An official website of the United States government
Many papers have addressed the problem of learning the behavior (i.e., the local interaction function at each node) of a networked system through active queries, assuming that the network topology is known. We address the problem of inferring both the network topology and the behavior of such a system through active queries. Our results are for systems where the state of each node is from {0, 1} and the local functions are Boolean. We present inference algorithms under both batch and adaptive query models for dynamical systems with symmetric local functions. These algorithms show that the structure and behavior of such dynamical systems can be learnt using only a polynomial number of queries. Further, we establish a lower bound on the number of queries needed to learn such dynamical systems. We also present experimental results obtained by running our algorithms on synthetic and real-world networks.
more »
« less
The Challenges of Inferring Organic Function from Structure and its Emulation in Biomechanics and Biomimetics
The discipline called biomimetics attempts to create synthetic systems that model the behavior and functions of biological systems. At a very basic level, this approach incorporates a philosophy grounded in modeling either the behavior or properties of organic systems based on inferences of structure–function relationships. This approach has achieved extraordinary scientific accomplishments, both in fabricating new materials and structures. However, it is also prone to misstep because (1) many organic structures are multifunctional that have reconciled conflicting individual functional demands (rather than maximize the performance of any one task) over evolutionary time, and (2) some structures are ancillary or entirely superfluous to the functions their associated systems perform. The important point is that we must typically infer function from structure, and that is not always easy to do even when behavioral characteristics are available (e.g., the delivery of venom by the fangs of a snake, or cytoplasmic toxins by the leaf hairs of the stinging nettle). Here, we discuss both of these potential pitfalls by comparing and contrasting how engineered and organic systems are operationally analyzed. We also address the challenges that emerge when an organic system is modeled and suggest a few methods to evaluate the validity of models in general.
more »
« less
- Award ID(s):
- 1718075
- PAR ID:
- 10295502
- Date Published:
- Journal Name:
- Biomimetics
- Volume:
- 6
- Issue:
- 21
- ISSN:
- 1059-0153
- Page Range / eLocation ID:
- 1-12
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
A rigorous physics-informed learning methodology is proposed for predictions of wave solutions and band structures in electronic and optical superlattice structures. The methodology is enabled by proper orthogonal decomposition (POD) and Galerkin projection of the wave equation. The approach solves the wave eigenvalue problem in POD space constituted by a finite set of basis functions (or POD modes). The POD ensures that the generated modes are optimized and tailored to the parametric variations of the system. Galerkin projection however enforces physical principles in the methodology to further enhance the accuracy and efficiency of the developed model. It has been demonstrated that the POD-Galerkin methodology offers an approach with a reduction in degrees of freedom by 4 orders of magnitude, compared to direct numerical simulation (DNS). A computing speedup near 15,000 times over DNS can be achieved with high accuracy for either of the superlattice structures if only the band structure is calculated without the wave solution. If both wave function solution and band structure are needed, a 2-order reduction in computational time can be achieved with a relative least square error (LSE) near 1%. When the training is incomplete or the desired eigenstates are slightly beyond the training bounds, an accurate prediction with an LSE near 1%-2% still can be reached if more POD modes are included. This reveals its remarkable learning ability to reach correct solutions with the guidance of physical principles provided by Galerkin projection.more » « less
-
Abstract The relationship between structure and function is a major constituent of the rules of life. Structures and functions occur across all levels of biological organization. Current efforts to integrate conceptual frameworks and approaches to address new and old questions promise to allow a more holistic and robust understanding of how different biological functions are achieved across levels of biological organization. Here, we provide unifying and generalizable definitions of both structure and function that can be applied across all levels of biological organization. However, we find differences in the nature of structures at the organismal level and below as compared to above the level of the organism. We term these intrinsic and emergent structures, respectively. Intrinsic structures are directly under selection, contributing to the overall performance (fitness) of the individual organism. Emergent structures involve interactions among aggregations of organisms and are not directly under selection. Given this distinction, we argue that while the functions of many intrinsic structures remain unknown, functions of emergent structures are the result of the aggregate of processes of individual organisms. We then provide a detailed and unified framework of the structure–function relationship for intrinsic structures to explore how their unknown functions can be defined. We provide examples of how these scalable definitions applied to intrinsic structures provide a framework to address questions on structure–function relationships that can be approached simultaneously from all subdisciplines of biology. We propose that this will produce a more holistic and robust understanding of how different biological functions are achieved across levels of biological organization.more » « less
-
In this work, we propose a balanced multicomponent and multilayer neural network (MMNN) structure to accurately and efficiently approximate functions with complex features in terms of both degrees of freedom and computational cost. The main idea is inspired by a multicomponent approach in which each component can be effectively approximated by a single-layer network, combined with a multilayer decomposition strategy to capture the complexity of the target function. Although MMNNs can be viewed as a simple modification of fully connected neural networks (FCNNs) or multilayer perceptrons (MLPs) by introducing balanced multicomponent structures, they achieve a significant reduction in training parameters, a much more efficient training process, and improved accuracy compared to FCNNs or MLPs. Extensive numerical experiments demonstrate the effectiveness of MMNNs in approximating highly oscillatory functions and their ability to automatically adapt to localized features. Our code and implementations are available at GitHub.more » « less
-
Glasses prepared by physical vapor deposition (PVD) are anisotropic, and the average molecular orientation can be varied significantly by controlling the deposition conditions. While previous work has characterized the average structure of thick PVD glasses, most experiments are not sensitive to the structure near an underlying substrate or interface. Given the profound influence of the substrate on the growth of crystalline or liquid crystalline materials, an underlying substrate might be expected to substantially alter the structure of a PVD glass, and this near-interface structure is important for the function of organic electronic devices prepared by PVD, such as organic light-emitting diodes. To study molecular packing near buried organic–organic interfaces, we prepare superlattice structures (stacks of 5- or 10-nm layers) of organic semiconductors, Alq3 (Tris-(8-hydroxyquinoline)aluminum) and DSA-Ph (1,4-di-[4-(N,N-diphenyl)amino]styrylbenzene), using PVD. Superlattice structures significantly increase the fraction of the films near buried interfaces, thereby allowing for quantitative characterization of interfacial packing. Remarkably, both X-ray scattering and spectroscopic ellipsometry indicate that the substrate exerts a negligible influence on PVD glass structure. Thus, the surface equilibration mechanism previously advanced for thick films can successfully describe PVD glass structure even within the first monolayer of deposition on an organic substrate.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- The DOI is not currently available.
