How is DenseNet different from ResNet?
mediumAnswer
- In a DenseNet block, each layer receives the concatenated feature maps of every preceding layer, not just the previous one.
- Massive feature reuse — each layer produces only 'growth rate' k channels (typically 12-32), so the whole network has surprisingly few parameters.
- Trade-offs: high memory usage (long concatenated feature maps), slower on GPU.
- Cool idea but ResNet-derived architectures dominate today.
Check yourself — multiple choice
- Same as ResNet
- Each layer concatenates all previous feature maps — massive reuse, few params, high memory cost
- No convolutions
- DenseNet uses attention
DenseNet: concat all prior features → massive reuse, few params, memory-heavy.
#architectures#computer-vision
Practise Deep Learning
214 interview questions in this topic.