Which Hyperparameters Directly Affect Overfitting in Deep Learning?
Exhibit
Refer to the exhibit. ``` Model: ResNet-50 Batch size: 32 Epochs: 10 Learning rate: 0.001 Optimizer: SGD Data: ImageNet subset Training accuracy: 0.99 Validation accuracy: 0.75 ```
A team trained a ResNet-50 model with the configuration shown. The high training accuracy and lower validation accuracy suggest overfitting. Which change to the training configuration is MOST likely to reduce overfitting?
⚠ Common exam trap
CompTIA often tests the misconception that increasing batch size or reducing epochs directly fixes overfitting, when in fact these changes can harm convergence or underfit, while regularization techniques like dropout are the correct solution.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Add dropout layers after convolutional layers.
Adding dropout layers after convolutional layers is a regularization technique that randomly drops a fraction of neurons during training, which forces the network to learn more robust features and reduces overfitting. This directly addresses the symptom of high training accuracy with lower validation accuracy by preventing the model from relying too heavily on specific neurons.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Reduce number of epochs to 5.
Why it's wrong here
Cutting epochs to five stops training early, which may leave the model underfit rather than genuinely regularised, and the configuration's overfitting would likely persist or become unstable. Epoch reduction suits trimming wasted compute once convergence is reached, not correcting a capacity or regularisation problem.
- ✗
Increase batch size to 64.
Why it's wrong here
Larger batches change gradient averaging and can speed training, but they do not constrain model capacity, so the network can still memorise the training set. Batch size tuning suits throughput and convergence stability; reducing overfitting needs regularisation or capacity limits, which this leaves untouched.
- ✗
Increase learning rate to 0.01.
Why it's wrong here
Raising the learning rate to 0.01 makes updates larger, which typically worsens instability and can amplify overfitting rather than reduce it. Higher learning rates suit escaping plateaus or speeding slow convergence, not constraining a model that already fits the training data too closely.
- ✓
Add dropout layers after convolutional layers.
Why this is correct
Dropout randomly deactivates neurons during training, forcing the network to learn redundant, generalisable features rather than memorising training samples. This directly counteracts the overfitting indicated by the high training accuracy and lower validation accuracy in the stem.
About these practice questions
This AI0-001 question is part of Courseiva's 962-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AI0-001 practice question is part of Courseiva's free CompTIA certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI0-001 exam.