Refine
Year of publication
- 2023 (2)
Document Type
- Master's Thesis (2) (remove)
Language
- English (2) (remove)
Has Fulltext
- yes (2)
Is part of the Bibliography
- no (2)
Keywords
- AI (1)
- Computer Vision (1)
- Deep learning (1)
- Diffusion Models (1)
- E-Learning (1)
- GAN (1)
- Generative Adversarial Network (1)
- Lernsoftware (1)
- Machine Learning (1)
- Maschinelles Lernen (1)
- Maschinelles Sehen (1)
- Pathologie (1)
- Pathology (1)
- Real-time Collaboration (1)
Institute
- Fakultät IV - Wirtschaft und Informatik (2) (remove)
Pathologists need to identify abnormal changes in tissue. With the developing digitalization, the used tissue slides are stored digitally. This enables pathologists to annotate the region of interest with the support of software tools. PathoLearn is a web-based learning platform explicitly developed for the teacher-student scenario, where the goal is that students learn to identify potential abnormal changes. Artificial intelligence (AI) and machine learning (ML) have become very important in medicine. Many health sectors already utilize AI and ML. This will only increase in the future, also in the field of pathology. Therefore, it is important to teach students the fundamentals and concepts of AI and ML early in their studies. Additionally, creating and training AI generally requires knowledge of programming and technical details. This thesis evaluates how this boundary can be overcome by comparing existing end-to-end AI platforms and teaching tools for AI. It was shown that a visual programming editor offers a fitting abstraction for creating neural networks without programming. This was extended with real-time collaboration to enable students to work in groups. Additionally, an automatic training feature was implemented, removing the necessity to know technical details about training neural networks.
In the last years generative models have gained large public attention due to their high level of quality in generated images. In short, generative models learn a distribution from a finite number of samples and are able then to generate infinite other samples. This can be applied to image data. In the past generative models have not been able to generate realistic images, but nowadays the results are almost indistinguishable from real images.
This work provides a comparative study of three generative models: Variational Autoencoder (VAE), Generative Adversarial Network (GAN) and Diffusion Models (DM). The goal is not to provide a definitive ranking indicating which one of them is the best, but to qualitatively and where possible quantitively decide which model is good with respect to a given criterion. Such criteria include realism, generalization and diversity, sampling, training difficulty, parameter efficiency, interpolating and inpainting capabilities, semantic editing as well as implementation difficulty. After a brief introduction of how each model works on the inside, they are compared against each other. The provided images help to see the differences among the models with respect to each criterion.
To give a short outlook on the results of the comparison of the three models, DMs generate most realistic images. They seem to generalize best and have a high variation among the generated images. However, they are based on an iterative process, which makes them the slowest of the three models in terms of sample generation time. On the other hand, GANs and VAEs generate their samples using one single forward-pass. The images generated by GANs are comparable to the DM and the images from VAEs are blurry, which makes them less desirable in comparison to GANs or DMs. However, both the VAE and the GAN, stand out from the DMs with respect to the interpolations and semantic editing, as they have a latent space, which makes space-walks possible and the changes are not as chaotic as in the case of DMs. Furthermore, concept-vectors can be found, which transform a given image along a given feature while leaving other features and structures mostly unchanged, which is difficult to archive with DMs.