Georgi Gaydadjiev

Rafaël Brandt, Daan Raatjens, Georgi Gaydadjiev

The rationale behind a deep learning model's output is often difficult to understand by humans. EXplainable AI (XAI) aims at solving this by developing methods that improve interpretability and explainability of machine learning models. Reliable evaluation metrics are needed to assess and compare different XAI methods. We propose a novel evaluation approach for benchmarking state-of-the-art XAI attribution methods. Our proposal consists of a synthetic classification model accompanied by its derived ground truth explanations allowing high precision representation of input nodes contributions. We also propose new high-fidelity metrics to quantify the difference between explanations of the investigated XAI method and those derived from the synthetic model. Our metrics allow assessment of explanations in terms of precision and recall separately. Also, we propose metrics to independently evaluate negative or positive contributions of inputs. Our proposal provides deeper insights into XAI methods output. We investigate our proposal by constructing a synthetic convolutional image classification model and benchmarking several widely used XAI attribution methods using our evaluation approach. We compare our results with established prior XAI evaluation metrics. By deriving the ground truth directly from the constructed model in our method, we ensure the absence of bias, e.g., subjective either based on the training set. Our experimental results provide novel insights into the performance of Guided-Backprop and Smoothgrad XAI methods that are widely in use. Both have good precision and recall scores among positively contributing pixels (0.7, 0.76 and 0.7, 0.77, respectively), but poor precision scores among negatively contributing pixels (0.44, 0.61 and 0.47, 0.75, resp.). The recall scores in the latter case remain close. We show that our metrics are among the fastest in terms of execution time.

Bahareh Ghari, Ali Tourani, Asadollah Shahbahrami et al.

Pedestrian detection remains a critical problem in various domains, such as computer vision, surveillance, and autonomous driving. In particular, accurate and instant detection of pedestrians in low-light conditions and reduced visibility is of utmost importance for autonomous vehicles to prevent accidents and save lives. This paper aims to comprehensively survey various pedestrian detection approaches, baselines, and datasets that specifically target low-light conditions. The survey discusses the challenges faced in detecting pedestrians at night and explores state-of-the-art methodologies proposed in recent years to address this issue. These methodologies encompass a diverse range, including deep learning-based, feature-based, and hybrid approaches, which have shown promising results in enhancing pedestrian detection performance under challenging lighting conditions. Furthermore, the paper highlights current research directions in the field and identifies potential solutions that merit further investigation by researchers. By thoroughly examining pedestrian detection techniques in low-light conditions, this survey seeks to contribute to the advancement of safer and more reliable autonomous driving systems and other applications related to pedestrian safety. Accordingly, most of the current approaches in the field use deep learning-based image fusion methodologies (i.e., early, halfway, and late fusion) for accurate and reliable pedestrian detection. Moreover, the majority of the works in the field (approximately 48%) have been evaluated on the KAIST dataset, while the real-world video feeds recorded by authors have been used in less than six percent of the works.

Nima Esmi, Maryam Nezhad-Moghaddam, Fatemeh Borhani et al.

With the significant expansion of the context window in Large Language Models (LLMs), these models are theoretically capable of processing millions of tokens in a single pass. However, research indicates a significant gap between this theoretical capacity and the practical ability of models to robustly utilize information within long contexts, especially in tasks that require a comprehensive understanding of numerous details. This paper evaluates the performance of four state-of-the-art models (Grok-4, GPT-4, Gemini 2.5, and GPT-5) on long short-context tasks. For this purpose, three datasets were used: two supplementary datasets for retrieving culinary recipes and math problems, and a primary dataset of 20K social media posts for depression detection. The results show that as the input volume on the social media dataset exceeds 5K posts (70K tokens), the performance of all models degrades significantly, with accuracy dropping to around 50-53% for 20K posts. Notably, in the GPT-5 model, despite the sharp decline in accuracy, its precision remained high at approximately 95%, a feature that could be highly effective for sensitive applications like depression detection. This research also indicates that the "lost in the middle" problem has been largely resolved in newer models. This study emphasizes the gap between the theoretical capacity and the actual performance of models on complex, high-volume data tasks and highlights the importance of metrics beyond simple accuracy for practical applications.