Showing 1–13 of 13 results for author: Jacobson, P

Doubly Robust Self-Training

Authors: Banghua Zhu, Mingyu Ding, Philip Jacobson, Ming Wu, Wei Zhan, Michael Jordan, Jiantao Jiao

Abstract: Self-training is an important technique for solving semi-supervised learning problems. It leverages unlabeled data by generating pseudo-labels and combining them with a limited labeled dataset for training. The effectiveness of self-training heavily relies on the accuracy of these pseudo-labels. In this paper, we introduce doubly robust self-training, a novel semi-supervised algorithm that provabl… ▽ More Self-training is an important technique for solving semi-supervised learning problems. It leverages unlabeled data by generating pseudo-labels and combining them with a limited labeled dataset for training. The effectiveness of self-training heavily relies on the accuracy of these pseudo-labels. In this paper, we introduce doubly robust self-training, a novel semi-supervised algorithm that provably balances between two extremes. When the pseudo-labels are entirely incorrect, our method reduces to a training process solely using labeled data. Conversely, when the pseudo-labels are completely accurate, our method transforms into a training process utilizing all pseudo-labeled data and labeled data, thus increasing the effective sample size. Through empirical evaluations on both the ImageNet dataset for image classification and the nuScenes autonomous driving dataset for 3D object detection, we demonstrate the superiority of the doubly robust loss over the standard self-training baseline. △ Less

Submitted 2 November, 2023; v1 submitted 31 May, 2023; originally announced June 2023.

Near-field localization of the boson peak on tantalum films for superconducting quantum devices

Authors: Xiao Guo, Zachary Degnan, Julian Steele, Eduardo Solano, Bogdan C. Donose, Karl Bertling, Arkady Fedorov, Aleksandar D. Rakić, Peter Jacobson

Abstract: Superconducting circuits are among the most advanced quantum computing technologies, however their performance is currently limited by losses found in surface oxides and disordered materials. Here, we identify and spatially localize a near-field signature of loss centers on tantalum films using terahertz scattering-type scanning near-field optical microscopy (s-SNOM). Making use of terahertz nanos… ▽ More Superconducting circuits are among the most advanced quantum computing technologies, however their performance is currently limited by losses found in surface oxides and disordered materials. Here, we identify and spatially localize a near-field signature of loss centers on tantalum films using terahertz scattering-type scanning near-field optical microscopy (s-SNOM). Making use of terahertz nanospectroscopy, we observe a localized excess vibrational mode around 0.5 THz and identify this resonance as the boson peak, a signature of amorphous materials. Grazing-incidence wide-angle x-ray scattering (GIWAXS) shows that oxides on freshly solvent-cleaned samples are amorphous, whereas crystalline phases emerge after aging in air. By localizing defect centers at the nanoscale, our characterization techniques and results will inform the optimization of fabrication procedures for new low-loss superconducting circuits. △ Less

Submitted 28 February, 2023; originally announced February 2023.

Comments: 15 pages including supplement. 13 figures (4 in main text, 9 in supplement)

Journal ref: J. Phys. Chem. Lett. 2023, 14, 20, 4892

Measurement-based Quantum Computation as a Tangram Puzzle

Authors: Ashlesha Patil, Yosef P. Jacobson, Don Towsley, Saikat Guha

Abstract: Measurement-Based Quantum Computing (MBQC), proposed in 2001 is a model of quantum computing that achieves quantum computation by performing a series of adaptive single-qubit measurements on an entangled cluster state. Our project is aimed at introducing MBQC to a wide audience ranging from high school students to quantum computing researchers through a Tangram puzzle with a modified set of rules,… ▽ More Measurement-Based Quantum Computing (MBQC), proposed in 2001 is a model of quantum computing that achieves quantum computation by performing a series of adaptive single-qubit measurements on an entangled cluster state. Our project is aimed at introducing MBQC to a wide audience ranging from high school students to quantum computing researchers through a Tangram puzzle with a modified set of rules, played on an applet. The rules can be understood without any background in quantum computing. The player is provided a quantum circuit, shown using gates from a universal gate set, which the player must map correctly to a playing board using polyominos. Polyominos or 'puzzle blocks' are the building blocks of our game. They consist of square tiles joined edge-to-edge to form different colored shapes. Each tile represents a single-qubit measurement basis, differentiated by its color. Polyominos rest on a square-grid playing board, which signifies a cluster state. We show that map** a quantum circuit to MBQC is equivalent to arranging a set of polyominos, each corresponding to a gate in the circuit on the playing board, subject to certain rules, which involve rotating and deforming polyominos. We state the rules in simple terms with no reference to quantum computing. The player has to place polyominos on the playing board conforming to the rules. Any correct solution creates a valid realization of the quantum circuit in MBQC. A higher-scoring correct solution fills up less space on the board, resulting in a lower-overhead embedding of the circuit in MBQC, an open and challenging research problem. △ Less

Submitted 20 October, 2022; originally announced October 2022.

Center Feature Fusion: Selective Multi-Sensor Fusion of Center-based Objects

Authors: Philip Jacobson, Yiyang Zhou, Wei Zhan, Masayoshi Tomizuka, Ming C. Wu

Abstract: Leveraging multi-modal fusion, especially between camera and LiDAR, has become essential for building accurate and robust 3D object detection systems for autonomous vehicles. Until recently, point decorating approaches, in which point clouds are augmented with camera features, have been the dominant approach in the field. However, these approaches fail to utilize the higher resolution images from… ▽ More Leveraging multi-modal fusion, especially between camera and LiDAR, has become essential for building accurate and robust 3D object detection systems for autonomous vehicles. Until recently, point decorating approaches, in which point clouds are augmented with camera features, have been the dominant approach in the field. However, these approaches fail to utilize the higher resolution images from cameras. Recent works projecting camera features to the bird's-eye-view (BEV) space for fusion have also been proposed, however they require projecting millions of pixels, most of which only contain background information. In this work, we propose a novel approach Center Feature Fusion (CFF), in which we leverage center-based detection networks in both the camera and LiDAR streams to identify relevant object locations. We then use the center-based detection to identify the locations of pixel features relevant to object locations, a small fraction of the total number in the image. These are then projected and fused in the BEV frame. On the nuScenes dataset, we outperform the LiDAR-only baseline by 4.9% mAP while fusing up to 100x fewer features than other fusion methods. △ Less

Submitted 26 April, 2023; v1 submitted 26 September, 2022; originally announced September 2022.

Comments: Accepted by ICRA 2023

Ternary Metal Oxide Substrates for Superconducting Circuits

Authors: Zach Degnan, Xin He, Alejandro Gomez Frieiro, Yauhen P. Sachkou, Arkady Fedorov, Peter Jacobson

Abstract: Substrate material imperfections and surface losses are one of the major factors limiting superconducting quantum circuitry from reaching the scale and complexity required to build a practicable quantum computer. One potential path towards higher coherence of superconducting quantum devices is to explore new substrate materials with a reduced density of imperfections due to inherently different su… ▽ More Substrate material imperfections and surface losses are one of the major factors limiting superconducting quantum circuitry from reaching the scale and complexity required to build a practicable quantum computer. One potential path towards higher coherence of superconducting quantum devices is to explore new substrate materials with a reduced density of imperfections due to inherently different surface chemistries. Here, we examine two ternary metal oxide materials, spinel (MgAl2O4) and lanthanum aluminate (LaAlO3), with a focus on surface and interface characterization and preparation. Devices fabricated on LaAlO3 have quality factors three times higher than earlier devices, which we attribute to a reduction in interfacial disorder. MgAl2O4 is a new material in the realm of superconducting quantum devices and, even in the presence of significant surface disorder, consistently outperforms LaAlO3. Our results highlight the importance of materials exploration, substrate preparation, and characterization to identify materials suitable for high-performance superconducting quantum circuitry. △ Less

Submitted 17 January, 2022; originally announced January 2022.

Comments: 5 pages, 6 figures

Journal ref: Mater. Quantum. Technol. 2 025004 (2022)

Few-Shot Image Classification Along Sparse Graphs

Authors: Joseph F Comer, Philip L Jacobson, Heiko Hoffmann

Abstract: Few-shot learning remains a challenging problem, with unsatisfactory 1-shot accuracies for most real-world data. Here, we present a different perspective for data distributions in the feature space of a deep network and show how to exploit it for few-shot learning. First, we observe that nearest neighbors in the feature space are with high probability members of the same class while generally two… ▽ More Few-shot learning remains a challenging problem, with unsatisfactory 1-shot accuracies for most real-world data. Here, we present a different perspective for data distributions in the feature space of a deep network and show how to exploit it for few-shot learning. First, we observe that nearest neighbors in the feature space are with high probability members of the same class while generally two random points from one class are not much closer to each other than points from different classes. This observation suggests that classes in feature space form sparse, loosely connected graphs instead of dense clusters. To exploit this property, we propose using a small amount of label propagation into the unlabeled space and then using a kernel PCA reconstruction error as decision boundary for the feature-space data distribution of each class. Using this method, which we call "K-Prop," we demonstrate largely improved few-shot learning performances (e.g., 83% accuracy for 1-shot 5-way classification on the RESISC45 satellite-images dataset) for datasets for which a backbone network can be trained with high within-class nearest-neighbor probabilities. We demonstrate this relationship using six different datasets. △ Less

Submitted 7 December, 2021; originally announced December 2021.

Testing for the Continuous Spectrum of X-Rays Predicted to Accompany the Photoejection of an Atomic Inner Shell Electron

Authors: Philip Jacobson, Andrija Rasovic, Arthur Campello, Chase Goddard, Matthew Dykes, Yuchao Chen, J. Y. Peter Ko, Stanislav Stoupin, Gwen Gardner, Justin Oh, Carl Franck

Abstract: Echoing classical physics, quantum electrodynamics predicts the release of a spectral continuum of electromagnetic radiation upon the sudden acceleration of charged particles in quantum matter. Despite apparent theoretical success in describing sister nuclear processes, known as internal bremsstrahlung, following nuclear beta decay and K capture, the situation of the photoejection of an electron f… ▽ More Echoing classical physics, quantum electrodynamics predicts the release of a spectral continuum of electromagnetic radiation upon the sudden acceleration of charged particles in quantum matter. Despite apparent theoretical success in describing sister nuclear processes, known as internal bremsstrahlung, following nuclear beta decay and K capture, the situation of the photoejection of an electron from an inner shell of an atom, intraatomic bremsstrahlung (IAB), is far from settled. In this paper we present fresh measurements which rely on contemporary signal processing as well as the high flux available from a synchrotron radiation source to revisit the problem by photoejecting electrons from the innermost shell of copper. For the first time we have sufficient sample statistics to measure the expected spectra at the level expected by contemporary theory. Furthermore, we employ sufficiently thin targets to overcome secondary scattering artifacts. Our approach applies the fluorescence coincidence method to guard against extraneous scattering and multiple incident photon processes. Our observations set a severe upper limit on the rate for IAB: We conclude that current theory overpredicts, by at least 5 sigma, the measured rate for K shell IAB in copper in the range of detected energies below the K fluorescence energy. △ Less

Submitted 3 November, 2021; originally announced November 2021.

Journal ref: Physical Review A 104, 042809 (2021)

Terahertz nanospectroscopy of plasmon polaritons for the evaluation of do** in quantum devices

Authors: Xiao Guo, Xin He, Zach Degnan, Chun-Ching Chiu, Bogdan C Donose, Karl Bertling, Arkady Fedorov, Aleksandar D Rakic, Peter Jacobson

Abstract: Terahertz (THz) waves are a highly sensitive probe of free carrier concentrations in semiconducting materials. However, most experiments operate in the far-field, which precludes the observation of nanoscale features that affect the material response. Here, we demonstrate the use of nanoscale THz plasmon polaritons as an indicator of surface quality in prototypical quantum devices properties. Usin… ▽ More Terahertz (THz) waves are a highly sensitive probe of free carrier concentrations in semiconducting materials. However, most experiments operate in the far-field, which precludes the observation of nanoscale features that affect the material response. Here, we demonstrate the use of nanoscale THz plasmon polaritons as an indicator of surface quality in prototypical quantum devices properties. Using THz near-field hyperspectral measurements, we observe polaritonic features in doped silicon near a metal-semiconductor interface. The presence of the THz surface plasmon polariton indicates the existence of a thin film doped layer on the device. Using a multilayer extraction procedure utilising vector calibration, we quantitatively probe the doped surface layer and determine its thickness and complex permittivity. The recovered multilayer characteristics match the dielectric conditions necessary to support the THz surface plasmon polariton. Applying these findings to superconducting resonators, we show that etching of this doped layer leads to an increase of the quality factor as determined by cryogenic measurements. This study demonstrates that THz scattering-type scanning near-field optical microscopy (s-SNOM) is a promising diagnostic tool for characterization of surface dielectric properties of quantum devices. △ Less

Submitted 2 August, 2023; v1 submitted 12 October, 2021; originally announced October 2021.

Comments: 16 pages with 12 figures

Journal ref: Nanophotonics, vol. 12, no. 10, 2023, pp. 1865-1875

Near-Field Terahertz Nanoscopy of Coplanar Microwave Resonators

Authors: Xiao Guo, Xin He, Zach Degnan, Bogdan C. Donose, Karl Bertling, Arkady Fedorov, Aleksandar D. Rakić, Peter Jacobson

Abstract: Superconducting quantum circuits are one of the leading quantum computing platforms. To advance superconducting quantum computing to a point of practical importance, it is critical to identify and address material imperfections that lead to decoherence. Here, we use terahertz Scanning Near-field Optical Microscopy (SNOM) to probe the local dielectric properties and carrier concentrations of wet-et… ▽ More Superconducting quantum circuits are one of the leading quantum computing platforms. To advance superconducting quantum computing to a point of practical importance, it is critical to identify and address material imperfections that lead to decoherence. Here, we use terahertz Scanning Near-field Optical Microscopy (SNOM) to probe the local dielectric properties and carrier concentrations of wet-etched aluminum resonators on silicon, one of the most characteristic components of the superconducting quantum processors. Using a recently developed vector calibration technique, we extract the THz permittivity from spectroscopy in proximity to the microwave feedline. Fitting the extracted permittivity to the Drude model, we find that silicon in the etched channel has a carrier concentration greater than buffer oxide etched silicon and we explore post-processing methods to reduce the carrier concentrations. Our results show that near-field THz investigations can be applied to quantitatively evaluate and identify potential loss channels in quantum devices. △ Less

Submitted 1 November, 2021; v1 submitted 24 June, 2021; originally announced June 2021.

Comments: 14 pages, 3 figures. This article may be downloaded for personal use only. Any other use requires prior permission of the author and AIP Publishing. This article appeared in Applied Physics Letters 119 (9), 091101 and may be found at https://aip.scitation.org/doi/full/10.1063/5.0061078

Journal ref: Appl. Phys. Lett. 119, 091101 (2021)

Bulk and surface characterization of In$_2$O$_3$(001) single crystals

Authors: Daniel R. Hagleitner, Peter Jacobson, Sara Blomberg, Karina Schulte, Edvin Lundgren, Markus Kubicek, Jürgen Fleig, Frank Kubel, Christoph Puls, Andreas Limbeck, Herbert Hutter, Lynn A. Boatner, Michael Schmid, Ulrike Diebold

Abstract: A comprehensive bulk and surface investigation of high-quality In$_2$O$_3$(001) single crystals is reported. The transparent-yellow, cube-shaped single crystals were grown using the flux method. Inductively coupled plasma mass spectrometry (ICP-MS) reveals small residues of Pb, Mg, and Pt in the crystals. Four-point-probe measurements show a resistivity of 2.0 $\pm$ 0.5 $\times$ 10$^5$ Ω cm, which… ▽ More A comprehensive bulk and surface investigation of high-quality In$_2$O$_3$(001) single crystals is reported. The transparent-yellow, cube-shaped single crystals were grown using the flux method. Inductively coupled plasma mass spectrometry (ICP-MS) reveals small residues of Pb, Mg, and Pt in the crystals. Four-point-probe measurements show a resistivity of 2.0 $\pm$ 0.5 $\times$ 10$^5$ Ω cm, which translates into a carrier concentration of $\approx$10$^{12}$ cm$^{-3}$. The results from x-ray diffraction (XRD) measurements revise the lattice constant to 10.1150(5) Å from the previously accepted value of 10.117 Å. Scanning tunneling microscopy (STM) images of a reduced (sputtered/annealed) and oxidized (exposure to atomic oxygen at 300 °C) surface show a step height of 5 Å, which indicates a preference for one type of surface termination. The surfaces stay flat without any evidence for macroscopic faceting under any of these preparation conditions. A combination of low-energy ion scattering (LEIS) and atomically resolved STM indicates an indium-terminated surface with small islands of 2.5 Å height under reducing conditions, with a surface structure corresponding to a strongly distorted indium lattice. Scanning tunneling spectroscopy (STS) reveals a pronounced surface state at the Fermi level ($E_F$). Photoelectron spectroscopy (PES) shows additional, deep-lying band gap states, which can be removed by exposure of the surface to atomic oxygen. Oxidation also results in a shoulder at the O 1s core level at a higher binding energy, possibly indicative of a surface peroxide species. A downward band bending of 0.4 eV is observed for the reduced surface, while the band bending of the oxidized surface is of the order of 0.1 eV or less. △ Less

Submitted 12 April, 2018; originally announced April 2018.

Journal ref: Phys. Rev. B 85, 115441 (2012)

Potential Energy Driven Spin Manipulation via a Controllable Hydrogen Ligand

Authors: Peter Jacobson, Matthias Muenks, Gennadii Laskin, Oleg O. Brovko, Valeri S. Stepanyuk, Markus Ternes, Klaus Kern

Abstract: Spin-bearing molecules can be stabilized on surfaces and in junctions with desirable properties such as a net spin that can be adjusted by external stimuli. Using scanning probes, initial and final spin states can be deduced from topographic or spectroscopic data, but how the system transitioned between these states is largely unknown. Here we address this question by manipulating the total spin o… ▽ More Spin-bearing molecules can be stabilized on surfaces and in junctions with desirable properties such as a net spin that can be adjusted by external stimuli. Using scanning probes, initial and final spin states can be deduced from topographic or spectroscopic data, but how the system transitioned between these states is largely unknown. Here we address this question by manipulating the total spin of magnetic cobalt hydride complexes on a corrugated boron nitride surface with a hydrogen- functionalized scanning probe tip by simultaneously tracking force and conductance. When the additional hydrogen ligand is brought close to the cobalt monohydride, switching between a corre- lated S = 1 /2 Kondo state, where host electrons screen the magnetic moment, and a S = 1 state with magnetocrystalline anisotropy is observed. We show that the total spin changes when the system is transferred onto a new potential energy surface defined by the position of the hydrogen in the junction. These results show how and why chemically functionalized tips are an effective tool to manipulate adatoms and molecules, and a promising new method to selectively tune spin systems. △ Less

Submitted 15 December, 2016; v1 submitted 2 September, 2016; originally announced September 2016.

Correlation Driven Transport Asymmetries Through Coupled Spins

Authors: Matthias Muenks, Peter Jacobson, Markus Ternes, Klaus Kern

Abstract: Correlation is a fundamental statistical measure of order in interacting quantum systems. In solids, electron correlations govern a diverse array of material classes and phenomena such as heavy fermion compounds, Hunds metals, high-Tc superconductors, and the Kondo effect. Spin-spin correlations, notably investigated by Kaufman and Onsager in the 1940s 6, are at the foundation of numerous theoreti… ▽ More Correlation is a fundamental statistical measure of order in interacting quantum systems. In solids, electron correlations govern a diverse array of material classes and phenomena such as heavy fermion compounds, Hunds metals, high-Tc superconductors, and the Kondo effect. Spin-spin correlations, notably investigated by Kaufman and Onsager in the 1940s 6, are at the foundation of numerous theoretical models but are challenging to measure experimentally. Reciprocal space methods can map correlations, but at the single atom limit new experimental probes are needed. Here, we determine the correlations between a strongly hybridized spin impurity and its electron bath by varying the coupling to a second magnetic impurity in the junction of a scanning tunneling microscope. Electronic transport through these coupled spins reveals an asymmetry in the differential conductance reminiscent of spin-polarized transport in a magnetic field. We show that at zero field, this asymmetry can be controlled by the coupling strength and is directly related to either ferromagnetic (FM) or antiferromagnetic (AFM) spin-spin correlations. △ Less

Submitted 9 May, 2016; originally announced May 2016.

Comments: 18 pages, 4 figures

Journal ref: Nat. Commun. 8, 14119 (2017)

Quantum Engineering of Spin and Anisotropy in Magnetic Molecular Junctions

Authors: Peter Jacobson, Tobias Herden, Matthias Muenks, Gennadii Laskin, Oleg Brovko, Valeri Stepanyuk, Markus Ternes, Klaus Kern

Abstract: Single molecule magnets and single spin centers can be individually addressed when coupled to contacts forming an electrical junction. In order to control and engineer the magnetism of quantum devices, it is necessary to quantify how the structural and chemical environment of the junction affects the spin center. Metrics such as coordination number or symmetry provide a simple method to quantify t… ▽ More Single molecule magnets and single spin centers can be individually addressed when coupled to contacts forming an electrical junction. In order to control and engineer the magnetism of quantum devices, it is necessary to quantify how the structural and chemical environment of the junction affects the spin center. Metrics such as coordination number or symmetry provide a simple method to quantify the local environment, but neglect the many-body interactions of an impurity spin when coupled to contacts. Here, we utilize a highly corrugated hexagonal boron nitride (h-BN) monolayer to mediate the coupling between a cobalt spin in CoHx (x=1,2) complexes and the metal contact. While the hydrogen atoms control the total effective spin, the corrugation is found to smoothly tune the Kondo exchange interaction between the spin and the underlying metal. Using scanning tunneling microscopy and spectroscopy together with numerical simulations, we quantitatively demonstrate how the Kondo exchange interaction mimics chemical tailoring and changes the magnetic anisotropy. △ Less

Submitted 9 May, 2015; originally announced May 2015.

Journal ref: Nature Communications 6, 8536 (2015)