Search NASA⌕ Search

SEARCH · Search NASA

Results for “box modeling”

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.

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

At least 361 records · Page 20
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Unusual modifications of protein biomarkers expressed by plasmid, prophage, and bacterial host of pathogenic Escherichia coli identified using top‐down proteomic analysis

Rationale Pathogenic bacteria often carry prophage (bacterial viruses) and plasmids (small circular pieces of DNA) that may harbor toxin, antibacterial, and antibiotic resistance genes. Proteomic characterization of pathogenic bacteria should include the identification of host proteins and proteins produced by prophage and plasmid genomes. Methods Protein biomarkers of two strains of Shiga toxin–producingEscherichia coli(STEC) were identified using antibiotic induction, matrix‐assisted laser desorption/ionization tandem time‐of‐flight (MALDI‐TOF‐TOF) tandem mass spectrometry (MS/MS) with post‐source decay (PSD), top‐down proteomic (TDP) analysis, and plasmid sequencing. Alphafold2 was also used to compare predicted in silico structures of the identified proteins to prominent fragment ions generated using MS/MS‐PSD. Strain samples were also analyzed with and without chemical reduction treatment to detect the attachment of pendant groups bound by thioester or disulfide bonds. Results Shiga toxin was detected and/or identified in both STEC strains. For the first time, we also identified the osmotically inducible protein (OsmY) whose sequence unexpectedly had two forms: a full and a truncated sequence. The truncated OsmY terminates in the middle of an α‐helix as determined by Alphafold2. A plasmid‐encoded colicin immunity protein was also identified with and without attachment of an unidentified cysteine‐bound pendant group (~307 Da). Plasmid sequencing confirmed top‐down analysis and the identification of a promoter upstream of the immunity gene that is activated by antibiotic induction, that is, SOS box. Conclusions TDP analysis, coupled with other techniques (e.g., antibiotic induction, chemical reduction, plasmid sequencing, and in silico protein modeling), is a powerful tool to identify proteins (and their modifications), including prophage‐ and plasmid‐encoded proteins, produced by pathogenic microorganisms.

Biochemistry & Molecular Biology↗
DOE OSTI2023

Increased Interpretability for Model-Driven Deception: MARS LDRD Project

Machine learning has been proposed as a solution to several cybersecurity solutions and one of the most promising applications is for digital twins for intrusion detection and driving deceptive defense. However, machine learning techniques often result in a black-box function that is difficult for end users to interpret which for deception limits their ability to effectively define decoys. In this report, an approach to validate the equations learned are accurate is provided and demonstrated. Following, begins the process of addressing this issue for a model-driven deception technology that produces equations representing the physical process controlled by operation technology devices. This research was performed by applying subject matter expert context to machine learned models.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Measurement of DC Arc-Flash Incident Energy in Large-Scale Photovoltaic Plants: A Basis for Standardization

The deployment of high-power dc equipment is increasing in solar photovoltaic (PV) plants, but very few studies have quantified dc arc-flash risks. Currently, PV plant owners and operators rely on theoretical, simplified models, such as those in NFPA-70E and other publications, for the assessment of risk associated with dc arc-flash. Here, this article presents an overview of arc-flash risks in a PV system based on a series of field experiments based on IEEE-1584 in two large-scale ground-mounted PV plants. The experiments include various high-power dc equipment of a PV plant, such as central inverters, combiner boxes, recombiner boxes, string inverters, and multiple configurations of electrodes in a 20-in calibration cube. The study reveals that none of the available dc arc-flash models are applicable for a PV plant. This work is an important first step toward developing an improved model that more accurately assesses dc arc-flash risk in a PV plant.

14 SOLAR ENERGY↗
DOE OSTI2023

Explicit model predictive control through robust optimization

A strategy that calculates an explicit state feedback policy to regulate constrained uncertain discrete‐time uncertain linear systems is presented. We consider uncertain processes, affected by box‐bounded multiplicative uncertainty as well as bounded additive uncertainty with linear state and inputs constraints. The proposed method includes (i) the calculation of a terminal set constraint and (ii) the robust reformulation of state constraints in the prediction horizon. These features allow the derivation of the desired policy by solving a single multiparametric quadratic programming problem that guarantees feasible operation in the presence of uncertainty. Additionally, we employ variable and constraint elimination approaches to enhance the computational performance of the strategy. We demonstrate the steps and benefits of these developments with a numerical example and a chemical engineering case study.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2024

AEOLUS: Advances in Experimental Design, Optimal Control, and Learning for Uncertain Complex Systems

Sustained advances in the mathematics of modeling and simulation have resulted in the capability today for routine simulation of a number of large scale complex DOE-relevant systems. As remarkable as this capability for solving the so-called forward problem is, it is typically only the first step-an inner loop within an outer loop that explores the simulation model's parameter space and decision space to characterize uncertainty in the model's predictions, learn unknown model parameters from data, design the most informative experiments, determine optimal control strategies, and create optimal designs. Broadly, what unifies all of these outer loop problems is that they are, in one form or another, optimization problems over parameter/control/design space that are constrained by complex uncertain models. To fully realize the power of scientific simulation as a basis for scientific discovery, technological innovation, and rational decision-making, it is imperative to move beyond simulation to tackle the outer loop of optimization for learning from data, experimental design, and control with complex uncertain models. When the models under consideration are large-scale and complex, and when the optimization variable and uncertain parameter spaces are high (or infinite) dimensional, this constitutes a grand challenge of the highest order, and is intractable with conventional methods. To overcome these challenges, the AEOLUS Center was established to develop a unified mathematical, computational, and statistical framework for (1) Learning predictive models from complex data via Bayesian inference and optimization, and (2) Optimizing experiments, processes, and designs using the resulting uncertain models. These problems are intractable with conventional methods, for several reasons: (1) The simulation problems that govern the inner loops of the optimization problems are expensive to execute (due to severe nonlinearity, heterogeneity, multiphysics/multiscale coupling); (2) The optimization variable and uncertain parameter spaces are high dimensional, often stemming from discretizations of infinite dimensional fields such as initial conditions, sources, or material properties. We argue that the key to overcoming these challenges is to develop new mathematical, computational, and statistical methods that exploit the structure of the Bayesian inference and optimization problems mediated by their underlying complex uncertain models. This structure includes the regularity, sparsity, geometry, low intrinsic dimensionality, and multifidelity nature of the maps from uncertain parameter/optimization variable spaces to the specific objectives targeted: Bayesian inference, optimal experimental design, and optimal control design. Black box methods developed as generic tools are incapable of exploiting this structure. To be successful, we must create, integrate, and cross-fertilize ideas across multiple areas of applied math--including approximation theory, Bayesian inference, data science, experimental design, information theory, machine learning, model reduction, optimal control theory, parallel algorithms, PDE-constrained optimization, randomized algorithms, stochastic optimization, and uncertainty quantification--all while exploiting the structure of the problems at hand. With this goal in mind, we have marshaled a team of leading authorities in these areas. While the methods we develop will be broadly applicable across a wide spectrum of DOE problems in which experiments inform models and the systems those models describe must be optimized under uncertainty, we have chosen a specific area, advanced manufacturing and materials, to drive our work. AMM is characterized by complex models across multiple scales, and is a rich source of challenging problems in inference, experimental design, and optimal control, requiring multifaceted and integrated advances in applied mathematics. As such, AMM serves as an excellent vehicle to motivate and demonstrate the advances in applied mathematics developed by our center.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Component-In-Grout Model Implementation for the E-Area Low-Level Waste Facility's Performance Assessment

The component-in-grout (CIG) disposal units are below grade earthen trenches that contain grout encapsulated waste components. Components disposed of within the CIG segments consist of large radioactively contaminated equipment and smaller waste forms (e.g., B-25 boxes and SeaLand containers) to fill the space around and above the large equipment. In the 2022 revision of the E-Area Low-Level Waste Facility’s (ELLWF) performance assessment, groundwater radionuclide contaminant transport through the vadose zone will be modeled using the PORFLOW software package for nine existing CIG segments located within the Slit Trench (ST) 23 footprint - no additional CIG segments are planned for at this time. The nine CIG segments have been placed within two of the nominally 20-foot-wide by 656-foot-long ST segments and are surrounded by no less than 1 foot of grout on all sides and up to 4 feet of backfill material. The two most recently placed CIG segments have an additional reinforced concrete mat to improve structural stability.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2021

The SvFUL2 transcription factor is required for inflorescence determinacy and timely flowering in Setaria viridis

Inflorescence architecture in cereal crops directly impacts yield potential through regulation of seed number and harvesting ability. Extensive architectural diversity found in inflorescences of grass species is due to spatial and temporal activity and determinacy of meristems, which control the number and arrangement of branches and flowers, and underlie plasticity. Timing of the floral transition is also intimately associated with inflorescence development and architecture, yet little is known about the intersecting pathways and how they are rewired during development. Here, we show that a single mutation in a gene encoding an AP1/FUL-like MADS-box transcription factor significantly delays flowering time and disrupts multiple levels of meristem determinacy in panicles of the C 4 model panicoid grass, Setaria viridis. Previous reports of AP1/FUL-like genes in cereals have revealed extensive functional redundancy, and in panicoid grasses, no associated inflorescence phenotypes have been described. In S. viridis, perturbation of SvFul2, both through chemical mutagenesis and gene editing, converted a normally determinate inflorescence habit to an indeterminate one, and also repressed determinacy in axillary branch and floral meristems. Our analysis of gene networks connected to disruption of SvFul2 identified regulatory hubs at the intersection of floral transition and inflorescence determinacy, providing insights into the optimization of cereal crop architecture.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2020

Black-Box Optimization for Design of Concentrating Solar Power and Photovoltaic Hybrid Systems with Optimal Dispatch Decisions

The hybridization of concentrating solar power (CSP) and photovoltaics (PV) can enable dispatchable renewable electricity generation at a lower price than current stand-alone CSP systems. However, designing a CSP-PV hybrid system can be challenging because of the many degrees of freedom in design that affect the internal and external system interactions and trade-offs. We develop a methodology to determine optimal designs for CSP-PV hybrids by implementing NLopt's derivative-free, or “black-box,” algorithms around pre-existing CSP-PV hybrid simulation software that utilizes the National Renewable Energy Laboratory’s System Advisor Model (SAM); we then employ a dispatch optimization model to determine operational decisions that maximize a plant’s profits. We present optimal designs for CSP-PV hybrid systems dispatching against four time-of-delivery (ToD) pricing structures. NLopt’s algorithms can improve the base case design’s power purchase agreement (PPA) price by 15% to 21%, depending on the ToD pricing structure. In addition, we present the resulting optimal CSP-PV hybrid design’s annual performance metrics, which tend to have capacity factors between 50% and 62%, but are able to generate electricity during the year’s highest-valued periods about 90% of the time. Lastly, we investigate the trade-offs between capacity factor and PPA price using Pareto fronts and demonstrate that, for some ToD pricing structures, the system capacity factor can increase by 20% but at the expense of a 2% increase in PPA price.

black-box↗
DOE OSTI2023

A small-box approach to the local crystal structure of Y 3 NbO 7

The local crystal structure of Y 3 NbO 7 was probed using neutron total scattering. Five different structural models were fit to the experimental pair distribution function of this material up to 15 Å. Fits of defect fluorite (Fm$\overline{3}$m), pyrochlore (Fd$\overline{3}$m), and orthorhombic C weberite models (C222 1 ) were attempted based on prior structural investigations of Y 3 NbO 7 and related fluorite-based materials. Orthorhombic P weberite (P222 1 ) and monoclinic weberite models (P112 1 ) were derived via symmetry reduction of the orthorhombic C weberite structure. All models except monoclinic weberite failed to reproduce the experimental pair distribution function. Misfits were clearly visible for peaks arising from atom pairs within the first coordination shell (defect fluorite and pyrochlore) and beyond (orthorhombic C and P weberite). By contrast, monoclinic weberite (γ ≈ 91.8°) provided an excellent fit, with no major misfits observed. This unit cell features distorted YO 8 dodecahedra, YO 7 monocapped octahedra, and NbO 6 octahedra as building blocks. Comparison between the topologies of the cationic and anionic substructures in monoclinic weberite and its parent orthorhombic C structure revealed displacements of the metal atoms along the [100] direction and a less compact spatial distribution of oxygen atoms along the a axis. In conclusion, the monoclinic weberite model presented herein may serve as a starting point to revisit structural modeling of fluorite-based materials with challenging local structures.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2021

FOXM1-activated SIRT4 inhibits NF-κB signaling and NLRP3 inflammasome to alleviate kidney injury and podocyte pyroptosis in diabetic nephropathy

Highlights: • FOXM1 is downregulated in kidney tissues of patients with DN. • Overexpression of FOXM1 alleviates kidney injury in DN modeled mice. • Overexpression of FOXM1 reduces pyroptosis of HG-treated podocytes. • FOXM1 activates SIRT4 to inhibit the NFκB pathway and the NLRP3 inflammasome. • Downregulation of SIRT4 blocks the protective roles of FOXM1 in podocytes and mice. Forkhead box M1 (FOXM1) has been reported to play a protective role against acute kidney injury by driving tubular regeneration. This study aims to probe the function of FOXM1 in diabetic nephropathy (DN) and the molecules involved. FOXM1 was poorly expressed in DN-diseased kidney tissues. A murine model of DN was established, and podocytes cells (MPC5) were treated with high-glucose (HG) for in vitro studies. FOXM1 overexpression improved kidney function and reduced pathological changes in mice, and it increased the expression of the podocyte marker Nephrin in kidney tissues. In vitro, FOXM1 increased viability and reduced pyroptosis of the HG-treated MPC5 cells, and it elevated the expression of the podocyte marker Nephrin whereas reduced the expression of pyroptosis-related NLRP3 inflammasome and cleaved caspase 1. FOXM1 bound to the promoter of sirtuin 4 (SIRT4) to induce transcriptional activation. Downregulation of SIRT4 blocked the protective roles of FOXM1 both in vivo and in vitro. Phosphorylation of nuclear factor-kappa B (NF-κB) in HG-treated cells was suppressed by FOXM1 but restored after SIRT4 inhibition. In conclusion, this study suggested that FOXM1 transcriptionally activates SIRT4 and inhibits NF-κB signaling and the NLRP3 inflammasome to alleviate kidney injury and podocyte pyroptosis in DN.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2020

Modeling the smoky troposphere of the southeast Atlantic: a comparison to ORACLES airborne observations from September of 2016

The southeast Atlantic is home to well-defined smoke outflow from Africa coinciding vertically with extensive marine boundary-layer cloud decks, both reaching their climatological maxima in spatial extent around September. A framework is put forth for evaluating the performance of a range of global and regional aerosol models against observations made during the NASA ORACLES (ObseRvations of Aerosols above CLouds and their intEractionS) airborne mission in September 2016. The sparse airborne observations are first aggregated into 2o grid boxes and into three vertical layers: the cloud-topped marine boundary layer (MBL), the layer from cloud top to 3 km, and the 3-6 km layer. Aerosol extensive properties simulated for the entire study region for all September suggest that the 2016 ORACLES observations are reasonably representative of the regional monthly average, with systematic deviations of 30% or less. All six models typically place the bottom of the smoke layer at lower altitudes than do the airborne lidar observations by 300-1400 m, whereas model aerosol top heights are within 0-500 m of the observations. All but one of the models that report carbonaceous aerosol masses underestimate the ratio of particulate extinction to the masses, a proxy for mass extinction efficiency, in 3-6 km. Notable findings on individual models include that WRF-CAM5 predicts the mass of black carbon and organic aerosols with minor (~10% or less) biases. GEOS-5 overestimates the carbonaceous particle masses in the MBL by a factor of 3-6. Extinction coefficients in the free troposphere (FT) and above-cloud aerosol optical depth (ACAOD) are 10-30% lower in WRF-CAM5, 30-50% lower in GEOS-5, 10-40% higher in GEOS-Chem, 10-20% higher in EAM-E3SM except for the practically unbiased 3-6 km extinction, and 20-70% lower in the Unified Model, than the airborne in situ, lidar and sunphotometer measurements. ALADIN-Climate also underestimates the ACAOD, by 30%. GEOS-5 and GEOS-Chem predict carbon monoxide in the MBL with small (10% or less) negative biases, despite their overestimates of carbonaceous aerosol masses. Overall, this study highlights a new approach to utilizing airborne aerosol measurements for model diagnosis.

Shinozuka, Yohei↗
DOE OSTI2022

How to see hidden patterns in metamaterials with interpretable machine learning

Machine learning models can assist with metamaterials design by approximating computationally expensive simulators or solving inverse design problems. However, past work has usually relied on black box deep neural networks, whose reasoning processes are opaque and require enormous datasets that are expensive to obtain. Here, in this work, we develop two novel machine learning approaches to metamaterials discovery that have neither of these disadvantages. These approaches, called shape-frequency features and unit-cell templates, can discover 2D metamaterials with user-specified frequency band gaps. Our approaches provide logical rule-based conditions on metamaterial unit-cells that allow for interpretable reasoning processes, and generalize well across design spaces of different resolutions. The templates also provide design flexibility where users can almost freely design the fine resolution features of a unit-cell without affecting the user’s desired band gap.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Probabilistic error estimation for non-intrusive reduced models learned from data of systems governed by linear parabolic partial differential equations

This work derives a residual-based a posteriori error estimator for reduced models learned with non-intrusive model reduction from data of high-dimensional systems governed by linear parabolic partial differential equations with control inputs. It is shown that quantities that are necessary for the error estimator can be either obtained exactly as the solutions of least-squares problems in a non-intrusive way from data such as initial conditions, control inputs, and high-dimensional solution trajectories or bounded in a probabilistic sense. Here, the computational procedure follows an offline/online decomposition. In the offline (training) phase, the high-dimensional system is judiciously solved in a black-box fashion to generate data and to set up the error estimator. In the online phase, the estimator is used to bound the error of the reduced-model predictions for new initial conditions and new control inputs without recourse to the high-dimensional system. Numerical results demonstrate the workflow of the proposed approach from data to reduced models to certified predictions.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Automated Energy-Dispersive X-ray Spectroscopy Analysis for Multi-Modal Few-Shot Learning

Scanning transmission electron microscopy (STEM) is a powerful tool that allows for the atomic-scale analysis of a materials’ structure, chemistry, and defect domains (Akers et al. 2021). The current generation of microscopes generate vast amounts of data, surpassing the limits of effective manual analysis traditionally performed by domain experts (Spurgeon et al. 2021). While recent strides in machine learning have significantly enhanced the processing of large and intricate datasets acquired through electron microscopy, the prevalent use of proprietary software packages for initial data collection poses a challenge. In many cases, these software packages act as a ‘black box’, constraining user functionality and hindering the output of data in a format that is conducive to seamless integration into machine learning models. This work addresses these challenges by adapting HyperSpy, an open-source Python library, for the analysis and quantification of raw energy dispersive spectroscopy (EDS) data acquired through STEM. The modified HyperSpy code successfully facilitates user-defined segmentation of the data, enabling the integration of atomic %, weight %, and raw EDS spectra for each segmented region into an existing few-shot machine learning model. While initial results reveal discrepancies in quantified atomic and weight percentages when compared to proprietary software, ongoing efforts aim to rectify this issue by refining the fit of the HyperSpy model to the EDS spectra. Overall, this research underscores the potential of open-source tools like HyperSpy to enhance the accessibility of analytical tools, fostering a transparent and user-friendly environment for seamlessly incorporating electron microscopy data into machine learning models.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Designing for Supply and Return Air System Interaction in Residential Buildings

Standard practice for HVAC return design has evolved from running a dedicated return to each room with a supply, to systems with returns in more centrally located areas of the home with jump ducts, transfer grilles, or simply door undercuts used as return air pathways from isolated rooms . In these latter systems, hallways and stairwells act as large open ducts for conveying air back to a single (or sometimes multiple) central return. When partition doors to an isolated room are closed, the return airflow resistance goes up, significantly impacting airflow balance if an adequate relief pathway is not installed. Airflow imbalances can lead to comfort and building durability issues and increased envelope leakage. To combat this, some jurisdictions have requirements for return air pathways. The supply system topology – or layout - impacts the airflow balance stability in response to adjustments of return pathway resistances. Branching supply topologies typically have reduced static pressure after each split. The static pressure at the final split will be lower than the primary supply plenum. Because of this, if there is a restriction in a room’s return path, supply airflow will tend to redistribute to adjacent ducts at the end of the branch. A properly designed trunk and branch supply system can effectively equalize static pressure by reducing the cross sectional area after each takeoff . Maintaining static pressure within the trunk will reduce the system’s sensitivity to changes in return paths. However, in practice, it is difficult to design and time consuming to install a complex supply plenum. Velocity effects and poor takeoff placement also impact airflow balance. This fact sheet considers return systems in three main categories: distributed, with a return duct to each room; multiple central, with one return grille on each floor of a home; and single central, with a single return grille located near the air handling unit. Three supply categories are also considered: radial splitter box, trunk and branch, and home-run with all ducts connecting directly to a central manifold. A complete description of the modeling work and results can be found in the companion technical report.

buildings↗
DOE OSTI2024

Probabilistic neural networks for improved analyses with phenomenological R -matrix

Here we present a method for measurement analyses based on probabilistic deep neural networks that provide several advantages over conventional analyses with phenomenological models. These include predicting physical quantities directly from data, the rapid generation of statistically robust uncertainties, and the ability to bypass some parameters that may induce ambiguities and complications in data analysis. As deep learning methods make predictions through “black boxes,” the uncertainty quantification is typically challenging. We use a probabilistic framework that provides thorough uncertainty quantification and is straightforward to follow in practice. With the network architecture based on the Transformer, we demonstrate the current method for predicting nuclear resonance parameters from scattering data using the phenomenological R-matrix model.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2026

Gradient-informed Hamiltonian Monte Carlo for multicomponent CALPHAD model optimization and uncertainty quantification

CALPHAD model parameter optimization is inherently challenging due to non-smooth objective functions, high-dimensional parameter spaces, and the need for uncertainty quantification (UQ). Traditional weighted nonlinear least squares approaches are computationally efficient but local, whereas black-box global optimizers and ensemble Markov Chain Monte Carlo (MCMC) methods provide broader exploration at substantial computational cost. The objective of this work is to combine the global exploration capability of gradient-informed Hamiltonian Monte Carlo – specifically the No-U-Turn Sampler (NUTS) – with local deterministic refinement using BFGS to efficiently optimize multicomponent CALPHAD models with minimal manual intervention. Analytic gradients are computed via the Jansson derivative framework. The methodology is demonstrated on the Cr—Fe binary system and extended to the Cr—Fe—Ni ternary system with 32 degrees of freedom. For Cr—Fe, NUTS achieves comparable or superior optimality relative to ensemble MCMC while requiring over an order-of-magnitude fewer likelihood evaluations. Parameter uncertainties are quantified through NUTS sampling and propagated to thermodynamic observables using local expansion, demonstrating a novel modular approach that combines binary and ternary parameter subsets without requiring global relaxation. These results establish gradient-informed exploration as a scalable strategy for multicomponent CALPHAD optimization and provide a practical route towards efficient higher-order database development with quantified uncertainty.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Exploring Gradient-Based Multi-directional Controls in GANs

Generative Adversarial Networks (GANs) have been widely applied in modeling diverse image distributions. However, despite its impressive applications, the structure of the latent space in GANs largely remains as a black-box, leaving its controllable generation an open problem, especially when spurious correlations between different semantic attributes exist in the image distributions. To address this problem, previous methods typically learn linear directions or individual channels that control semantic attributes in the image space. However, they often suffer from imperfect disentanglement, or are unable to obtain multi-directional controls. Furthermore, in this work, in light of the above challenges, we propose a novel approach that discovers nonlinear controls, which enables multi-directional manipulation as well as effective disentanglement, based on gradient information in the learned GAN latent space. More specifically, we first learn interpolation directions by following the gradients from classification networks trained separately on the attributes, and then navigate the latent space by exclusively controlling channels activated for the target attribute in the learned directions. Empirically, with small training data, our approach is able to gain fine-grained controls over a diverse set of bi-directional and multi-directional attributes, and we showcase its ability to achieve disentanglement significantly better than state-of-the-art methods both qualitatively and quantitatively.

97 MATHEMATICS AND COMPUTING↗