Applied AICollège Unica
3.4SpecializationSession 3 · Deep Learning & Applied Machine Learning

Supervised learning algorithms

20classes60htotalTH2 · PR2 · PNW2weighting per class

Class-by-class breakdown

20 classes

TH = theory · PR = practical · PNW = personal work. These are ministry weighting codes (not hours) used to split each class's minutes. Class length = course hours ÷ class count; the three time-boxes below always sum to that length.

0%

Quiz progress

0 of 0 classes attempted

First 2 classes free with a referral 19 more with enrollment180 min per classTheory 60mPractical 60mPersonal Work 60m
1

Where Supervised Learning Fits in an Agentic-AI Stack

FreeTH2PR2PNW2

An AI Integration Manager knows GenAI agents aren't the whole solution — supervised models route queries, score relevance, and guard outputs.

Theory60m
  • Supervised learning: learning from labeled examples
  • Where classical ML supports GenAI: routing, classification, evaluation
  • Why agents still need fast, cheap, deterministic models
Practical60m
  • Map one agent workflow to where a supervised model would help
  • List 3 agent tasks better suited to classical ML than an LLM
Personal Work60m
  • Write a short note on one supervised model that would support a sample agent
  • Micro-task: submit your note
2

Labels and Features — Building Training Data for Agent Support Models

LockedTH2PR2PNW2

Locked class preview

A Data/Business Analyst who builds clean labeled data directly controls model quality — no labels, no supervised model

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
3

Train/Validation/Test Splits — Evaluating Honestly

LockedTH2PR2PNW2

Locked class preview

A Project Manager who trusts a model's accuracy without a proper test split is trusting a number that may be meaningless

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
4

Linear Regression — Predicting a Number (and Agent Latency)

LockedTH2PR2PNW2

Locked class preview

A Data/Business Analyst who predicts agent response latency with regression can warn stakeholders before SLAs break

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
5

Logistic Regression — Binary Classification for Agent Guardrails

LockedTH2PR2PNW2

Locked class preview

An AI Integration Manager who builds a toxicity/hallucination classifier with logistic regression adds a fast guardrail around the agent

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
6

Intent Classification — Routing User Queries to the Right Agent

LockedTH2PR2PNW2

Locked class preview

A Project Manager who routes queries to the right specialist agent with a classifier saves cost and improves answer quality

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
7

Decision Trees — Interpretable Models Stakeholders Trust

LockedTH2PR2PNW2

Locked class preview

A Data/Business Analyst who ships an interpretable decision tree can explain exactly why the model routed a query the way it did

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
8

Random Forests — Stronger, More Robust Classifiers

LockedTH2PR2PNW2

Locked class preview

An AI Integration Manager who uses a random forest gets a more accurate, less overfit classifier for retrieval-relevance scoring

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
9

Evaluation Metrics — Accuracy, Precision, Recall, F1

LockedTH2PR2PNW2

Locked class preview

A Data/Business Analyst who reports only accuracy hides the model's real failures — precision and recall reveal what matters for agents

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
10

Classification for Hallucination and Toxicity Detection

LockedTH2PR2PNW2

Locked class preview

An AI Integration Manager who classifies agent outputs as hallucinated or toxic adds a safety layer users can trust

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
11

Scoring Retrieval Relevance — Supervised RAG Evaluation

LockedTH2PR2PNW2

Locked class preview

A Data/Business Analyst who trains a relevance scorer can automatically evaluate RAG retrieval quality at scale

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
12

Feature Engineering with Embeddings — ML on GenAI Representations

LockedTH2PR2PNW2

Locked class preview

A Project Manager who uses embeddings as features gives classical ML models a powerful signal derived from GenAI

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
13

Cross-Validation — Trusting Your Evaluation

LockedTH2PR2PNW2

Locked class preview

A Data/Business Analyst who uses cross-validation avoids the trap of a lucky test split — the result is more trustworthy

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
14

Class Imbalance — When Most Agent Answers Are 'Fine'

LockedTH2PR2PNW2

Locked class preview

An AI Integration Manager who ignores class imbalance ships a model that says 'fine' to everything — imbalance breaks guardrails

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
15

Model Selection and Hyperparameter Tuning

LockedTH2PR2PNW2

Locked class preview

A Data/Business Analyst who tunes hyperparameters squeezes real performance from a model — defaults rarely win

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
16

Deploying a Supervised Model as an Agent Microservice

LockedTH2PR2PNW2

Locked class preview

A Project Manager who deploys the model as a service lets the agent call it as a tool — classical ML wrapped as an agent tool

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
17

Mini-Project — Build a Supervised Model That Supports an Agent

LockedTH2PR2PNW2

Locked class preview

This capstone rehearsal proves a Data/Business Analyst can build a supervised model that genuinely improves an agent workflow

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
18

Monitoring Model Drift — When the Agent's Support Model Goes Stale

LockedTH2PR2PNW2

Locked class preview

An AI Integration Manager who monitors drift catches the moment the classifier starts misrouting queries as user behavior changes

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
19

Naive Bayes and Text Classification for Agent Inputs

LockedTH2PR2PNW2

Locked class preview

A Data/Business Analyst who uses Naive Bayes gets a fast, cheap text classifier for routing or tagging agent inputs at scale

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions
20

Present and Reflect — Your Agent-Support Model

LockedTH2PR2PNW2

Locked class preview

A Data/Business Analyst presenting a supervised model proves classical ML still has a vital role supporting GenAI agents

Enroll to unlock every class in this course — the full agenda, the AI tutor, and the quiz.

Talk to admissions

Unlock all 20 classes

Get the full agenda, AI tutor, and quiz for every class in 3.4 · Supervised learning algorithms.

Apply via admissions

Online checkout not yet live — contact admissions to enroll.

Previous course

3.3 · Data transformation and manipulation

Next course

3.5 · Reinforcement learning algorithms

Back to Session 3