SEARCH · Search NASA

Results for “network”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 343 records · Page 19

All sources NASA NTRS DOE OSTI

DOE OSTI2024

GAINN: The Galaxy Assembly and Interaction Neural Networks for High-redshift JWST Observations

We present the Galaxy Assembly and Interaction Neural Networks (Gainn), a series of artificial neural networks for predicting the redshift, stellar mass, halo mass, and mass-weighted age of simulated galaxies based on James Webb Space Telescope (JWST) photometry. Our goal is to determine the best neural network for predicting these variables at 11 < z < 15. The parameters of the optimal neural network can then be used to estimate these variables for real, observed galaxies. The inputs of the neural networks are JWST filter magnitudes of a subset of five broadband filters (F150W, F200W, F277W, F356W, and F444W) and two medium-band filters (F162M and F182M). We compare the performance of the neural networks using different combinations of these filters, as well as different activation functions and numbers of layers. The best neural network predicted redshift with a normalized rms error of $0.010^{+0.003}_{-0.001}$, stellar mass with rms = $0.089^{+0.044}_{-0.022}$, halo mass with a mean-squared error of $0.022^{+0.014}_{-0.008}$, and mass-weighted age with rms = $12.466^{+5.065}_{-2.408}$. We also test the performance of Gainn on real data from MACS0647JD, an object observed by JWST. Predictions from Gainn for the first projection of the object (JD1) have normalized bias $\langle$Δz$\rangle$ < 0.00228, which is significantly smaller than found with template-fitting methods. We find that the optimal filter combination is F277W, F356W, F162M, and F200W when considering both theoretical accuracy and observational resources from JWST.

97 MATHEMATICS AND COMPUTING↗

NASA NTRS2000

Neural network recognition of chemical class information in mobility spectra obtained at high temperatures

A minimal neural network was applied to a large library of high-temperature mobility spectra drawn from 16 chemical classes including 154 substances with 2000 spectra at various concentrations. A genetic algorithm was used to create a representative subset of points from the mobility spectrum as input to a cascade-type back-propagation network. This network demonstrated that significant information specific to chemical class was located in the spectral region near the reactant ions. This network failed to generalize the solution to unfamiliar compounds necessitating the use of complete spectra in network processing. An extended back-propagation network classified unfamiliar chemicals by functional group with a mean for average values of 0.83 without sulfides and 0.79 with sulfides. Further experiments confirmed that chemical class information was resident in the spectral region near the reactant ions. Deconvolution of spectra demonstrated the presence of ions, merged with the reactant ion peaks that originated from introduced samples. The ability of the neural network to generalize the solution to unfamiliar compounds suggests that these ions are distinct and class specific.

NASA Discipline Environmental Health↗

NASA NTRS2011

Enhanced Communication Network Solution for Positive Train Control Implementation

The commuter and freight railroad industry is required to implement Positive Train Control (PTC) by 2015 (2012 for Metrolink), a challenging network communications problem. This paper will discuss present technologies developed by the National Aeronautics and Space Administration (NASA) to overcome comparable communication challenges encountered in deep space mission operations. PTC will be based on a new cellular wireless packet Internet Protocol (IP) network. However, ensuring reliability in such a network is difficult due to the "dead zones" and transient disruptions we commonly experience when we lose calls in commercial cellular networks. These disruptions make it difficult to meet PTC s stringent reliability (99.999%) and safety requirements, deployment deadlines, and budget. This paper proposes innovative solutions based on space-proven technologies that would help meet these challenges: (1) Delay Tolerant Networking (DTN) technology, designed for use in resource-constrained, embedded systems and currently in use on the International Space Station, enables reliable communication over networks in which timely data acknowledgments might not be possible due to transient link outages. (2) Policy-Based Management (PBM) provides dynamic management capabilities, allowing vital data to be exchanged selectively (with priority) by utilizing alternative communication resources. The resulting network may help railroads implement PTC faster, cheaper, and more reliably.

Policy-Based Management (PBM)↗

NASA NTRS2014

A Framework for Dimensioning VDL-2 Air-Ground Networks

This paper describes a framework developed at MITRE for dimensioning a Very High Frequency (VHF) Digital Link Mode 2 (VDL-2) Air-to-Ground network. This framework was developed to support the FAA's Data Communications (Data Comm) program by providing estimates of expected capacity required for the air-ground network services that will support Controller-Pilot-Data-Link Communications (CPDLC), as well as the spectrum needed to operate the system at required levels of performance. The Data Comm program is part of the FAA's NextGen initiative to implement advanced communication capabilities in the National Airspace System (NAS). The first component of the framework is the radio-frequency (RF) coverage design for the network ground stations. Then we proceed to describe the approach used to assess the aircraft geographical distribution and the data traffic demand expected in the network. The next step is the resource allocation utilizing optimization algorithms developed in MITRE's Spectrum ProspectorTM tool to propose frequency assignment solutions, and a NASA-developed VDL-2 tool to perform simulations and determine whether a proposed plan meets the desired performance requirements. The framework presented is capable of providing quantitative estimates of multiple variables related to the air-ground network, in order to satisfy established coverage, capacity and latency performance requirements. Outputs include: coverage provided at different altitudes; data capacity required in the network, aggregated or on a per ground station basis; spectrum (pool of frequencies) needed for the system to meet a target performance; optimized frequency plan for a given scenario; expected performance given spectrum available; and, estimates of throughput distributions for a given scenario. We conclude with a discussion aimed at providing insight into the tradeoffs and challenges identified with respect to radio resource management for VDL-2 air-ground networks.

Aviation↗

NASA NTRS2011

Intelligent Wireless Sensor Networks for System Health Monitoring

Wireless sensor networks (WSN) based on the IEEE 802.15.4 Personal Area Network (PAN) standard are finding increasing use in the home automation and emerging smart energy markets. The network and application layers, based on the ZigBee 2007 Standard, provide a convenient framework for component-based software that supports customer solutions from multiple vendors. WSNs provide the inherent fault tolerance required for aerospace applications. The Discovery and Systems Health Group at NASA Ames Research Center has been developing WSN technology for use aboard aircraft and spacecraft for System Health Monitoring of structures and life support systems using funding from the NASA Engineering and Safety Center and Exploration Technology Development and Demonstration Program. This technology provides key advantages for low-power, low-cost ancillary sensing systems particularly across pressure interfaces and in areas where it is difficult to run wires. Intelligence for sensor networks could be defined as the capability of forming dynamic sensor networks, allowing high-level application software to identify and address any sensor that joined the network without the use of any centralized database defining the sensors characteristics. The IEEE 1451 Standard defines methods for the management of intelligent sensor systems and the IEEE 1451.4 section defines Transducer Electronic Datasheets (TEDS), which contain key information regarding the sensor characteristics such as name, description, serial number, calibration information and user information such as location within a vehicle. By locating the TEDS information on the wireless sensor itself and enabling access to this information base from the application software, the application can identify the sensor unambiguously and interpret and present the sensor data stream without reference to any other information. The application software is able to read the status of each sensor module, responding in real-time to changes of PAN configuration, providing the appropriate response for maintaining overall sensor system function, even when sensor modules fail or the WSN is reconfigured. The session will present the architecture and technical feasibility of creating fault-tolerant WSNs for aerospace applications based on our application of the technology to a Structural Health Monitoring testbed. The interim results of WSN development and testing including our software architecture for intelligent sensor management will be discussed in the context of the specific tradeoffs required for effective use. Initial certification measurement techniques and test results gauging WSN susceptibility to Radio Frequency interference are introduced as key challenges for technology adoption. A candidate Developmental and Flight Instrumentation implementation using intelligent sensor networks for wind tunnel and flight tests is developed as a guide to understanding key aspects of the aerospace vehicle design, test and operations life cycle.

networks↗

NASA NTRS2022

Impacts of Climate Change on Global Food Trade Networks

Countries' reliance on global food trade networks implies that regionally different climate change impacts on crop yields will be transmitted across borders. This redistribution constitutes a significant challenge for climate adaptation planning and may affect how countries engage in cooperative action. This paper investigates the long-term (2070–2099) potential impacts of climate change on global food trade networks of three key crops: wheat, rice and maize. We propose a simple network model to project how climate change impacts on crop yields may be translated into changes in trade. Combining trade and climate impact data, our analysis proceeds in three steps. First, we use network community detection to analyse how the concentration of global production in present-day trade communities may become disrupted with climate change impacts. Second, we study how countries may change their network position following climate change impacts. Third, we study the total climate-induced change in production plus import within trade communities. Results indicate that the stability of food trade network structures compared to today differs between crops, and that countries' maize trade is least stable under climate change impacts. Results also project that threats to global food security may depend on production change in a few major global producers, and whether trade communities can balance production and import loss in some vulnerable countries. Overall, our model contributes a baseline analysis of cross-border climate impacts on food trade networks.

climate change↗

NASA NTRSDate not supplied

Emulated Spacecraft Communication Testbed for Evaluating Cognitive Networking Technology

The ability to emulate the full space protocol stack is an essential aspect required to evaluate and mature cognitive communication capabilities. The interaction between the physical layer and network layers is key to developing network optimizations for a dynamic and complex environment. We present a laboratory testbed for the evaluation of cognitive radio and networking techniques applied to space communications. The testbed is a high fidelity, flight-like hardware testbed consisting of software-defined radios, channel emulators, modems, and orbital analysis and scheduling software. The testbed uses RF links with signal quality, propagation delay, and Doppler effects driven by orbital mechanics simulations of emulated spacecraft. Our framework enables control of link bidirectionality, data rates, and interference sources. In addition to hardware radio nodes, the testbed can incorporate virtualized emulated nodes for larger and more challenging network scenarios. Our approach to a cognitive communication system uses delay tolerant networking (DTN) to mitigate the challenges of the space environment. While many DTN networks use only preplanned schedules, our system uses User-Initiated Service (UIS) to dynamically schedule service providers. Software-defined radio allows the system to adapt to a variety of service providers. Integration of DTN, UIS, and software-defined radio technologies provides a framework for the implementation of a cognitive communication system. This paper describes the testbed capabilities, network emulation approach, component integration, and initial end-to-end testing results.

cognitive radio↗

NASA NTRSDate not supplied

Large-Scale Space Network Simulator for Performance-Optimized DTNs

High-rate Delay Tolerant Networking (HDTN) is a performance-optimized Delay Tolerant Networking (DTN) implementation that can provide multigigabit per second data rates in disruptive and high-latency space networks. Routing, especially in large-scale space networks, remains challenging due to network topologies that evolve over time. This paper presents a simulation tool that enables HDTN implementation testing at accelerated speeds, which is key for routing in large-scale space networks.

High data-rate Delay Tolerant Networking↗

NASA NTRSDate not supplied

NASA Space Communications and Navigation: One Network Evolution

The NASA Space Communications and Navigation (SCaN) Program is responsible for providing the essential connectivity to robotic and human space explorers. The missions relying on SCaN range from suborbital and balloon missions to those traveling beyond the edge of the solar system. The demands for communications and navigation services enabled by SCaN (and its affiliated partners) are projected to increase and outpace the current network capacity. At the same time, the Agency finds itself surrounded by a burgeoning commercial space marketplace, technological advancement, and other government agencies that share common interests in space resiliency, robustness, and performance. As a result, SCaN has begun pivoting toward commercial services and collaborating with partners to close capacity and capability gaps. Given these growing demands of the Agency there is increasing need for multi-network solutions. Future mission concepts will rely on both government and commercial capabilities, both Near Space Network capacity and Deep Space Network capacity. Integrating these diverse support services together from a technical, programmatic and implementation standpoint will be key to meet the growing needs of the future. To accomplish this, a more substantive shift is required, and SCaN is reshaping itself to be a customer-centric, service-oriented, high-performance leader in the space communications community. This paper outlines the SCaN One Team, One Mission, One Network approach, and provides a vision for future mission community experience that includes streamlined mission commitment interfaces and clear processes, dynamic network scheduling and load balancing, and higher efficiency data transport and delivery through the integration of cloud infrastructure and services.

Near Space Network↗

DOE OSTI2026

Input specific neural networks

Neural networks have emerged as powerful tools for mapping between inputs and outputs. However, their black-box nature limits the ability to encode or impose specific structural relationships between inputs and outputs. Many scientific and engineering problems, such as constitutive modeling in solid mechanics, require networks that can enforce convexity, monotonicity, or other structural constraints to ensure physical consistency. Here, we introduce the Input Specific Neural Network (ISNN), a new architecture that enables multiple, distinct constraints to be imposed on different input subsets for scalar-valued outputs. This framework unifies convex, monotone–convex, monotone, and arbitrary mappings within a single network for the first time. Two ISNN architectures with analytical first- and second-order derivatives are developed. We demonstrate the performance on synthetic toy problems, inverse problems in isotropic hyperelasticity, and finite element simulations. ISNNs achieve improved extrapolation behavior, require fewer invariant inputs than standard input convex networks for polyconvex potentials, and enable significant computational savings via manual differentiation. We also show how ISNNs can be used to learn structural relationships between inputs and outputs via a binary gating mechanism. Particularly, ISNNs are employed to model a homogenized anisotropic free energy potential in a decoupled multiscale setting. The network learns whether or not the potential should be modeled as polyconvex and retains only the relevant layers while using the minimum number of inputs. ISNNs provide a flexible foundation for embedding structural priors into neural networks, enhancing both interpretability and stability. They are broadly applicable across computational mechanics and other scientific domains requiring constrained functional relationships.

Jadoon, Asghar A. [Univ. of Texas, Austin, TX (Uni↗

DOE OSTI2025

Data-Efficient Strategies for Probabilistic Voltage Envelopes under Network Contingencies

This work presents an efficient data-driven method to construct probabilistic voltage envelopes (PVE) using power flow learning in grids with network contingencies. First, a network-aware Gaussian process (GP) termed Vertex-Degree Kernel (VDK-GP), developed in prior work, is used to estimate voltage–power functions for a few network configurations. The paper introduces a novel multi-task vertex degree kernel (MT-VDK) that amalgamates the learned VDK-GPs to determine power flows for unseen networks, with a significant reduction in the computational complexity and hyperparameter requirements compared to alternate approaches. Simulations on the IEEE 30-Bus network demonstrate the retention and transfer of power flow knowledge in both N-1 and N-2 contingency scenarios. The MT-VDK-GP approach achieves over 50 % reduction in mean prediction error for novel N-1 contingency network configurations in low training data regimes (50–250 samples) over VDK-GP. Additionally, MT-VDK-GP outperforms a hyper-parameter based transfer learning approach in over 75 % of N-2 contingency network structures, even without historical N-2 outage data. Furthermore, the proposed method demonstrates the ability to achieve PVEs using sixteen times fewer power flow solutions compared to Monte-Carlo sampling-based methods.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2024

Synergistic Combination of Living Ring-Opening Metathesis Polymerization and Atom Transfer Radical Polymerization to Synthesize Structurally Tailored and Engineered Macromolecular Networks

Structurally tailored and engineered macromolecular (STEM) networks are attractive materials for soft robotics, stretchable electronics, tissue engineering, and 3D printing due to their tunable properties. To date, STEM networks have been synthesized by atom transfer radical polymerization (ATRP) or the combination of reversible addition–fragmentation chain-transfer (RAFT) polymerization and ATRP. RAFT polymerization could have limited selectivity with ATRP inimer sites that can participate in radical-transfer processes. On the other hand, living ring-opening metathesis polymerization (ROMP) can produce a polymeric network with latent ATRP initiator sites in high selectivity. Herein, for the first time, we report the syntheses of STEM zero-generation (STEM-0) networks using a monomer, a cross-linker, and an ATRP/ROMP inimer via living ROMP, followed by their modification using a second monomer via ATRP to synthesize STEM first-generation (STEM-1) networks. The mechanical property and swelling capacity analyses of these networks were carried out. A change in mechanical properties and swelling capacity of these networks was observed due to their structural modification.

Absorption↗

DOE OSTI2025

Structural and Dynamical Insights into the Formation Process of a Cross-Linked Polymer Network in Acrylic Adhesives During Thermal Curing

Many modern adhesives, sealants, and coatings rely on the controlled transition from a liquid to a solid state by forming a three-dimensional cross-linked polymer network, often referred to as curing. The curing process, which is initiated by the mixing of reactive components or an external trigger, defines the structure of a network and further controls the final mechanical properties of cured materials. However, the curing mechanism is not fully understood yet due to the lack of experimental tools capable of directly probing the structure and dynamics of a network over relevant time- and length scales. Here, in this paper, we report the curing process of a commercial two-component methyl methacrylate (MMA) adhesive using in operando X-ray photon correlation spectroscopy (XPCS), a method that closely simulates the target manufacturing environment of the adhesive. The results are then integrated with those obtained by rheology, differential scanning calorimetry (DSC), and transmission electron microscopy to establish the structure–dynamics–process–property relationship. The XPCS results identify four distinct stages in the curing process after the mixing, extrusion, and deposition of the acrylic adhesive: (i) At a cure time (or “aging time”, t age ) of less than 1 min, nanodomains of polymerized MMA are formed within a liquid monomeric MMA matrix. The average size is several nm and remains constant over t age , while the dynamics of the nanodomains are slowed down with t age due to an increase in the viscosity of the MMA matrix. (ii) After t age > 1 min, the size of the nanodomains increases with t age until the gel point (= 6.3 min after mixing as determined by rheology). The dynamics of the nanodomains also increase due to the heat generated by the exothermic reaction. (iii) At the gel point, the nanodomains begin to interconnect each other, resulting in a network structure with a characteristic length of about 100 nm. This characteristic network size does not change for the rest of the curing process up to t age = 500 min. The dynamics of the network structure, however, show a rapid slowing down with t age up to t age ≈ 12 min, corresponding to the onset of vitrification (as determined by rheology). (iv) At t age > 12 min, when the DSC and rheology data can no longer provide meaningful information, the XPCS data show a further slowing down of the network dynamics associated with vitrification. Our results provide rich and complex insights into the physics and material design of thermosets in commercially relevant processes, which are essential for future industrial applications.

36 MATERIALS SCIENCE↗

DOE OSTI2026

Cooperative effect of local active stresses on the macroscopic contractility of elastic fiber networks

The collective action of actively contractile units embedded in elastic biopolymer networks plays a crucial role in regulating the network's macroscopic mechanical response. Here, in this study, we investigate how the macroscopic boundary stress in model elastic fiber networks depends on the number and nature of embedded contractile units, each exerting an isotropic force dipole, as well as on the bending stiffness of fibers. We find that the macroscopic stress increases nonlinearly with the number of dipoles due to mutual stiffening of initially soft, bending-dominated networks. Using effective medium theory, we relate this enhanced contractility to an increase in the effective average network coordination number due to constraints imposed by the force dipoles. By comparing three distinct force dipole models that differ in their local structures, we demonstrate that the specific manner in which an active unit constrains the network strongly influences the onset and nature of the stiffening transition. Our results highlight that not only the quantity but also the local geometry of force-generating units critically determines the macroscopic mechanical behavior. This framework provides a physical basis for understanding how biological systems—such as molecular motors in the cytoskeleton, or adherent cells in the extracellular matrix—can modulate network-scale nonlinear elastic properties through local tuning of active force-generating units.

Biological and medical sciences↗

DOE OSTI2024

Abiotic Stress Reorganizes Rhizosphere and Endosphere Network Structure of Sorghum bicolor

Sorghum bicolor is a promising bioenergy feedstock with high biomass production and unusual tolerance for stresses, such as water and nutrient limitation. Although the membership of the sorghum microbiome in response to stress has been explored, relatively little is known about how microbe–microbe networks change under water- or nutrient-limited conditions. This is important because network changes can indicate impacts on the functionality and stability of microbial communities. We performed network-based analysis on the core bacterial and archaeal community of an agronomically promising high biomass bioenergy genotype, Grassl, grown under nitrogen and water stress. Stress caused relatively minor changes in bacterial abundances within soil, rhizosphere, and endosphere communities but led to significant changes in bacterial network structure and modularity. We found a complete reorganization of network roles in all plant compartments, as well as an increase in the modularity and proportion of positive associations, which potentially could represent coexistence and cooperation in the sorghum bacterial/archaeal community under stress. Although stressors are often believed to be destabilizing, we found stressed networks were as or more stable than non-stressed networks, likely due to their redundancy and compartmentalization. Together, these findings support the idea that both sorghum and its bacterial/archaeal community can be resilient to future environmental stressors.

09 BIOMASS FUELS↗

DOE OSTI2025

Real-World Cyber Security Demonstration for Networked Electric Drives

In this article, we present the design and implementation of a cyber-physical security testbed for networked electric drive systems, aimed at conducting real-world security demonstrations. To our knowledge, this is one of the first security testbeds for networked electric drives, seamlessly integrating the domains of power electronics and computer science, and cybersecurity. By doing so, the testbed offers a comprehensive platform to explore and understand the intricate and often complex interactions between cyber and physical systems. The core of our testbed consists of four electric machine drives, meticulously configured to emulate small-scale but realistic information technology (IT) and operational technology (OT) networks. This setup both provides a controlled environment for simulating a wide array of cyber-attacks, and mirrors potential real-world attack scenarios with a high degree of fidelity. The testbed serves as an invaluable resource for the study of cyber-physical security, offering a practical and dynamic platform for testing and validating cybersecurity measures in the context of networked electric drive systems. As a concrete example of the testbed's capabilities, we have developed and implemented a Python-based script designed to execute step-stone attacks over a wireless local area network (WLAN). This script leverages a sequence of target IP addresses, simulating a real-world attack vector that could be exploited by adversaries. To counteract such threats, we demonstrate the efficacy of our developed cyber-attack detection algorithms, which are integral to our testbed's security framework. Furthermore, the testbed incorporates a real-time visualization system using InfluxDB and Grafana, providing a dynamic and interactive representation of networked electric drives and their associated security monitoring mechanisms. This visualization component not only enhances the testbed's usability but also offers insightful, real-time data for researchers and practitioners, thereby facilitating a deeper understanding of cyber-physical security dynamics in networked electric drive systems.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2026

Comprehensive framework for assessing and optimizing existing research networks

Conservation, monitoring, and research networks, or collections of ecological research sites unified under a common mission of data collection or a research mission, are essential infrastructure for understanding large landscapes. However, most networks developed opportunistically over decades rather than through systematic design, creating potential limitations in the ability to address conservation challenges across entire regions. We developed a framework to evaluate how well an existing research network represents the environmental conditions its members study and devised an approach to rank sites of priority for strategic expansion. Our approach measures performance through environmental representativeness, geographic coverage, and adequacy for scientific inference and thus optimizes limited monitoring resources to maximize scientific impact. We demonstrated this approach with the U.S. Department of Agriculture (USDA) Forest Service Experimental Forests and Ranges Network (EFRN), a 79‐site network across the United States that grew opportunistically over a century. At the national scale, the network effectively captured high‐biomass forests important for carbon cycle research; 82% of forest biomass was in well‐represented areas. Some areas in Texas, Florida, the Rocky Mountains, and the West Coast had no relevant EFRN sites, which limits the ability to make regional inferences. A fundamental challenge for the EFRN was that sites improving regional extent coverage sometimes provided minimal national benefits, which can create conflicts between local and global priorities. Adding the highest‐ranked candidate site provided a relevant site for 17% of currently poorly represented 1‐km pixel cells nationally, but regional and national site rankings varied considerably due to nested spatial inference. This framework provides quantitative tools for strategic infrastructure decision‐making, ensures that limited monitoring resources maximize conservation impact, and can be applied broadly to address the widespread challenge of optimizing conservation and monitoring networks worldwide.

additional site↗

DOE OSTI2026

Measuring the Energy Consumption and Efficiency of Deep Neural Networks: An Empirical Analysis and Design Recommendations

Addressing the "Red-AI" trend of rising energy consumption by large-scale neural networks, this study investigates the measured energy consumption of training various fully connected neural network architectures. We introduce the BUTTER-E dataset, an augmentation to the BUTTER Empirical Deep Learning dataset, containing energy consumption and performance data from 41,129 individual experimental runs spanning 30,582 distinct configurations: 13 datasets, 20 sizes (trainable parameters), 8 "shapes", and 14 depths on both CPUs and GPUs using node-level watt-meters. This dataset reveals the complex relationship between dataset size, network structure, and energy use. Our analysis uncovers a surprising, hardware-mediated non-linear relationship between energy efficiency and network design, challenging the assumption that reducing the number of parameters or FLOPs is the best way to achieve greater energy efficiency. We propose a straightforward and effective energy model that accounts for network size, computing, and memory hierarchy. Highlighting the need for cache-considerate algorithm development, we suggest a codesign approach to energy efficient network, algorithm, and hardware design. This work contributes to the fields of sustainable computing and Green AI, offering practical guidance for creating more energy-efficient neural networks and promoting sustainable AI.

97 MATHEMATICS AND COMPUTING↗