantum computer, which would break RSA encryption — the foundation of internet security. Post-quantum cryptography is an active research area, with NIST standardising new algorithms (CRYSTALS-Kyber, CRYSTALS-Dilithium) that resist quantum attacks. Indian researchers at IISER, IISc, and TIFR are contributing to both quantum computing hardware and post-quantum cryptographic algorithms.

AI safety and alignment is another frontier with direct connections to building a neural network from scratch: the complete implementation. As AI systems become more capable, ensuring they behave as intended becomes critical. This involves formal verification (mathematically proving system properties), interpretability (understanding WHY a model makes certain decisions), and robustness (ensuring models do not fail catastrophically on edge cases). The Alignment Research Center and organisations like Anthropic are working on these problems, and Indian researchers are increasingly contributing.

Edge computing and the Internet of Things present new challenges: billions of devices with limited compute and connectivity. India's smart city initiatives and agricultural IoT deployments (soil sensors, weather stations, drone imaging) require algorithms that work with intermittent connectivity, limited battery, and constrained memory. This is fundamentally different from cloud computing and requires rethinking many assumptions.

Finally, the ethical dimensions: facial recognition in public spaces (deployed in several Indian cities), algorithmic bias in loan approvals and hiring, deepfakes in political campaigns, and data sovereignty questions about where Indian citizens' data should be stored. These are not just technical problems — they require CS expertise combined with ethics, law, and social science. The best engineers of the future will be those who understand both the technical implementation AND the societal implications. Your study of building a neural network from scratch: the complete implementation is one step on that path.

Key Vocabulary

Here are important terms from this chapter that you should know:

🏗️ Architecture Challenge

Design the backend for India's election results system. Requirements: 10 lakh (1 million) polling booths reporting simultaneously, results must be accurate (no double-counting), real-time aggregation at constituency and state levels, public dashboard handling 100 million concurrent users, and complete audit trail. Consider: How do you ensure exactly-once delivery of results? (idempotency keys) How do you aggregate in real-time? (stream processing with Apache Flink) How do you serve 100M users? (CDN + read replicas + edge computing) How do you prevent tampering? (digital signatures + blockchain audit log) This is the kind of system design problem that separates senior engineers from staff engineers.

The Frontier

You now have a deep understanding of building a neural network from scratch: the complete implementation — deep enough to apply it in production systems, discuss tradeoffs in system design interviews, and build upon it for research or entrepreneurship. But technology never stands still. The concepts in this chapter will evolve: quantum computing may change our assumptions about complexity, new architectures may replace current paradigms, and AI may automate parts of what engineers do today.

What will NOT change is the ability to think clearly about complex systems, to reason about tradeoffs, to learn quickly and adapt. These meta-skills are what truly matter. India's position in global technology is only growing stronger — from the India Stack to ISRO to the startup ecosystem to open-source contributions. You are part of this story. What you build next is up to you.

Crafted for Class 10–12 • Neural Network Fundamentals • Aligned with NEP 2020 & CBSE Curriculum