# Analysis and Optimization of Convolutional Neural Network Architectures

arXiv: https://arxiv.org/abs/1707.09725 · HTML: https://mlsanity.com/paper/1707.09725/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 31 Jul 2017.

## Built on by (most cited first, 7 of 7)

- [Is Appearance Free Action Recognition Possible?](https://mlsanity.com/paper/2207.06261/) (2022)
- [BotSIM: An End-to-End Bot Simulation Framework for Commercial Task-Oriented Dialog Systems](https://mlsanity.com/paper/2211.11982/) (2022)
- [WSAM: Visual Explanations from Style Augmentation as Adversarial Attacker and Their Influence in Image Classification](https://mlsanity.com/paper/2308.14995/) (2023)
- [Visualizing CoAtNet Predictions for Aiding Melanoma Detection](https://mlsanity.com/paper/2205.10515/) (2022)
- [Introducing Federated Learning into Internet of Things ecosystems -- preliminary considerations](https://mlsanity.com/paper/2207.07700/) (2022)
- [BotSIM: An End-to-End Bot Simulation Toolkit for Commercial Task-Oriented Dialog Systems](https://mlsanity.com/paper/2211.15916/) (2022)
- [Effectiveness of L2 Regularization in Privacy-Preserving Machine Learning](https://mlsanity.com/paper/2412.01541/) (2024)

## Abstract (arXiv)

Convolutional Neural Networks (CNNs) dominate various computer vision tasks since Alex Krizhevsky showed that they can be trained effectively and reduced the top-5 error from 26.2 % to 15.3 % on the ImageNet large scale visual recognition challenge. Many aspects of CNNs are examined in various publications, but literature about the analysis and construction of neural network architectures is rare. This work is one step to close this gap. A comprehensive overview over existing techniques for CNN analysis and topology construction is provided. A novel way to visualize classification errors with confusion matrices was developed. Based on this method, hierarchical classifiers are described and evaluated. Additionally, some results are confirmed and quantified for CIFAR-100. For example, the positive impact of smaller batch sizes, averaging ensembles, data augmentation and test-time transformations on the accuracy. Other results, such as the positive impact of learned color transformation on the test accuracy could not be confirmed. A model which has only one million learned parameters for an input size of 32x32x3 and 100 classes and which beats the state of the art on the benchmark dataset Asirra, GTSRB, HASYv2 and STL-10 was developed.