- Browse by Author
Browsing by Author "Dale, Ashley S."
Now showing 1 - 9 of 9
Results Per Page
Sort Options
Item 3D Object Detection Using Virtual Environment Assisted Deep Network Training(2020-12) Dale, Ashley S.; Christopher, Lauren; King, Brian; Salama, PaulAn RGBZ synthetic dataset consisting of five object classes in a variety of virtual environments and orientations was combined with a small sample of real-world image data and used to train the Mask R-CNN (MR-CNN) architecture in a variety of configurations. When the MR-CNN architecture was initialized with MS COCO weights and the heads were trained with a mix of synthetic data and real world data, F1 scores improved in four of the five classes: The average maximum F1-score of all classes and all epochs for the networks trained with synthetic data is F1∗ = 0.91, compared to F1 = 0.89 for the networks trained exclusively with real data, and the standard deviation of the maximum mean F1-score for synthetically trained networks is σ∗ = 0.015, compared to σ_F1 = 0.020 for the networks trained exclusively with real F1 data. Various backgrounds in synthetic data were shown to have negligible impact on F1 scores, opening the door to abstract backgrounds and minimizing the need for intensive synthetic data fabrication. When the MR-CNN architecture was initialized with MS COCO weights and depth data was included in the training data, the net- work was shown to rely heavily on the initial convolutional input to feed features into the network, the image depth channel was shown to influence mask generation, and the image color channels were shown to influence object classification. A set of latent variables for a subset of the synthetic datatset was generated with a Variational Autoencoder then analyzed using Principle Component Analysis and Uniform Manifold Projection and Approximation (UMAP). The UMAP analysis showed no meaningful distinction between real-world and synthetic data, and a small bias towards clustering based on image background.Item CNN-based network has Network Anisotropy -work harder to learn rotated feature than non-rotated feature(IEEE, 2022-10) Dale, Ashley S.; Qui, Mei; Christopher, Lauren; Krogg, Wen; William, Albert; Electrical and Computer Engineering, School of Engineering and TechnologySuccessful object identification and classification in a generic Convolutional Neural Network (CNN) depends on object orientation. We expect CNN-based architectures to work harder to learn a rotated version of a feature than when learning the same feature in its default orientation. We name this phenomenon “Network Anisotropy”. A data set of 6000 RGB and grayscale images was created with rotated orientations of a feature predetermined and evenly distributed across four classes: 0°, 30°, 60°, 90°. Four ResNet (18, 34, 50, 101) classifier architectures were trained and the confidence scores were used to represent prediction accuracy. The results show that in all networks, training performance lags several epochs for the 30° and 60° rotation predictions compared to the 0° and 90° rotations, indicating a quantifiable network anisotropy. Because 0° and 90° both lie along a single rectilinear axis that coincides with the convolutional kernel of the CNN, we expect the classifier to do better on these two classes than on 30° and 60° classes. This work confirms that CNN architectures may have weaker performance based on feature orientation alone, independent of the feature distribution within the data set or the correlation of features within an image.Item Developing an approach to improve beta-phase properties in ferroelectric pvd-hfp thin films(2020-05) Dale, Ashley S.; Cheng, Ruihua; Petrache, Horia; Wassall, StephenImproved fabrication of poly(vinylindenefluoride)-hexafluoropropylene (PVDF-HFP) thin films is of particular interest due to the high electric coercivity found in the beta-phase structure of the thin film. We show that it is possible to obtain high-quality, beta-phase dominant PVDF-HFP thin films using a direct approach to Langmuir-Blodgett deposition without the use of annealing or additives. To improve sample quality, an automated Langmuir-Blodgett thin film deposition system was developed; a custom dipping trough was fabricated, a sample dipping mechanism was designed and constructed, and the system was automated using custom LabVIEW software. Samples were fabricated in the form of ferroelectric capacitors on substrates of glass and silicon, and implement a unique step design with a bottom electrode of copper with an aluminum wetting layer and a top electrode of gold with an aluminum wetting layer. Samples were then characterized using a custom ferroelectric measurement program implemented in LabVIEW with a Keithley picoammeter/voltage supply to confirm electric coercivity properties. Further characterization using scanning electron microscopy and atomic force microscopy confirmed the improvement in thin film fabrication over previous methods.Item Direct observation of the magnetic anisotropy of an Fe(II) spin crossover molecular thin film(IOPP, 2023-07) Dale, Ashley S.; Yazdani, Saeed; Ekanayaka, Thinlini K.; Mishra, Esha; Hu, Yuchen; Dowben, Peter A.; Freeland, John W.; Zhang, Jian; Cheng, Ruihua; Physics, School of ScienceIn this work, we provide clear evidence of magnetic anisotropy in the local orbital moment of a molecular thin film based on the SCO complex [Fe(H2B(pz)2)2(bipy)] (pz = pyrazol−1−yl, bipy = 2,2'−bipyridine). Field dependent x-ray magnetic circular dichroism measurements indicate that the magnetic easy axis for the orbital moment is along the surface normal direction. Along with the presence of a critical field, our observation points to the existence of an anisotropic energy barrier in the high-spin state. The estimated nonzero coupling constant of ∼2.47 × 10−5 eV molecule−1 indicates that the observed magnetocrystalline anisotropy is mostly due to spin–orbit coupling. The spin- and orbital-component anisotropies are determined to be 30.9 and 5.04 meV molecule−1, respectively. Furthermore, the estimated g factor in the range of 2.2–2.45 is consistent with the expected values. This work has paved the way for an understanding of the spin-state-switching mechanism in the presence of magnetic perturbations.Item Evidence of dynamical effects and critical field in a cobalt spin crossover complex(Royal Society of Chemistry, 2022-01) Ekanayaka, Thilini K.; Wang, Ping; Yazdani, Saeed; Phillips, Jared Paul; Mishra, Esha; Dale, Ashley S.; N'Diaye, Alpha T.; Klewe, Christoph; Shafer, Padraic; Freeland, John; Streubel, Robert; Wampler, James Paris; Zapf, Vivien; Cheng, Ruihua; Shatruk, Michael; Dowben, Peter A.; Physics, School of ScienceThe [Co(SQ)2(4-CN-py)2] complex exhibits dynamical effects over a wide range of temperature. The orbital moment, determined by X-ray magnetic circular dichroism (XMCD) with decreasing applied magnetic field, indicates a nonzero critical field for net alignment of magnetic moments, an effect not seen with the spin moment of [Co(SQ)2(4-CN-py)2].Item Magnetic Field Perturbations to a Soft X-ray-Activated Fe (II) Molecular Spin State Transition(MDPI, 2021-10) Hao, Guanhua; N’Diaye, Alpha T.; Ekanayaka, Thilini K.; Dale, Ashley S.; Jiang, Xuanyuan; Mishra, Esha; Mellinger, Corbyn; Yazdani, Saeed; Freeland, John W.; Zhang, Jian; Cheng, Ruihua; Xu, Xiaoshan; Dowben, Peter A.; Physics, School of ScienceThe X-ray-induced spin crossover transition of an Fe (II) molecular thin film in the presence and absence of a magnetic field has been investigated. The thermal activation energy barrier in the soft X-ray activation of the spin crossover transition for [Fe{H2B(pz)2}2(bipy)] molecular thin films is reduced in the presence of an applied magnetic field, as measured through X-ray absorption spectroscopy at various temperatures. The influence of a 1.8 T magnetic field is sufficient to cause deviations from the expected exponential spin state transition behavior which is measured in the field free case. We find that orbital moment diminishes with increasing temperature, relative to the spin moment in the vicinity of room temperature.Item Nonvolatile Voltage Controlled Molecular Spin-State Switching for Memory Applications(MDPI, 2021-03) Ekanayaka, Thilini K.; Hao, Guanhua; Mosey, Aaron; Dale, Ashley S.; Jiang, Xuanyuan; Yost, Andrew J.; Sapkota, Keshab R.; Wang, George T.; Zhang, Jian; N’Diaye, Alpha T.; Marshall, Andrew; Cheng, Ruihua; Naeemi, Azad; Xu, Xiaoshan; Dowben, Peter A.; Physics, School of ScienceNonvolatile, molecular multiferroic devices have now been demonstrated, but it is worth giving some consideration to the issue of whether such devices could be a competitive alternative for solid-state nonvolatile memory. For the Fe (II) spin crossover complex [Fe{H2B(pz)2}2(bipy)], where pz = tris(pyrazol-1-yl)-borohydride and bipy = 2,2′-bipyridine, voltage-controlled isothermal changes in the electronic structure and spin state have been demonstrated and are accompanied by changes in conductance. Higher conductance is seen with [Fe{H2B(pz)2}2(bipy)] in the high spin state, while lower conductance occurs for the low spin state. Plausibly, there is the potential here for low-cost molecular solid-state memory because the essential molecular thin films are easily fabricated. However, successful device fabrication does not mean a device that has a practical value. Here, we discuss the progress and challenges yet facing the fabrication of molecular multiferroic devices, which could be considered competitive to silicon.Item Probing the unpaired Fe spins across the spin crossover of a coordination polymer(RSC, 2021) Ekanayaka, Thilini K.; Kurz, Hannah; Dale, Ashley S.; Hao, Guanhua; Mosey, Aaron; Mishra, Esha; N'Diaye, Alpha T.; Cheng, Ruihua; Weber, Birgit; Dowben, Peter A.; Physics, School of ScienceFor the spin crossover coordination polymer [Fe(L1)(bipy)]n (where L1 is a N2O22− coordinating Schiff base-like ligand bearing a phenazine fluorophore and bipy = 4,4′-bipyridine), there is compelling additional evidence of a spin state transition. Both Fe 2p X-ray absorption and X-ray core level photoemission spectroscopies confirm that a spin crossover takes place, as observed by magnetometry. Yet the details of the temperature dependent changes of the spin state inferred from both X-ray absorption and X-ray core level photoemission, differ from magnetometry, particularly with regard to the apparent critical transition temperatures and the cooperative nature of the curve progression in general. Comparing the experimental spin crossover data to Ising model simulations, a transition activation energy in the region of 160 to 175 meV is indicated, along with a nonzero exchange J. Overall, the implication is that there may be perturbations to the bistability of spin states, that are measurement dependent or that the surface differs from the bulk with regard to the cooperative effects observed upon spin transition.Item Quantitative Study of the Energy Changes in Voltage-Controlled Spin Crossover Molecular Thin Films(ACS, 2020-09) Mosey, Aaron; Dale, Ashley S.; Hao, Guanhua; N'Diaye, Alpha; Dowben, Peter A.; Cheng, Ruihua; Physics, School of ScienceVoltage-controlled nonvolatile isothermal spin state switching of a [Fe{H2B(pz)2}2(bipy)] (pz = tris(pyrazol-1–1y)-borohydride, bipy = 2,2′-bipyridine) film, more than 40 to 50 molecular layers thick, is possible when it is adsorbed onto a molecular ferroelectric substrate. Accompanying this high-spin and low-spin state switching, at room temperature, we observe a remarkable change in conductance, thereby allowing not only nonvolatile voltage control of the spin state (“write”) but also current sensing of the molecular spin state (“read”). Monte Carlo Ising model simulations of the high-spin state occupancy, extracted from X-ray absorption spectroscopy, indicate that the energy difference between the low-spin and high-spin state is modified by 110 meV. Transport measurements demonstrate that four terminal voltage-controlled devices can be realized using this system.