indiadigital
Home / Code / AI & ML Basics
🤖 BEGINNER → INTERMEDIATE · CLASS 9+ · FREE COURSE

AI & ML Basics

AI और ML की बुनियाद — ब्राउज़र में छोटे-छोटे मॉडल ट्रेन कीजिए — कोई जादू नहीं, बस JavaScript।

Build real machine-learning ideas from scratch in plain JavaScript: a mango-or-lemon classifier, a line that predicts marks from study hours, a perceptron, a fairness check and a next-word predictor for cricket commentary. You need a little JavaScript ([our JS course](/code/javascript/) is enough); never coded? Start with [AI Shiksha](/learn-ai/), our no-code AI course.

Start lesson 1 →Final project
  • 13lessons
  • 26challenges
  • 10hto finish

Syllabus

Unit 1: Data: what models learn from डेटा: मॉडल किससे सीखते हैं3 lessons

  1. What is machine learning?Examples instead of rules; features and labels.1 challenge · 12 min
  2. Summaries and scalingMean, min, max and normalising to 0–1.2 challenges · 14 min
  3. Train and testKeep some data hidden to check the model honestly.1 challenge · 10 min

Unit 2: Classifiers: which group is it? क्लासिफ़ायर: यह किस समूह का है?3 lessons

  1. Distance and nearest neighbourEuclidean distance and the simplest classifier.2 challenges · 14 min
  2. k-nearest neighboursAsk k neighbours and take a vote.2 challenges · 18 min
  3. Is the model any good?Spam features, accuracy and the confusion counts.3 challenges · 18 min

Unit 3: Learning a line एक रेखा सीखना3 lessons

  1. Predicting numbers with a liney = m·x + c and mean squared error.1 challenge · 15 min
  2. Gradient descentWalk downhill on the error, step by step.3 challenges · 22 min
  3. The perceptron: a tiny neuronWeights, bias, a step function and the update rule.2 challenges · 20 min

Unit 4: Overfitting and fairness Overfitting और निष्पक्षता2 lessons

  1. Memorising vs understandingA perfect score on training can still fail the test.2 challenges · 18 min
  2. Bias and fairnessCheck a model group by group, not just overall.2 challenges · 20 min

Unit 5: Language models भाषा मॉडल2 lessons

  1. From text to countsTokens, word counts and a bigram table.3 challenges · 20 min
  2. ★ Predicting the next wordPick the likeliest next word, then build the full predictor.2 challenges · 30 min

Search by task, format or tool name.