Angular Encoding Research Articles

Hybrid quantum neural network structures for image multi-classification

Image classification is a fundamental problem in computer vision, and neural networks provide an effective solution. With the advancement of quantum technology, quantum neural networks have attracted a lot of attention. However, they are only suitable for low-dimensional data and require dimensionality reduction and quantum encoding. Two image classification methods have recently emerged: one employs PCA for dimensionality reduction and angular encoding, and the other integrates QNN into a CNN to improve performance. Despite numerous algorithms, the differences between them remain unclear. This study explores these algorithms’ performance in multi-class image classification and proposes an optimized hybrid quantum neural network suitable for the current environment. As the number of classes increases, research on PCA-based quantum algorithms reveals the barren plateau problem of QNN, which is not suitable for multiple classes in a hybrid setting. Our proposed model combining traditional CNN with QNN addresses QNN’s multi-class training difficulties to some extent and achieves satisfactory classification results. Nevertheless, its accuracy remains below that of the top-performing CNN models. Furthermore, we investigate transfer learning in hybrid quantum neural network models and assess the performance of our models on the quantum hardware from IBM. In conclusion, quantum neural networks show promise but require further research and optimization, facing challenges ahead.

Fusing Higher-Order Features in Graph Neural Networks for Skeleton-Based Action Recognition.

Skeleton sequences are lightweight and compact and thus are ideal candidates for action recognition on edge devices. Recent skeleton-based action recognition methods extract features from 3-D joint coordinates as spatial-temporal cues, using these representations in a graph neural network for feature fusion to boost recognition performance. The use of first- and second-order features, that is, joint and bone representations, has led to high accuracy. Nonetheless, many models are still confused by actions that have similar motion trajectories. To address these issues, we propose fusing higher-order features in the form of angular encoding (AGE) into modern architectures to robustly capture the relationships between joints and body parts. This simple fusion with popular spatial-temporal graph neural networks achieves new state-of-the-art accuracy in two large benchmarks, including NTU60 and NTU120, while employing fewer parameters and reduced run time. Our source code is publicly available at: https://github.com/ZhenyueQin/Angular-Skeleton-Encoding.

Non-intrusive multi-label load monitoring via transfer and contrastive learning architecture

To achieve the goal of peaking carbon emissions globally and carbon neutrality, smart energy management is a promising way to boost energy conservation and estimate the residential potential for regional demand response, among which the non-intrusive load monitoring technique is highlighted due to its effectiveness on the residential side. However, the identification of multi-label appliance switching operations is still a challenge in this field, which may critically affect the total identification results due to the few-shot learning problem and the complicated overlap of features belonging to different appliances. Therefore, this paper proposed a transfer and contrastive learning architecture to identify multi-label appliances effectively. In the first stage, Gramian angular field encoding is implemented to visualize power sequences to highlight the correlation between timestamps and enhance the feature extraction efficiency. Secondly, a contrastive learning architecture is implemented to learn the general distinguishing features between samples of different labels, and density-based spatial clustering of applications with noise clustering is utilized to detect multi-label samples. Thirdly, transfer learning is utilized to enhance the multi-label identification capacity of contrastive learning structures based on the existing trained model. Finally, the effectiveness of the proposed algorithm is verified through two low-frequency non-intrusive load monitoring public datasets and real-world measurements from a pilot project in China. The results show that the proposed architecture can achieve the efficacy of deep features extraction and few-shot learning in identifying multi-label appliance switching operations.

Aspect Ratio-Based Bidirectional Label Encoding for Square-Like Rotation Detection

Rotation object detection is of great importance in remote sensing imagery where the orientation is arbitrary and objects are densely distributed. However, there are several challenges that need to be overcome, such as the angular boundary problem in the regression-based methods and the square-like problem in the classification-based methods. For square-like object rotation detection, classification-based methods (e.g. CSL) suffer from the inconsistencies between angular coding and evaluation mechanisms due to the variation of aspect ratio. To address the angular inconsistencies of square-like object existing in current classification methods, we design a novel angular encoding mechanism based on aspect ratio. We make optimisations and improvements in the following two aspects: i) proposing an aspect ratio-based bidirectional coded label (AR-BCL) to replace circular smooth label (CSL) for angle coding of square-like object, which significantly improves detection accuracy for square-like object. ii) designing a cross-fusion decoupled head (CF-DH) based on angular classification to replace the existing coupled head (CH), which can help extract features that suitable for angular classification. Extensive experiments on DOTA, a large-scale public dataset for aerial images, demonstrate the effectiveness of our method for square-like object detection.

Multiqudit quantum hashing and its implementation based on orbital angular momentum encoding

A new version of quantum hashing technique is developed wherein a quantum hash is constructed as a sequence of single-photon high-dimensional states (qudits). A proof-of-principle implementation of the high-dimensional quantum hashing protocol using orbital-angular momentum encoding of single photons is implemented. It is shown that the number of qudits decreases with increase of their dimension for an optimal ratio between collision probability and decoding probability of the hash. Thus, increasing dimension of information carriers makes quantum hashing with single photons more efficient.

Deploying Machine Learning Techniques for Human Emotion Detection.

Emotion recognition is one of the trending research fields. It is involved in several applications. Its most interesting applications include robotic vision and interactive robotic communication. Human emotions can be detected using both speech and visual modalities. Facial expressions can be considered as ideal means for detecting the persons' emotions. This paper presents a real-time approach for implementing emotion detection and deploying it in the robotic vision applications. The proposed approach consists of four phases: preprocessing, key point generation, key point selection and angular encoding, and classification. The main idea is to generate key points using MediaPipe face mesh algorithm, which is based on real-time deep learning. In addition, the generated key points are encoded using a sequence of carefully designed mesh generator and angular encoding modules. Furthermore, feature decomposition is performed using Principal Component Analysis (PCA). This phase is deployed to enhance the accuracy of emotion detection. Finally, the decomposed features are enrolled into a Machine Learning (ML) technique that depends on a Support Vector Machine (SVM), k-Nearest Neighbor (KNN), Naïve Bayes (NB), Logistic Regression (LR), or Random Forest (RF) classifier. Moreover, we deploy a Multilayer Perceptron (MLP) as an efficient deep neural network technique. The presented techniques are evaluated on different datasets with different evaluation metrics. The simulation results reveal that they achieve a superior performance with a human emotion detection accuracy of 97%, which ensures superiority among the efforts in this field.

Multigrained Angle Representation for Remote-Sensing Object Detection

Arbitrary-oriented object detection (AOOD) plays a significant role for image understanding in remote sensing scenarios. The existing AOOD methods face the challenges of ambiguity and high costs in angle representation. To this end, a multi-grained angle representation (MGAR) method, consisting of coarse-grained angle classification (CAC) and fine-grained angle regression (FAR), is proposed. Specifically, the designed CAC avoids the ambiguity of angle prediction by discrete angular encoding (DAE) and reduces complexity by coarsening the granularity of DAE. Based on CAC, FAR is developed to refine the angle prediction with much lower costs than narrowing the granularity of DAE. Furthermore, an Intersection over Union (IoU) aware FAR-Loss (IFL) is designed to improve accuracy of angle prediction using an adaptive re-weighting mechanism guided by IoU. Extensive experiments are performed on several public remote sensing datasets, which demonstrate the effectiveness of the proposed MGAR. Moreover, experiments on embedded devices demonstrate that the proposed MGAR is also friendly for lightweight deployments.

A Nonlinear Observability Analysis of Ambient Wind Estimation with Uncalibrated Sensors, Inspired by Insect Neural Encoding.

Estimating the direction of ambient fluid flow is key for many flying or swimming animals and robots, but can only be accomplished through indirect measurements and active control. Recent work with tethered flying insects indicates that their sensory representation of orientation, apparent wind, direction of movement, and control is represented by a 2-dimensional angular encoding in the central brain. This representation simplifies sensory integration by projecting the direction (but not scale) of measurements with different units onto a universal polar coordinate frame. To align these angular measurements with one another and the motor system does, however, require a calibration of angular gain and offset for each sensor. This calibration could change with time due to changes in the environment or physical structure. The circumstances under which small robots and animals with angular sensors and changing calibrations could self-calibrate and estimate the direction of ambient fluid flow while moving remains an open question. Here, a methodical nonlinear observability analysis is presented to address this. The analysis shows that it is mathematically feasible to continuously estimate flow direction and perform self-calibrations by adopting frequent changes in course (or active prevention thereof) and orientation, and requires fusion and temporal differentiation of three sensory measurements: apparent flow, orientation (or its derivative), and direction of motion (or its derivative). These conclusions are consistent with the zigzagging trajectories exhibited by many plume tracking organisms, suggesting that perhaps flow estimation is a secondary driver of their trajectory structure.

Demonstration of free-space one-to-many multicasting link from orbital angular momentum encoding.

Multicasting is necessary when distributing signals between multiple users. In this Letter, we demonstrate an orbital angular momentum (OAM) encoding-based free-space one-to-many multicasting link, where digital signals are encoded into a series of time-varying OAM states and transmitted from one transmitter to multiple receivers with various locations. Moreover, encoding N various signals simultaneously in one transmitter and sending them at the same time to N various receivers separately, is also demonstrated. As a proof-of-concept, four different gray images are coded by one transmitter simultaneously and multicast to four various receivers separately. The favorable decoding results returned by the four receivers show good multicasting performance.

Digitalized Geometric Phases for Parallel Optical Spin and Orbital Angular Momentum Encoding

Geometric phases have attracted considerable attention in recent years, due to their capability of arbitrary beam shaping in a most efficient and compact way, while traditional geometric phases are usually limited to handling single-structured beams and lack the capability of parallel manipulation. Here, we propose a digitalized geometric phase enabling parallel optical spin and orbital angular momentum encoding. The concept is demonstrated in inhomogeneous anisotropic media by imprinting a particularly designed binary phase into a space-variant geometric phase. We theoretically analyze its spin–orbit interaction of light and experimentally created higher-order Poincare sphere beam lattices, the order number and symmetry of which can be flexibly manipulated. Special lattices of cylindrical vector beams and orbital angular momentum modes with square and hexagonal symmetry are presented. This work discloses a new insight in programming geometric phases for tailoring the optical field and inspires various ph...

Method of ultrafast beam rotation for single-shot, time-resolved measurements.

Ultrafast optical beam rotation is proposed for single-shot, time-resolved measurements. A pump-probe configuration is considered using a diffraction grating and focusing optics to create angular encoding of the time delay between the pump and probe pulses. The characteristic time t(ap) of the grating-lens system is derived as a function of dispersion, NA, and time window T. An analytical equation for time resolution is obtained that incorporates t(ap)laser pulse width, and beam crossing, enabling optimum selection of optical components. For commercial standard gratings with width W≤50 mm, laser λ=800 nm, and NA=.05, a 160 ps time window can be achieved, and t(ap)=23 fs for T=1 ps.

Polarizability tensor retrieval for magnetic and plasmonic antenna design

A key quantity in the design of plasmonic antennas and metasurfaces, as well as metamaterials, is the electrodynamic polarizability of a single scattering building block. In particular, in the current merging of plasmonics and metamaterials, subwavelength scatterers are judged by their ability to present a large, generally anisotropic electric and magnetic polarizability, as well as a bi-anisotropic magnetoelectric polarizability. This bi-anisotropic response, whereby a magnetic dipole is induced through electric driving, and vice versa, is strongly linked to the optical activity and chiral response of plasmonic metamolecules. We present two distinct methods to retrieve the polarizibility tensor from electrodynamic simulations. As a basis for both, we use the surface integral equation (SIE) method to solve for the scattering response of arbitrary objects exactly. In the first retrieval method, we project scattered fields onto vector spherical harmonics with the aid of an exact discrete spherical harmonic Fourier transform on the unit sphere. In the second, we take the effective current distributions generated by SIE as a basis to calculate dipole moments. We verify that the first approach holds for scatterers of any size, while the second is only approximately correct for small scatterers. We present benchmark calculations, revisiting the zero-forward scattering paradox of Kerker et al (1983 J. Opt. Soc. Am. 73 765–7) and Alù and Engheta (2010 J. Nanophoton. 4 041590), relevant in dielectric scattering cancelation and sensor cloaking designs. Finally, we report the polarizability tensor of split rings, and show that split rings will strongly influence the emission of dipolar single emitters. In the context of plasmon-enhanced emission, split rings can imbue their large magnetic dipole moment on the emission of simple electric dipole emitters. We present a split ring antenna array design that is capable of converting the emission of a single linear dipole emitter in forward and backward beams of directional emission of opposite handedness. This design can, for instance, find application in the spin angular momentum encoding of quantum information.

Single-beam data encoding using a holographic angular multiplexing technique

We propose a novel configuration for angular multiplexing holographic encoding in which the signal beam and the reference beam are combined into a single beam. By using a spatial light modulator based on twisted nematic liquid crystals, the signal and the reference beams are modulated in amplitude mode and phase mode, respectively. The multiplexed interference patterns with the reference beams of different incident angles are recorded near the Fourier transform plane, and then the signals are selectively reconstructed by the corresponding reference beam. Both the simulation and the experiment of single-beam angular multiplexed holography are performed with consistent results. Compared with the traditional angular multiplexing holographic recording system, the single-beam configuration is more compact, easier to adjust, and less sensitive to the vibration of the environment. Therefore, it will be more attractive for potential applications in many fields, such as high-density signal recording and data encryption.

Experimental generation and characterization of single-photon hybrid ququarts based on polarization and orbital angular momentum encoding

High-dimensional quantum states, or qudits, represent a promising resource in the quantum information field. Here we present the experimental generation of four-dimensional quantum states, or ququarts, encoded in the polarization and orbital angular momentum of a single photon. Our novel technique, based on the q-plate device, allows to prepare and measure the ququart in all five mutually unbiased bases. We report the reconstruction of the four dimensional density matrix through the tomographic procedure for different ququart states.

Larmor procession reflectometry for studying multilayers

A new experimental method (angular encoding) based on the combination of reflectometry and neutron spin-echo methods, which makes it possible to use widely collimated neutron beams and provides an angular resolution of about 0.01°, has been proposed. It was used to study magnetic thin films. In particular, the effect of Larmor pseudoprecession of neutron polarization upon reflection from a 57Fe film with a thickness of about 100 nm was investigated, and it was shown that this method allows one to determine the absolute magnetization direction; this cannot be done using conventional reflectometry.

Angular encoding in attosecond recollision

We describe a general concept of using the spatial information encoded in the time-dependent polarization of high harmonic radiation generated by orthogonally polarized two-color laser fields. The main properties of recolliding electron wave packets driven by such fields are reviewed. It is shown that in addition to the recollision energy the angle of recollision of such wave packets, which is directly mapped onto the polarization direction of the emitted high harmonic radiation, varies on a sub-laser-cycle time-scale. Thus, a mapping between the polarization angle and the frequency of the emitted radiation is established on an attosecond time scale. While the polarization angle encodes the spatial properties of the recollision process, the frequency is linked to time via the well-known dispersion relations of high harmonic generation. Based on these principles, we show that in combination with polarization selective detection the use of orthogonally polarized drive pulses for high harmonic generation permit one to construct spatially resolved attosecond measurements. Here, we present two examples of possible applications: (i) a method for isolating a single attosecond pulse from an attosecond pulse train which is more efficient than the cut-off selection method, and (ii) a technique for orbital tomography of molecules with attosecond resolution.

Magnetic Field Inhomogeneity and Torque in High Temperature Superconducting Magnetic Bearings

We give experimental evidence for the connection between magnetic field inhomogeneity of a permanent magnet and torque on the rotor in a high temperature superconducting bearing. Spin-down measurements below 14 Hz are used to demonstrate a high degree of correlation between variations in the angular speed of the rotor within a single period of rotation with the measured spatial structure of the magnetic field of the rotor. At frequencies below /spl sim/1 Hz the fractional speed variation within a single period of rotation is inversely proportional to the square of the mean frequency of rotation. We propose that a dipole-dipole interaction gives rise to the torques that lead to speed variations and we show that this interaction explains the observed functional dependence on frequency. At frequencies above /spl sim/1 Hz the measured magnitude is about 1% of the mean frequency of rotation, consistent with the noise level in the experiment. The results imply that arcminute accuracy angular encoding of the rotor can be achieved with a single measurement of angle in each period.

Angular encoding with Larmor precession

The first test of angular encoding of a neutron beam with Larmor precession was performed at the spin-echo set-up of ZETA on IN3 (ILL, Grenoble, France). The polymer multilayer sample exhibits specular and off-specular scattering previously measured with reflectometry. The Yoneda scattering was discriminated from the specular reflection employing the angular encoding. The scans of the reflected and the off-specular scattered intensities as a function of the spin-echo length reveal its sinusoidal intensity dependence and the phase relation with respect to each other. The phase shift between the off-specular scattering and specular reflected intensity produced the wanted and necessary angular resolution. An estimation of the effective resolution abilities of this method in reflectometry kinematics yields that the angular encoding including subsequent deconvolution is useful for any divergent or convergent incoming or outgoing beam in order to separate signals.

Increased diffusion sensitivity with hyperechos.

It is shown that the introduction of a 180 degrees refocusing pulse into a standard diffusion weighted stimulated echo sequence is equivalent to the simplest hyperecho sequence with identical diffusion weighting but equal or greater signal-to-noise (SNR) and thus equal or greater diffusion contrast. For high b-value imaging, the hyperecho sequence thus possesses the high diffusion contrast in the presence of small T(1)/T(2) ratios characteristic of stimulated echo sequences but with less than the 50% loss in SNR that is associated with the stimulated echo. For low b-value imaging, the hyperecho signal converges to that of the standard spin echo. The advantages of the two-pulse diffusion weighted hyperecho sequence are demonstrated theoretically. Experimental results are shown in the application to high angular resolution diffusion encoding (HARD) in normal human brain.

A UHV compatible two-crystal monochromator for synchrotron radiation

For obvious reasons, monochromatic synchrotron radiation studies at soft X-ray energies require that the monochromator be ultra-high vacuum compatible. We will describe the design, testing and performance of a linkage based, two-crystal, non-dispersive monochromator for use at synchrotrns. The use of mechanical linkage reducesthe degrees of freedom and minimized the number of motion feedthroughs into the vacuum. The linkage chosen maintains a constant output beam position, and assures that the second crystal is always properly positioned. Pertinent design features include direct angular encoding and a feedback-controlled correction system to eliminate the effects of mechanical imperfections.