Landslide is a common form of natural disaster in a tropical country such as Malaysia, and its presence is mostly concentrated during the wet season. Human activities have been known to influence landslides as development often causes displacement of the original slope. Western Sarawak is a region where development is currently taking place at a steady rate, with the highest population density in Sarawak settlement is expected to increase in the coming years, increasing the impact of human activity on landslide occurrences in the region. However, the most prevalent representation of human activity in the area has not been discussed. Through a Machine Learning approach of an Artificial Neural Network, the commonly used human activity is represented by Distance from Road (DFR), Normalised Density Vegetation Index (NDVI), and Land Use and Land Cover (LULC) for landslide predictions. It was determined through Garson’s algorithm that the most prevalent human activity variable in the region is best represented by LULC, followed by NDVI, and DFR with a weightage of 0.235, 0.063, and 0.042 respectively.
The Hill Cipher (HC) algorithm is a symmetric technique with a simple structure and fast computations; however, it has a low level of security because both parties need to use and share the same private key through an insecure channel. In this study, a double encryption technique for grayscale images has been proposed by combining the HC with self-invertible (SI) keys. In this method, the encryption and decryption processes are carried out using a key matrix that is SI, where the elements of the key are Integer Matrices with Integer Eigenvalues (IMIE). To assess the quality and resilience of the encryption, a thorough analysis of the grayscale image that has been encrypted twice is performed using metrics such as Entropy, Peak Signal-to-Noise Ratio (PSNR), and Unified Average Changing Intensity (UACI). As a result, the transmission of grayscale images through this new variant of the HC is found to be competitive in terms of security and efficiency compared to several existing encryption techniques, where the proposed method achieves an entropy of 7.9947, PSNR of 8.5230, and UACI of 30.0486\% for the double encrypted image, demonstrating its competitiveness compared to existing techniques.
A polynomial equation with two or more unknowns for which the integer solutions are sought out is called a Diophantine equation. When exponents are introduced into the equation, a simple linear Diophantine equation transforms into a more complex exponential Diophantine equation. This paper concentrates on finding the solution to the exponential Diophantine equation $p^{2m} + {(6r+1)}^n = z^{2}$ where $p, m, r, n, z \in \mathbb{Z}^+$. There is no integral solution to the equation when $p$ is an odd prime, $m \leq 5$, $r \leq 25$ and $n \leq 10$.
Student performance prediction is essential for personalized learning support in online education. This paper proposes DeepFM-AE, a hybrid autoencoder architecture that learns comprehensive student representations for accurate performance score prediction. The model combines factorization machines (FM) to capture educationally meaningful feature interactions with deep neural networks (DNN) to learn complex nonlinear patterns, all within an autoencoder framework that ensures robust representation learning. Evaluated on the Open University Learning Analytics Dataset (OULAD) with 26,727 students, DeepFM-AE achieves a Mean Absolute Error (MAE) of 0.1554, representing a 4.78% improvement over the competitive DeepFM baseline (MAE=0.1632). While this entails a trade-off in other regression metrics, we establish MAE as the primary metric of interest for this educational task. Ablation studies reveal that the reconstruction constraint is crucial for this performance gain. The learned representations show clear semantic structure when visualized, validating the effectiveness of our dual-encoder design for educational data.
Optimization involves finding the optimal solution of the objective function. One of the best optimization methods for solving unconstrained optimization problems is Conjugate gradient (CG) method [1]. CG method is implemented in various applications such as robotic motion, image restoration and regression analysis. CG methods are classified into several categories, including scaled, three-term and hybrid CG methods. This research focuses on the performance of three term CG method (TTCG) under strong Wolfe line search and its applicability in robotic motion control. The performance of four TTCG methods, TTLAMR, TTRMIL, TTSMAR, and TTKMAR coefficients are tested using 15 standard test functions with different initial points and variables ranging from 2 to 10,000. The numerical results are evaluated based on the number of iterations (NOI) and CPU time. These results are plotted using a performance profile to evaluate efficiency and robustness. Numerically, TTLAMR outperforms other methods by solving all test functions and it is followed by TTSMAR (99.38%), TTRMIL (97.84%), and TTKMAR (93.83%). Lastly, TTLAMR is implemented in robotic motion control. It shows that TTLAMR is able to be applied to the motion control of two joint planar robotic manipulators.
