Gradient Boosted Decision Trees (GBDT’s) are a powerful tool for classification and regression tasks in Big Data. Researchers should be familiar with the strengths and weaknesses of current implementations of GBDT’s in order to use them effectively and make successful contributions. CatBoost is a member of the family of GBDT machine learning ensemble techniques. Since its debut in late 2018, researchers have successfully used CatBoost for machine learning studies involving Big Data. We take this opportunity to review recent research on CatBoost as it relates to Big Data, and learn best practices from studies that cast CatBoost in a positive light, as well as studies where CatBoost does not outshine other techniques, since we can learn lessons from both types of scenarios. Furthermore, as a Decision Tree based algorithm, CatBoost is well-suited to machine learning tasks involving categorical, heterogeneous data. Recent work across multiple disciplines illustrates CatBoost’s effectiveness and shortcomings in classification and regression tasks. Another important issue we expose in literature on CatBoost is its sensitivity to hyper-parameters and the importance of hyper-parameter tuning. One contribution we make is to take an interdisciplinary approach to cover studies related to CatBoost in a single work. This provides researchers an in-depth understanding to help clarify proper application of CatBoost in solving problems. To the best of our knowledge, this is the first survey that studies all works related to CatBoost in a single publication.

/pdf/catboost-for-big-data-an-interdisciplinary-review-utp91h5i0y.pdf

CatBoost for big data: an interdisciplinary review

Self-assembled peptide and protein nanostructures for anti-cancer therapy: Targeted delivery, stimuli-responsive devices and immunotherapy.

We examine the effectiveness of the ground-plane invisibility cloak generated from quasiconformal mapping of electromagnetic space. This cloak without anisotropy will generally lead to a lateral shift of the scattered wave, whose value is comparable to the height of the cloaked object, making the object detectable. This can be explained by the fact that the corresponding virtual space is thinner and wider than it should be. Ray tracing on a concrete model shows that, for a bump with a maximum height of 0.2 units to be hidden, the lateral shift of a ray with 45° incidence is around 0.15 units.

Lateral Shift Makes a Ground-Plane Cloak Detectable

Multi-bit graphene-based bias-encoded metasurfaces for real-time terahertz wavefront shaping: From controllable orbital angular momentum generation toward arbitrary beam tailoring

Recent years have witnessed an extraordinary spurt in attention toward the wave-manipulating strategies revealed by phase-amplitude metasurfaces. Recently, it has been shown that, when two different phase-encoded metasurfaces responsible for doing separate missions are added together based on the superposition theorem, the mixed digital phase distribution will realize both missions at the same time. In this paper, via a semi-analytical procedure, we demonstrate that such a theorem is not necessarily valid when using phase-only metasurfaces or ignoring the element pattern functions. We introduce the concept of asymmetric spatial power divider (ASPD) with arbitrary power ratio levels in which modulating both amplitude and phase of the meta-atoms is inevitable to fully control the power intensity pattern of a reflective metasurface. Numerical simulations illustrate that the proposed ASPD designed by proper phase and amplitude distribution over the surface can directly generate a desired number of beams with predetermined orientations and power budgets. The C-shaped Pancharatnam-Berry meta-atoms locally realize the optimal phase and amplitude distribution in each case, and the good conformity between simulations and theoretical predictions verifies the presented formalism. A prototype of our ASPD designs is also fabricated and measured, and the experimental results corroborate well our numerical and semi-analytical predictions. Our findings not only offer possibilities to realize arbitrary spatial power dividers over subwavelength scale but also reveal an economical and simple alternative for a beamforming array antenna.

https://pubs.acs.org/doi/pdf/10.1021/acsomega.9b02195

Asymmetric Spatial Power Dividers Using Phase-Amplitude Metasurfaces Driven by Huygens Principle.

https://iopscience.iop.org/article/10.1088/2040-8986/aacdc1/pdf

Experimental demonstration of metamaterial-assisted antenna beam deflection through folded transformation optics

In this letter, a novel bifunctional coating layer is proposed for a cylindrical antenna using transformation optics. This layer not only cloaks the omnidirectional antenna from an external source, but also enhances its directivity at the same time. To achieve this, an appropriate coordinate transformation is established so that the cylindrical wavefronts originating from the antenna would be transformed into plane waves. The material properties of the proposed cover are numerically derived by solving the Laplace's equation. Full-wave simulations are performed to verify both the directivity enhancement and cloaking functions of the designed coating layer. Thus, it can be effectively utilized in highly scattering multiple-antenna environments as a shielding mechanism.

Design of Coating Materials for Cloaking and Directivity Enhancement of Cylindrical Antennas Using Transformation Optics

The wall-effect compensation is one of the most annoying steps in modern through wall imaging (TWI) problems. Presence of wall not only increases the computational burden due to clutter rejection and complex imaging algorithms, but also blurs the reconstructed image. To mitigate this drawback, different wall-clutter rejection methods have been subject of numerous researches in recent years. However, these wall-clutter rejection methods mainly focus on removing the first strong reflection of wall. In this study, a novel wall-removal method (not only wall-clutter rejection) based on transformation optics (TO) is addressed. This is the first time that TO has been applied to TWI problem. Using this new method, TWI is cast to a free space imaging problem which is computationally more effective and results in more focused images. Real-time imaging is also more available in this case. Without loss of generality, a two-dimensional TWI configuration is assumed to evaluate the provided technique. A single point-like scatterer hidden behind a wall is used to obtain the imaging system point spread function resulted by the proposed technique. Finally, it is shown that the presented method is valid even for lossy walls. Reconstructed images show satisfactory focusing that highlight the capabilities of presented method.

https://ietresearch.onlinelibrary.wiley.com/doi/pdf/10.1049/iet-map.2016.0514

Removing the wall effects using electromagnetic complex coating layer for ultra-wideband through wall imaging

Coordinate transformation (CT) theory has shown great potentials in manipulating both time-varying and static fields for different physics ranging from electromagnetism and acoustics to electrostatic and thermal science. Nevertheless, as inhomogeneous and anisotropic materials are required to be realized for the implementation of CT-based devices, the applicability of this method is restricted due to difficulties in the fabrication process. In this paper, based on transformation electrostatic (TE) methodology, the design principle of an arbitrary shape dc electric concentrator is established which yields the enhancement of static electric fields in a predefined region with only one homogeneous conductivity, named as dc null medium (DNM). It is shown that one constant DNM is sufficient for localizing steady electric current in any arbitrary shape region, which in turn obviates the tedious mathematical calculations that conventional methods suffer from. In other words, the same DNM can be used for different concentrators regardless of their cross-section geometries, which makes the presented approach suitable for scenarios where reconfigurability is of utmost importance. Several numerical simulations are performed in order to demonstrate the capability of the proposed dc electric concentrator in localizing steady electric fields into the desired region. Moreover, by utilizing the analogy between electrically conducting materials and resistor networks, the attained DNM is realized with low-cost resistors and then exploited for fabricating a square shape dc electric concentrator on a printed circuit board (PCB). It is demonstrated that the measurement results agree well with the theoretical predictions and numerical simulations, which corroborate the effectiveness of the propounded method. The presented idea of this paper could find applications in scenarios where highly confined electric fields/currents are of critical importance such as electronic skin devices and electrical impedance tomography.

/pdf/experimental-demonstration-of-an-arbitrary-shape-dc-electric-n0qzgjnepz.pdf

Experimental demonstration of an arbitrary shape dc electric concentrator.

Based on the transformation acoustic methodology, an innovative approach for designing arbitrary shape concentrators is proposed. Unlike previous works, which utilized inhomogeneous and anisotropic materials to concentrate the incident acoustic waves in an arbitrary domain, the same functionality is attained by introducing only one homogeneous anisotropic medium, which is called the acoustic null medium (ANM). The other advantage of this method is that the obtained materials are not dependent on the shape of the concentrator. That is, regardless of the device geometry, a constant ANM will be used for each new shape and the output results do not alter. This will circumvent the conventional transformation acoustics' sophisticated and tedious calculations and could be easily implemented in real-life scenarios. To show these competencies, several numerical full-wave simulations are performed. In addition to the bulk material simulations, an arbitrary shape concentrator is realized via effective medium theory and it is shown that an ANM, as a kind of extreme medium, could be implemented with natural materials.

Mohammad Hosein Fakheri

Papers

Experimental demonstration of metamaterial-assisted antenna beam deflection through folded transformation optics

Design of Coating Materials for Cloaking and Directivity Enhancement of Cylindrical Antennas Using Transformation Optics

Removing the wall effects using electromagnetic complex coating layer for ultra-wideband through wall imaging

Experimental demonstration of an arbitrary shape dc electric concentrator.

Arbitrary Shaped Acoustic Concentrators Enabled by Null Media