Multi-label classification in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding Multi-label Classification

Which statement best describes multi-label classification?

A Each instance is assigned exactly one label from multiple classes.

B Each instance can be assigned multiple labels simultaneously.

C It is a regression problem predicting continuous values.

D Labels are predicted sequentially, one after another.

💡 Hint

Think about whether an example can belong to more than one category at the same time.

Multi-label classification means each example can have multiple labels at once, unlike single-label classification where only one label is assigned.

❓ Model Choice

intermediate

Choosing a Model for Multi-label Classification

You want to build a multi-label classifier. Which model architecture is most suitable?

A A neural network with multiple output neurons using sigmoid activation.

B A decision tree that outputs one label per instance.

C A linear regression model predicting continuous values.

D A neural network with a single output neuron using softmax activation.

💡 Hint

Consider how the output layer should handle multiple labels independently.

Multi-label classification requires independent probabilities for each label, so sigmoid activation on multiple outputs is appropriate.

❓ Metrics

advanced

Evaluating Multi-label Classification Performance

Which metric is best suited to evaluate a multi-label classification model?

A Mean Squared Error between predicted and true labels.

B Accuracy calculated as (correct predictions / total samples).

C Hamming Loss measuring the fraction of wrong labels to total labels.

D R-squared score measuring variance explained by the model.

💡 Hint

Think about a metric that counts label-wise errors in multi-label settings.

Hamming Loss counts how many labels are incorrectly predicted per instance, making it suitable for multi-label tasks.

❓ Predict Output

advanced

Output of Multi-label Prediction Code

What is the output of the following Python code snippet?

ML Python

import numpy as np
from sklearn.preprocessing import MultiLabelBinarizer

labels = [['cat', 'dog'], ['dog'], ['mouse', 'cat'], []]
mlb = MultiLabelBinarizer(classes=['cat', 'dog', 'mouse'])
encoded = mlb.fit_transform(labels)
print(encoded.tolist())

A [[1, 1, 1], [0, 1, 0], [1, 0, 1], [0, 0, 0]]

B [[1, 0, 1], [0, 1, 0], [1, 1, 0], [0, 0, 0]]

C [[0, 1, 1], [1, 0, 0], [0, 1, 1], [0, 0, 0]]

D [[1, 1, 0], [0, 1, 0], [1, 0, 1], [0, 0, 0]]

💡 Hint

Check the order of classes and how MultiLabelBinarizer encodes labels.

MultiLabelBinarizer encodes each label as a binary vector in the order of classes given.

🔧 Debug

expert

Debugging Multi-label Classification Training Issue

Given this training code snippet for a multi-label classifier, what is the main cause of the error?

import torch
import torch.nn as nn

model = nn.Sequential(
    nn.Linear(10, 3),
    nn.Softmax(dim=1)
)

criterion = nn.BCEWithLogitsLoss()
inputs = torch.randn(5, 10)
targets = torch.randint(0, 2, (5, 3)).float()
outputs = model(inputs)
loss = criterion(outputs, targets)

A Using Softmax activation before BCEWithLogitsLoss causes incorrect input range.

B Targets should be integers, not floats, for BCEWithLogitsLoss.

C The input tensor shape does not match the model output shape.

D BCEWithLogitsLoss requires outputs to be probabilities between 0 and 1.

💡 Hint

Consider what BCEWithLogitsLoss expects as input and what Softmax outputs.

BCEWithLogitsLoss expects raw logits, so applying Softmax before it is incorrect and causes training issues.