arXiv subjects

Shyam Kumar M

Publications and source records attributed to Shyam Kumar M.

3 recordsLinked to original sources

arXiv2026

A Theoretical and Experimental Study of a Novel Adaptive Learning Algorithm

A crucial component of machine learning algorithms is minimizing loss functions with less computational cost and less oscillations. While adaptive learning rate-based optimizers have been widely used for real-world tasks, they do not guarantee convergence, which is why AMSGrad was later introduced to investigate the non-convergence behaviour of Adam. In this paper, popular adaptive optimization methods like Adam and AMSGrad are critically reviewed with an emphasis on their fundamental design concepts. To address limitations of the above mentioned optimizers, a new optimizer variant, C-Adam, is proposed based on the line of sight approach. A theoretical proof for convergence is also provided and the optimizer is validated through a number of real-life based numerical experiments.

cs.LG↗

arXiv2024

A Review of 3D Particle Tracking and Flow Diagnostics Using Digital Holography

Advanced three-dimensional (3D) tracking methods are essential for studying particle dynamics across a wide range of complex systems, including multiphase flows, environmental and atmospheric sciences, colloidal science, biological and medical research, and industrial manufacturing processes. This review provides a comprehensive summary of 3D particle tracking and flow diagnostics using Digital Holography (DH). We begin by introducing the principles of DH, accompanied by a detailed discussion on numerical reconstruction. The review then explores various hardware setups used in DH, including inline, off-axis, and dual or multiple-view configurations, outlining their advantages and limitations. We also delve into different hologram processing methods, categorized into traditional multi-step, inverse, and machine learning-based approaches, providing in-depth insights into their applications for 3D particle tracking and flow diagnostics across multiple studies. The review concludes with a discussion on future prospects, emphasizing the significant role of machine learning in enabling accurate DH-based particle tracking and flow diagnostic techniques across diverse fields, such as manufacturing, environmental monitoring, and biological sciences.

physics.flu-dyn↗

arXiv2023

Visualization and Characterization of Agricultural Sprays Using Machine Learning based Digital Inline Holography

Accurate characterization of agricultural sprays is crucial to predict in field performance of liquid applied crop protection products. Here we introduce a robust and efficient machine learning (ML) based Digital In-line Holography (DIH) to accurately characterize the droplet field for a wide range of agricultural spray nozzles. Compared to non-ML methods, our method enhances accuracy, generalizability, and processing speed. Our approach employs two neural networks: a modified U-Net to obtain the 3D droplet field from the numerically reconstructed optical field, followed by a VGG16 classifier to reduce false positives from the U-Net prediction. The modified U-Net is trained using holograms generated using a single spray nozzle at three spray locations; center, half-span, and the spray edge to create training data with various number densities and droplet size ranges. VGG16 is trained via the minimum intensity projection of the droplet 3D point spread function. Data augmentation is used to increase the efficiency of classification and make the algorithm generalizable for different measurement settings. The model is validated via NIST traceable glass beads and six agricultural spray nozzles representing various spray characteristics. The results demonstrate a high accuracy rate, with over 90% droplet extraction and less than 5% false positives. Compared to traditional spray measurement techniques, our method offers a significant leap forward in spatial resolution and generalizability. In particular, our method can extract the real cumulative volume distribution of the NIST beads, where the laser diffraction is biased towards droplets moving at slower speeds. Additionally, the ML-based DIH enables the estimation of mass and momentum flux at different locations and the calculation of relative velocities of droplet pairs, which are difficult to obtain via conventional techniques.

physics.flu-dyn↗