---
title: "Learning framework with example"
subject: "Artificial Intelligence"
chapter: 6
key: "semester_6/artificial_intelligence:learning-framework"
source: https://www.bctengineer.com/q/semester_6/artificial_intelligence/learning-framework
site: BCT Engineering
---

# Learning framework with example

Artificial Intelligence · Chapter 6: Machine Learning

- Asked in: 2069 Bhadra, 6 marks
- Answer checked by Claude Opus 5.5 (high effort) on 2026-10-14
- Page: https://www.bctengineer.com/q/semester_6/artificial_intelligence/learning-framework
- Whole chapter: https://www.bctengineer.com/md/semester_6/artificial_intelligence/chapter-6

## Question

Explain the learning framework with suitable example.

## Short answer

A learning system consists of an environment providing examples, a learning element that updates a knowledge base using feedback from a critic, and a performance element that uses the knowledge to act. For a spam filter, the environment supplies labeled emails, the performance element classifies new mail, the critic compares to user corrections, and the learning element adjusts word weights to reduce errors.

## Answer

A **learning system** improves its performance at some task with experience. Rich & Knight and Russell & Norvig describe a general framework made of a few basic components.

#### Block diagram
```text
   Environment
       |  (examples / experience)
       v
 +-----------+   knowledge   +-------------+
 | Learning  |-------------->| Knowledge   |
 | element   |<--------------| base        |
 +-----------+    feedback   +------+------+
       ^                             |
       |                             v
 +-----+------+  evaluation   +-------------+
 |   Critic   |<--------------|Performance  |--> action
 +------------+               | element     |
                              +-------------+
```

#### Components
1. **Environment:** supplies the training examples or situations.
2. **Learning element:** improves the knowledge base using the information from the environment and the critic's feedback.
3. **Knowledge base:** stores what has been learned (rules, weights, decision trees).
4. **Performance element:** uses the knowledge to carry out the task and choose actions.
5. **Critic:** compares performance with a standard and sends feedback to the learning element.
6. **Problem generator (optional):** suggests new experiments to explore.

#### Design issues
- Which component is to be improved, and how it is represented (rules, networks, trees).
- What feedback is available (labelled examples, rewards, none).
- Type of learning: rote, by advice, from examples (induction), analogy, discovery.

#### Example: spam filter
- Environment: incoming emails marked "spam" or "not spam" by users.
- Performance element: classifier that labels a new email.
- Critic: compares the label with the user's correction.
- Learning element: updates word weights/rules to reduce errors.
- Knowledge base: the word weights.

After many emails the filter's accuracy improves, which shows learning.

## Check yourself

The answer's key points as multiple-choice questions, with the correct option and why.

1. What is the role of the critic in the learning framework?
   A. It compares performance with a standard and sends feedback to the learning element
   B. It stores the learned rules
   C. It carries out actions in the environment
   D. It suggests new experiments to explore
   Answer: A. The critic evaluates performance and the learning element uses its feedback. (Key point: Critic role)

2. What does the learning element do?
   A. Chooses the action to perform in the task
   B. Improves the knowledge base using information from the environment and the critic's feedback
   C. Supplies the training examples
   D. Marks the outputs as right or wrong against a standard
   Answer: B. The learning element modifies knowledge; the performance element uses it. (Key point: Learning element)

3. What is the optional problem generator for?
   A. Generating the final answer
   B. Deleting wrong knowledge
   C. Suggesting new experiments so the system explores and gains new experience
   D. Measuring accuracy
   Answer: C. It encourages exploration. (Key point: Problem generator)

4. In the spam-filter example, which part is the knowledge base?
   A. The incoming emails
   B. The user's correction
   C. The classifier that labels a new email
   D. The word weights or rules that classify emails
   Answer: D. The performance element is the classifier; the knowledge base holds the word weights it uses. (Key point: Spam filter example)