Azar Alili
Research Subjects
Azar Alili has a total of 1 co-authors. Here are the top co-authors by number of shared publications:
- Guman Garayev
Pasha Holding, Azerbaijan
5 shared publications
Latest Publications
- Open Access
- 23 October 2024
- Published by MDPI AG in Preprints.org
AI Learning through SUSY Field Theory
We propose a SUSY-inspired loss framework to address the generalizationrobustness trade-off in AI. Combining bosonic and fermionic components, the loss ensures smooth learning and robustness against adversarial attacks. Parallel transport stabilizes weight updates across non-Euclidean loss landscapes. Validation on CIFAR-10 demonstrates stable convergence and enhanced performance under FGSM and PGD attacks, confirming the effectiveness of the proposed approach. Keywords: Supersymmetry, Generalization, Robustness, Adversarial Attacks, Parallel Transport, Loss Optimization, CIFAR-10.
- 23 October 2024
- Published by MDPI AG in Preprints.org
Open AccessAI Learning through SUSY Field Theory
We propose a SUSY-inspired loss framework to address the generalizationrobustness trade-off in AI. Combining bosonic and fermionic components, the loss ensures smooth learning and robustness against adversarial attacks. Parallel transport stabilizes weight updates across non-Euclidean loss landscapes. Validation on CIFAR-10 demonstrates stable convergence and enhanced performance under FGSM and PGD attacks, confirming the effectiveness of the proposed approach. Keywords: Supersymmetry, Generalization, Robustness, Adversarial Attacks, Parallel Transport, Loss Optimization, CIFAR-10.
- Open Access
- 16 October 2024
- Published by MDPI AG in Preprints.org
Stable GANs via Fermi-Dirac Statistics and Ruppeiner Geometry
Generative Adversarial Networks (GANs) are powerful models for generative tasks but face significant challenges such as instability, mode collapse, and inefficient convergence. This paper introduces a theoretical framework leveraging Fermi-Dirac statistics, Ruppeiner geometry, phase transition dynamics, and graph theory to address these issues. The proposed theorems provide a robust mathematical foundation for enhancing GAN stability, diversity, and convergence. Experimental validation on the MNIST dataset demonstrates that the application of these theorems results in improved Inception Scores, reduced Fréchet Inception Distances, and stable training dynamics, as indicated by Ruppeiner curvature analysis. The findings suggest that these theoretical insights offer a comprehensive solution to the persistent challenges in GAN training, paving the way for more reliable and effective generative models.
- 16 October 2024
- Published by MDPI AG in Preprints.org
Open AccessStable GANs via Fermi-Dirac Statistics and Ruppeiner Geometry
Generative Adversarial Networks (GANs) are powerful models for generative tasks but face significant challenges such as instability, mode collapse, and inefficient convergence. This paper introduces a theoretical framework leveraging Fermi-Dirac statistics, Ruppeiner geometry, phase transition dynamics, and graph theory to address these issues. The proposed theorems provide a robust mathematical foundation for enhancing GAN stability, diversity, and convergence. Experimental validation on the MNIST dataset demonstrates that the application of these theorems results in improved Inception Scores, reduced Fréchet Inception Distances, and stable training dynamics, as indicated by Ruppeiner curvature analysis. The findings suggest that these theoretical insights offer a comprehensive solution to the persistent challenges in GAN training, paving the way for more reliable and effective generative models.
- Open Access
- 26 July 2024
- Published by MDPI AG in Preprints.org
Stable GANs via Fermi-Dirac Statistics and Ruppeiner Geometry
Generative Adversarial Networks (GANs) are powerful models for generative tasks but face significant challenges such as instability, mode collapse, and inefficient convergence. This paper introduces a theoretical framework leveraging Fermi-Dirac statistics, Ruppeiner geometry, phase transition dynamics, and graph theory to address these issues. The proposed theorems provide a robust mathematical foundation for enhancing GAN stability, diversity, and convergence. Experimental validation on the MNIST dataset demonstrates that the application of these theorems results in improved Inception Scores, reduced Fr´echet Inception Distances, and stable training dynamics, as indicated by Ruppeiner curvature analysis. The findings suggest that these theoretical insights offer a comprehensive solution to the persistent challenges in GAN training, paving the way for more reliable and effective generative models.
- 26 July 2024
- Published by MDPI AG in Preprints.org
Open AccessStable GANs via Fermi-Dirac Statistics and Ruppeiner Geometry
Generative Adversarial Networks (GANs) are powerful models for generative tasks but face significant challenges such as instability, mode collapse, and inefficient convergence. This paper introduces a theoretical framework leveraging Fermi-Dirac statistics, Ruppeiner geometry, phase transition dynamics, and graph theory to address these issues. The proposed theorems provide a robust mathematical foundation for enhancing GAN stability, diversity, and convergence. Experimental validation on the MNIST dataset demonstrates that the application of these theorems results in improved Inception Scores, reduced Fr´echet Inception Distances, and stable training dynamics, as indicated by Ruppeiner curvature analysis. The findings suggest that these theoretical insights offer a comprehensive solution to the persistent challenges in GAN training, paving the way for more reliable and effective generative models.
- Open Access
- 11 July 2024
- Published by MDPI AG in Preprints.org
On Renormalization Group Based Deep Q-Network
In This paper we introduce the integration of Renormalization Group (RG) methods with Deep Q-Networks (DQNs) to improve reinforcement learning in high-dimensional state spaces. RG methods provide multi-scale analysis, enhancing state representation, learning stability, and exploration. The proposed RG-DQN algorithm uses hierarchical Q-value estimation and multi-scale representations, demonstrating superior performance on synthetic genomic data compared to traditional DQNs.}
- 11 July 2024
- Published by MDPI AG in Preprints.org
Open AccessOn Renormalization Group Based Deep Q-Network
In This paper we introduce the integration of Renormalization Group (RG) methods with Deep Q-Networks (DQNs) to improve reinforcement learning in high-dimensional state spaces. RG methods provide multi-scale analysis, enhancing state representation, learning stability, and exploration. The proposed RG-DQN algorithm uses hierarchical Q-value estimation and multi-scale representations, demonstrating superior performance on synthetic genomic data compared to traditional DQNs.}
- Open Access
- 19 April 2024
- Published by MDPI AG in Preprints.org
QKD Security Through Three-Dimensional Angular Momentum Representation
This paper proposes an enhanced Quantum Key Distribution (QKD) protocol by incorporating a Three-Dimensional Angular Momentum Representation (3D AMR) to address vulnerabilities inherent in traditional two-dimensional polarization states-based systems. The 3D AMR method expands the state space, potentially increasing security against sophisticated quantum attacks. We present a theoretical and computational framework that transitions from the conventional two-dimensional polarization to a three-dimensional angular momentum space, which inherently increases the complexity and security of the quantum states used in QKD.
- 19 April 2024
- Published by MDPI AG in Preprints.org
Open AccessQKD Security Through Three-Dimensional Angular Momentum Representation
This paper proposes an enhanced Quantum Key Distribution (QKD) protocol by incorporating a Three-Dimensional Angular Momentum Representation (3D AMR) to address vulnerabilities inherent in traditional two-dimensional polarization states-based systems. The 3D AMR method expands the state space, potentially increasing security against sophisticated quantum attacks. We present a theoretical and computational framework that transitions from the conventional two-dimensional polarization to a three-dimensional angular momentum space, which inherently increases the complexity and security of the quantum states used in QKD.