Adaptive Physics-Guided Neural Network
Authors:
David Shulman,
Itai Dattner
Abstract:
This paper introduces an adaptive physics-guided neural network (APGNN) framework for predicting quality attributes from image data by integrating physical laws into deep learning models. The APGNN adaptively balances data-driven and physics-informed predictions, enhancing model accuracy and robustness across different environments. Our approach is evaluated on both synthetic and real-world datase…
▽ More
This paper introduces an adaptive physics-guided neural network (APGNN) framework for predicting quality attributes from image data by integrating physical laws into deep learning models. The APGNN adaptively balances data-driven and physics-informed predictions, enhancing model accuracy and robustness across different environments. Our approach is evaluated on both synthetic and real-world datasets, with comparisons to conventional data-driven models such as ResNet. For the synthetic data, 2D domains were generated using three distinct governing equations: the diffusion equation, the advection-diffusion equation, and the Poisson equation. Non-linear transformations were applied to these domains to emulate complex physical processes in image form.
In real-world experiments, the APGNN consistently demonstrated superior performance in the diverse thermal image dataset. On the cucumber dataset, characterized by low material diversity and controlled conditions, APGNN and PGNN showed similar performance, both outperforming the data-driven ResNet. However, in the more complex thermal dataset, particularly for outdoor materials with higher environmental variability, APGNN outperformed both PGNN and ResNet by dynamically adjusting its reliance on physics-based versus data-driven insights. This adaptability allowed APGNN to maintain robust performance across structured, low-variability settings and more heterogeneous scenarios. These findings underscore the potential of adaptive physics-guided learning to integrate physical constraints effectively, even in challenging real-world contexts with diverse environmental conditions.
△ Less
Submitted 15 November, 2024;
originally announced November 2024.
A Guided FP-growth algorithm for multitude-targeted mining of big data
Authors:
Lior Shabtay,
Rami Yaari,
Itai Dattner
Abstract:
In this paper we present the GFP-growth (Guided FP-growth) algorithm, a novel method for multitude-targeted mining: finding the count of a given large list of itemsets in large data. The GFP-growth algorithm is designed to focus on the specific multitude itemsets of interest and optimizes the time and memory costs. We prove that the GFP-growth algorithm yields the exact frequency-counts for the re…
▽ More
In this paper we present the GFP-growth (Guided FP-growth) algorithm, a novel method for multitude-targeted mining: finding the count of a given large list of itemsets in large data. The GFP-growth algorithm is designed to focus on the specific multitude itemsets of interest and optimizes the time and memory costs. We prove that the GFP-growth algorithm yields the exact frequency-counts for the required itemsets. We show that for a number of different problems, a solution can be devised which takes advantage of the efficient implementation of multitude-targeted mining for boosting the performance. In particular, we study in detail the problem of generating the minority-class rules from imbalanced data, a scenario that appears in many real-life domains such as medical applications, failure prediction, network and cyber security, and maintenance. We develop the Minority-Report Algorithm that uses the GFP-growth for boosting performance. We prove some theoretical properties of the Minority-Report Algorithm and demonstrate its performance gain using simulations and real data.
△ Less
Submitted 4 July, 2018; v1 submitted 18 March, 2018;
originally announced March 2018.
