Match the following: Item Learning type Match Description a. Supervised…
2012
Match the following:
Item | Learning type | Match | Description |
|---|---|---|---|
a. | Supervised learning | 1. | The decision system receives rewards for its action at the end of a sequence of steps. |
b. | Unsupervised learning | 2. | Manual labels of inputs are not used. |
c. | Reinforcement learning | 3. | Manual labels of inputs are used. |
d. | Inductive learning | 4. | System learns by example |
Answer: D. a b c d 3 2 1 4 — ConceptA learning paradigm is defined by the kind of feedback the learner receives while it is being trained, not by the algorithm or the model it uses. Four…
- A.
a
b
c
d
1
2
3
4
- B.
a
b
c
d
2
3
1
4
- C.
a
b
c
d
3
2
4
1
- D.
a
b
c
d
3
2
1
4
Attempted by 28 students.
Show answer & explanation
Correct answer: D
Concept
A learning paradigm is defined by the kind of feedback the learner receives while it is being trained, not by the algorithm or the model it uses. Four standard paradigms are distinguished this way:
Supervised learning — every training input is accompanied by a manually assigned target value, so the learner is corrected against a known answer on each example.
Unsupervised learning — the same inputs arrive with no target values attached, so the learner has to expose structure (groups, densities, compressed representations) from the inputs alone.
Reinforcement learning — there is no target for any single input; an agent acts, and a scalar reward arrives only after a whole sequence of steps, so credit for that reward has to be spread back over the actions that earned it.
Inductive learning — the general principle of inferring a general rule or hypothesis from specific instances that are shown to the learner.
Applying it to this match list
Learning type | Matched description | Why it fits |
|---|---|---|
Supervised learning (a) | Manual labels of inputs are used (3) | Its training set is a set of input-and-target pairs, so the targets have to be labelled by hand before training starts. |
Unsupervised learning (b) | Manual labels of inputs are not used (2) | The labels are withheld by definition; the learner groups or compresses the inputs by similarity alone. |
Reinforcement learning (c) | Rewards are received at the end of a sequence of steps (1) | Feedback is a delayed reward over a trajectory of actions, not a target attached to each input. |
Inductive learning (d) | System learns by example (4) | Induction is exactly the move from specific worked examples to a general hypothesis. |
So the accepted matching is a-3, b-2, c-1, d-4.
Cross-check
The two label descriptions are exact complements, so they must go to supervised and unsupervised learning in that order; handing “manual labels are used” to unsupervised learning contradicts its own definition.
The delayed-reward description mentions a sequence of steps and no per-input target, which is the signature of reinforcement learning; supervised learning would need a target on every input, which that description never provides.
“Learns by example” describes generalisation from instances, which is induction; reading it as reinforcement learning would drop the reward signal that defines reinforcement learning.
Checking all four assignments together, a-3, b-2, c-1, d-4 is the pairing under which every definition holds at once.
Explore the full course: Mppsc Assistant Professor Computer Science Paper 2