// bridge system

useState()→Model Weights·Event Propagation→Forward Pass·Array.map()→Tensor Operation·React diff→Loss Function L = Σ(y−ŷ)²·transition-duration→Learning Rate η·CSS clamp()→σ(x) Activation·Re-render cycle→Training Epoch·Event bubbling→Backpropagation ∂L/∂w·useCallback→Gradient Caching·Promise.all()→Batch Inference·Redux store→Weight Matrix·DevTools profiler→Loss Landscape·useState()→Model Weights·Event Propagation→Forward Pass·Array.map()→Tensor Operation·React diff→Loss Function L = Σ(y−ŷ)²·transition-duration→Learning Rate η·CSS clamp()→σ(x) Activation·Re-render cycle→Training Epoch·Event bubbling→Backpropagation ∂L/∂w·useCallback→Gradient Caching·Promise.all()→Batch Inference·Redux store→Weight Matrix·DevTools profiler→Loss Landscape·

Commander Mirza
Before we run the deep scan, I want all sensor data normalized. We can't have one sensor's scale drowning out another — each data source must contribute equally.

Dr. Farah
Just like CSS resets create a consistent baseline for styling, normalization creates a consistent baseline for data. Without it, the analysis system will be biased toward sensors with larger numerical ranges.

Learning Objectives

○Understand why normalization is critical for ML
○Apply z-score normalization (zero mean, unit variance)
○Apply min-max normalization (scale to 0-1 range)
○Recognize when to use each normalization strategy

The CSS Reset of Data Science

Frontend

CSS normalize/reset

* { margin: 0; padding: 0; box-sizing: border-box; }

Machine Learning

Feature Normalization

const normalized = data.sub(mean).div(std)

Intuition Bridge

Every frontend developer knows CSS resets. Without them, browsers apply inconsistent default styles. Normalization is the data equivalent — without it, features with large values (like pixel counts in the thousands) dominate features with small values (like percentages from 0-1).

normalization.tstypescript

import * as tf from '@tensorflow/tfjs';

// Sensor data: temperature (°C) and signal strength (dB)
const temperature = tf.tensor([20, 22, 19, 35, 28]);  // range: 19-35
const signal = tf.tensor([0.1, 0.8, 0.3, 0.9, 0.5]);  // range: 0.1-0.9

// Without normalization, temperature dominates (values 20-35 vs 0.1-0.9)

// Z-score normalization: center at 0, spread to ~1
function zNormalize(tensor: tf.Tensor) {
const mean = tensor.mean();
const std = tensor.std();  // standard deviation
return tensor.sub(mean).div(std);
}

const tempNorm = zNormalize(temperature);
// [-0.65, -0.26, -0.85, 1.27, 0.49] — centered around 0

const signalNorm = zNormalize(signal);
// [-1.06, 0.88, -0.50, 1.16, -0.22] — same scale!

// Min-max normalization: squeeze to [0, 1] range
function minMaxNormalize(tensor: tf.Tensor) {
const min = tensor.min();
const max = tensor.max();
return tensor.sub(min).div(max.sub(min));
}

const tempMinMax = minMaxNormalize(temperature);
// [0.06, 0.19, 0.0, 1.0, 0.56] — all between 0 and 1

When to Use Which

Z-score (mean=0, std=1): Use when you need to preserve the spread of your data. Most common for neural network inputs.

Min-max (range 0-1): Use when you need bounded values, like pixel data or probabilities.

Think of it like choosing between normalize.css and a full reset — each has its place depending on what consistency you need.

Challenge

Normalize the Archimedes sensor data using z-score normalization.

⚡

Exercise

Advanced~12 min

Normalize Sensor Data

Apply z-score normalization to a tensor of sensor readings.

const data = tf.tensor([10, 20, 30, 40, 50]);
const normalized = null; // apply z-score normalization

Key Takeaways

✓Normalization is the CSS reset of data — it creates a consistent baseline
✓Z-score: subtract mean, divide by std — centers data at 0
✓Min-max: scales data to a 0-1 range
✓Always normalize before training — unnormalized data causes biased models

Need a hint?

🧭 Guidance

✅ Solution

Data Normalization — Deep Orbit | Tensorcraft

Data Normalization

Context