Hamza A. A. Gardi

Hamza Gardi, Sebastian F. Walter, Christoph S. Garbe

This paper presents a technique for finding the surface normal of an object from a set of images obtained under different lighting positions. The method presented is based on the principles of Photometric Stereo (PS) combined with Optimum Experimental Design (OED) and Parameter Estimation (PE). Unclear by the approach of photometric stereo, and many models based thereon, is how to position the light sources. So far, this is done by using heuristic approaches this leads to suboptimal and non-data driven positioning of the light sources. But what if the optimal positions of the light sources are calculated for photometric stereo? To this end, in this contribution, the effect of positioning the light sources on the quality of the normal vector for PS is evaluated. Furthermore, a new approach in this direction is derived and formulated. For the calculation of the surface normal of a Lambertian surface, the approach based on calibrated photometric stereo; for the estimation the optimal position of the light sources the approach is premised on parameter estimation and optimum experimental design. The approach is tested using synthetic and real-data. Based on results it can be seen that the surface normal estimated with the new method is more detailed than with conventional methods.

Silas Baumann, Karl A. Busch, Hamza A. A. Gardi

This study investigates the impact of varying sequence lengths on the accuracy of predicting cryptocurrency returns using Artificial Neural Networks (ANNs). Utilizing the Mean Absolute Error (MAE) as a threshold criterion, we aim to enhance prediction accuracy by excluding returns that are smaller than this threshold, thus mitigating errors associated with minor returns. The subsequent evaluation focuses on the accuracy of predicted returns that exceed this threshold. We compare four sequence lengths 168 hours (7 days), 72 hours (3 days), 24 hours, and 12 hours each with a return prediction interval of 2 hours. Our findings reveal the influence of sequence length on prediction accuracy and underscore the potential for optimized sequence configurations in financial forecasting models.

Florian Zager, Hamza A. A. Gardi

Deep neural networks have achieved remarkable success across a range of tasks, however their computational demands often make them unsuitable for deployment on resource-constrained edge devices. This paper explores strategies for compressing and adapting models to enable efficient inference in such environments. We focus on GhostNetV3, a state-of-the-art architecture for mobile applications, and propose GhostNetV3-Small, a modified variant designed to perform better on low-resolution inputs such as those in the CIFAR-10 dataset. In addition to architectural adaptation, we provide a comparative evaluation of knowledge distillation techniques, including traditional knowledge distillation, teacher assistants, and teacher ensembles. Experimental results show that GhostNetV3-Small significantly outperforms the original GhostNetV3 on CIFAR-10, achieving an accuracy of 93.94%. Contrary to expectations, all examined distillation strategies led to reduced accuracy compared to baseline training. These findings indicate that architectural adaptation can be more impactful than distillation in small-scale image classification tasks, highlighting the need for further research on effective model design and advanced distillation techniques for low-resolution domains.

Liam Ressel, Hamza A. A. Gardi

The Engineers' Salary Prediction Challenge requires classifying salary categories into three classes based on tabular data. The job description is represented as a 300-dimensional word embedding incorporated into the tabular features, drastically increasing dimensionality. Additionally, the limited number of training samples makes classification challenging. Linear dimensionality reduction of word embeddings for tabular data classification remains underexplored. This paper studies Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). We show that PCA, with an appropriate subspace dimension, can outperform raw embeddings. LDA without regularization performs poorly due to covariance estimation errors, but applying shrinkage improves performance significantly, even with only two dimensions. We propose Partitioned-LDA, which splits embeddings into equal-sized blocks and performs LDA separately on each, thereby reducing the size of the covariance matrices. Partitioned-LDA outperforms regular LDA and, combined with shrinkage, achieves top-10 accuracy on the competition public leaderboard. This method effectively enhances word embedding performance in tabular data classification with limited training samples.