Set Of Prototypes Research Articles

Deep Metric Learning for K Nearest Neighbor Classication

KNN has gained popularity in machine learning due to its simplicity and good performance. However, kNN faces two problems with classification tasks. The first is that an appropriate distance measurement is required to compute distances between test sample and training samples. The other is the highly computational complexity due to the requirement of searching the nearest neighbors in the whole training data. In order to mitigate these two problems, we propose a novel method named KCNN to enhance the performance of kNN. KCNN uses convolutional neural networks to learn a suitable distance metric as well as prototype reduction to learn a reduced set of prototypes which can represent the original set. It has several superiorities compared with related methods. The combination of CNN and kNN empowers it to extract discriminative hierarchical features with which kNN can easily classify. KCNN learns spatial information on an image instead of considering it as a vector to learn distance metric. Moreover, KCNN simultaneously learns a reduced set of prototypes, which help improve classification efficiency and avoid noisy samples of the massive training set. The proposed method has a better robustness and convergence than CNN, especially when projecting input data into a low-dimension space.

Developing Prototype Simulants for Surface Materials and Morphology of Near Earth Asteroid 2016 HO3

There are a variety of applications for asteroid simulants in asteroid studies for science advances as well as technology maturation. For specific purpose, it usually requires purpose-specialized simulant. In this study, we designed and developed a set of prototype simulants as S-type asteroid surface materials analogue based on H, L, and LL ordinary chondrites’ mineralogy and terrestrial observations of near-earth asteroid 2016 HO3, which is the Chinese sample return mission target. These simulants are able to simulate morphology and reflectance characteristics of asteroid (469219) 2016 HO3 and, thus, to be used for engineering evaluation of the optical navigation system and the sampling device of the spacecraft during the mission phase. Meanwhile, these prototype simulants are easily to modify to reflect new findings on the asteroid surface when the spacecraft makes proximate observations.

Histogram Entropy Representation and Prototype Based Machine Learning Approach for Malware Family Classification

The number of malware has steadily increased as malware spread and evasion techniques have advanced. Machine learning has contributed to making malware analysis more efficient by detecting various behavioral and evasion patterns. However, when analyzing large-scale malware datasets, malware analysis through learning models has both high temporal and spatial complexity. In order to address these problems, this work proposes a low-dimensional feature using histogram entropy and a prototype selection algorithm using hyperrectangles. The low-dimensional feature forms an $L \times 256$ map according to the preselected parameter $L$ . The prototype selection algorithm divides the input space into overlapping subspaces where each subspace is decided by its hyperrectangle that becomes a prototype in the same class. A set cover optimization algorithm is employed to select a small number of prototypes that construct a new training dataset. A set of prototypes selected by the prototype selection algorithm is used to classify malware families. The experiment compares the performance of machine learning models for the histogram entropy feature using both the BIG 2015 dataset and the collected dataset. The integrated approach is evaluated using learning algorithms, such as Decision Tree, Random Forest, XGBoost, and CNN. The experimental results indicate that learning models perform competitively when compared to the entire dataset, while the proposed selection approach benefits from smaller datasets and lower time complexity.

Applying the new conicotomy kit is the key to solve the problem of asphyxia at the prehospital stage

The purpose of this study is to analyze the incidence of asphyxia, therapeutic measures, the outcomes of treatment of the wounded with impaired secure airway and to assess the effectiveness of the prototype: «one-time set to eliminate asphyxia (conicotomy)» UK-02 (state defense order LLC «NOVOPLAST-M»). The basis of the analysis was the material of 2047 case histories-clinical test data on armed conflicts of Military Field Surgery Department of the Military Medical Academy named after S. M. Kirov. Development and evaluation of the effectiveness of the prototype set of UK-02 to eliminate asphyxia in the course of experimental studies was carried out on 6 biological objects (pigs). The incidence of asphyxia is 1.5% according to national data and 8-13% according to the results of studies by foreign authors. Complete and service equipment-a one-time set of UK-02 (to eliminate asphyxia (conicotomy) will reduce prehospital mortality and significantly improve the outcomes of treatment of the wounded with severe damage to organs and tissues of the maxillofacial region, ENT organs and pathological processes accompanied by impaired secure airway.

A Study on the Prototype Setting for Energy Independent Site Planning

A Study on the Prototype Setting for Energy Independent Site Planning

On the generation of multi-label prototypes

Data reduction techniques play a key role in instance-based classification to lower the amount of data to be processed. Prototype generation aims to obtain a reduced training set in order to obtain accurate results with less effort. This translates into a significant reduction in both algorithms’ spatial and temporal burden. This issue is particularly relevant in multi-label classification, which is a generalization of multiclass classification that allows objects to belong to several classes simultaneously. Although this field is quite active in terms of learning algorithms, there is a lack of data reduction methods. In this paper, we propose several prototype generation methods from multi-label datasets based on Granular Computing. The simulations show that these methods significantly reduce the number of examples to a set of prototypes without significantly affecting classifiers’ performance.

Prototype REBCO Z1 and Z2 shim coils for ultra high-field high-temperature superconducting NMR magnets

We present promising results of novel high-temperature superconducting (HTS) shim coil prototypes that circumvent the size and strength limitation of our earlier innovative HTS shim concept based on 46-mm wide REBCO tape. The HTS shim coil is placed inside the HTS magnet, mainly for ultra-high-field (> 1 GHz or 23.5 T) NMR magnets, and thus unaffected from the windings’ diamagnetic wall effects. One full-scale version will be applied to clean up Z1 and Z2 harmonic errors in the MIT 1.3-GHz high-resolution NMR magnet composed of an 835-MHz HTS insert, while another version for an MIT 1-GHz microcoil NMR magnet whose small-scale model we are currently building. The prototype sets were wound with a 2-pile, 1.03-mm wide, 0.30-mm thick REBCO conductor. Operated at 77 K, the Z1 shim set generated a 1st harmonic field strength of 179 kHz/cm at 70 A, while the Z2 shim set, composed of two pairs, Z21 and Z22, generated the 2nd harmonic field of 141 kHz/cm2 at 50 A. Together with discussion on technical challenges for this REBCO shim coil concept, we demonstrate its feasibility for the next generation of ultra-high-field (UHF) HTS NMR magnets.

Experimental and Theoretical Dynamic Investigation of MEMS Polymer Mass-Spring Systems

This paper presents the experimental, analytical and numerical dynamic analysis of SU-8 polymer MEMS mass-spring systems using two different attachment beams, dual and quad. Beam thicknesses of $4.48~\mu \text{m}$ and $8.21~\mu \text{m}$ produce two different sets of prototypes using dual-beam attachments. Those using quad-beam are only $8.21~\mu \text{m}$ thick. Using Lazer Doppler Vibrometer (LDV), dynamic characterization is performed to measure respective resonance frequencies of 1325 Hz and 1813 Hz for dual and quad-beam attachments having a thickness of $8.21~\mu \text{m}$ . Numerical and analytical dynamic analysis are performed to investigate the frequency response of systems using both attachment structures. Using measured key geometric dimensions, numerical and analytical resonance frequencies are evaluated. Theoretical and experimental values are in good agreement with maximum errors of 16% and 12% for dual and quad-beam structures, respectively. The low-cost microfabrication process can be used to incorporate a piezoresistive material/nanomaterial in locations of maximum stress on the beams producing dual and quad-beam MEMS piezoresistive sensors. Numerical analysis is performed to show that the resonance frequency is not affected which proves that the developed dynamic analysis stays valid.

Analysis of Oil Spill Pollution in the Sea Based on Hyperspectral Laser Radar

Li, J.; Che, Y.; Song, N.; Chen, D., and Li, J., 2020. Analysis of oil spill pollution in the sea based on hyperspectral laser radar. In: Yang, D.F. and Wang, H. (eds.), Recent Advances in Marine Geology and Environmental Oceanography. Journal of Coastal Research, Special Issue No. 108, pp. 156–160. Coconut Creek (Florida), ISSN 0749-0208.The oceans, which account for 71% of the earth's surface area, play a very important role in the development of human society and provide rich resources for mankind. The pollution from oil spills on the sea surface will cause great harm to marine natural ecology and marine living resources. In this paper, according to the characteristics and harms of petroleum pollutants, combined with the development trend of domestic and foreign super-continuous spectrum LIDAR technology, a set of hyperspectral LIDAR prototypes for monitoring oil spill pollution in the sea is designed. The LIDAR can cover the band range of 400 nm∼1100 nm and measure the spectral irradiance of oil spills on the surface of water with different thicknesses at daytime and at night, respectively. According to the experimental data, the super-continuous spectrum LIDAR prototype can be well-qualified for the monitoring of oil spills on the sea surface at night. The experimental data show that the LIDAR prototype can be applied to the monitoring of oil spill pollution.

A Measurement Model of Value of Data for Decision Making in the Digital Era

Despite burgeoning opportunities for data-driven decisions, research shows that decision makers are failing to make sense of data within a broader context of organisational change which presents the following pertinent questions: 1) How can decision makers measure the value of data by giving a holistic account?; 2) How should the organisation-specific blending of machine and human rationality be factored in the measurement model? This study tackles these questions by proposing a novel approach that combines system dynamics and the ability to incorporate data science methods. In addition to a conceptual description, this paper also describes a feasibility test conducted on a small-scale prototype set in a supply chain context. The results show that the use of sophisticated models that have a local scope ('locally rational') might have unintended global consequences. It underscores the need for a holistic model in the decision makers' toolkit providing the ability to run targeted simulations to assess digital investments.

An analytical approach of designing compact microstrip patch antenna using metal-ring superstrate for wideband and broadside radiations

A compact design of microstrip patch antenna is presented to realize wideband characteristics. An annular metal-ring has been used as a superstrate as well as an inductive load to the rectangular patch. Robust analytical technique is applied to choose the dimensions of the ring and the patch where they resonate very close to each other. This strategic approach ensures a wide matching bandwidth with a broadside radiation pattern. An additional enhancement of bandwidth has been achieved by employing a unique design strategy. A set of antenna prototype has been fabricated for experimental validation. As much as 34% matching bandwidth has been revealed experimentally using a compact antenna (~λ × λ). The boresight gain is maintained about 10 dBi over the operating frequency band. The cross-polarized isolation in E- and H-planes are maintained above 40 dB and 30 dB respectively over the operating bandwidth. The back-radiation of the antenna is also maintained below −20 dB over the wide matching bandwidth.

NOVA Mobility Assistive System: Developed and Remotely Controlled with IOPT-Tools

In this paper, a Mobility Assistive System (NOVA-MAS) and a model-driven development approach are proposed to support the acquisition and analysis of data, infrastructures control, and dissemination of information along public roads. A literature review showed that the work related to mobility assistance of pedestrians in wheelchairs has a gap in ensuring their safety on road. The problem is that pedestrians in wheelchairs and scooters often do not enjoy adequate and safe lanes for their circulation on public roads, having to travel sometimes side by side with vehicles and cars moving at high speed. With NOVA-MAS, city infrastructures can obtain information regarding the environment and provide it to their users/vehicles, increasing road safety in an inclusive way, contributing to the decrease of the accidents of pedestrians in wheelchairs. NOVA-MAS not only supports information dissemination, but also data acquisition from sensors and infrastructures control, such as traffic light signs. For that, it proposed a development approach that supports the acquisition of data from the environment and its control while using a tool framework, named IOPT-Tools (Input-Output Place-Transition Tools). IOPT-Tools support controllers’ specification, validation, and implementation, with remote operation capabilities. The infrastructures’ controllers are specified through IOPT Petri net models, which are then simulated using computational tools and verified using state-space-based model-checking tools. In addition, an automatic code generator tool generates the C code, which supports the controllers’ implementation, avoiding manual codification errors. A set of prototypes were developed and tested to validate and conclude on the feasibility of the proposals.

YOLACT++ Better Real-Time Instance Segmentation.

We present a simple, fully-convolutional model for real-time ( fps) instance segmentation that achieves competitive results on MS COCO evaluated on a single Titan Xp, which is significantly faster than any previous state-of-the-art approach. Moreover, we obtain this result after training on only one GPU. We accomplish this by breaking instance segmentation into two parallel subtasks: (1) generating a set of prototype masks and (2) predicting per-instance mask coefficients. Then we produce instance masks by linearly combining the prototypes with the mask coefficients. We find that because this process doesn't depend on repooling, this approach produces very high-quality masks and exhibits temporal stability for free. Furthermore, we analyze the emergent behavior of our prototypes and show they learn to localize instances on their own in a translation variant manner, despite being fully-convolutional. We also propose Fast NMS, a drop-in 12 ms faster replacement for standard NMS that only has a marginal performance penalty. Finally, by incorporating deformable convolutions into the backbone network, optimizing the prediction head with better anchor scales and aspect ratios, and adding a novel fast mask re-scoring branch, our YOLACT++ model can achieve 34.1 mAP on MS COCO at 33.5 fps, which is fairly close to the state-of-the-art approaches while still running at real-time.

Initializing k-means Clustering by Bootstrap and Data Depth

The k-means algorithm is widely used in various research fields because of its fast convergence to the cost function minima; however, it frequently gets stuck in local optima as it is sensitive to initial conditions. This paper explores a simple, computationally feasible method, which provides k-means with a set of initial seeds to cluster datasets of arbitrary dimensions. Our technique consists of two stages: firstly, we use the original data space to obtain a set of prototypes (cluster centers) by applying k-means to bootstrap replications of the data and, secondly, we cluster the space of centers, which has tighter (thus easier to separate) groups, and search the deepest point in each assembled cluster using a depth notion. We test this method with simulated and real data, compare it with commonly used k-means initialization algorithms, and show that it is feasible and more efficient than previous proposals in many situations.

A novel method to assess intraarticular screw penetration into the joint space.

Objectives This study aims to propose a novel method to detect articular penetration of screws by relying on their electrical conductivity properties and control the validity of this method.Materials and methods In this ex vivo study, conducted between June 2017 and August 2017, we used five fresh sheep shoulder joints. First, the shoulder joint space was filled with saline solution. An insulated cannula was placed in the joint capsule, and a conductive wire was introduced into the joint via this cannula. A single titanium screw was inserted from the tuberculum majus into the posteroinferior quadrant of the humeral head under fluoroscopic observation. Conductivity was continuously measured using a digital multimeter. When a sudden decrease in conduction resistance was detected, fluoroscopic images were obtained in the anteroposterior (AP) and lateral directions. These images were assessed for penetration by a blinded surgeon. Penetration was confirmed by dissection of the joint.Results There was a significant decrease in electrical resistance when screw penetration occurred (p<0.001). All penetration events were confirmed using our novel method. For all five of the specimens, either AP or lateral images could not be used to confirm penetration. For two of these specimens, penetration was undetectable in both AP and lateral fluoroscopic images, but a decrease in resistance was recorded.ConclusionThe described method exhibits greater sensitivity and accuracy for metal penetration to joint, and it is effective in detecting screws in the joints. The novel method described in this paper was applied in a prototype setting, and we believe that this concept can continue to be developed.

Wideband Design of Circular Microstrip Patch Antenna using Rectangular Metal Sheet Superstrate for X-Band Applications

A wideband circular microstrip patch antenna (CMPA) has been presented employing a rectangular metal sheet superstrate. The proposed concept follows a unique, simple, and a flexible design approach to enhance the bandwidth of a circular patch. A simple change in the conventional antenna geometry has been suggested by adding a rectangular metal sheet superstrate, placed symmetrically above the patch. A cylindrical shaped foam spacer has been used to provide mechanical support to the optimized superstrate. The proposed antenna offers about 36% of impedance matching bandwidth ranging between 8.46 GHz to 12.06 GHz with a total bandwidth of 3.6 GHz. Whereas, a conventional circular patch, resonating at 9.96 GHz, hardly shows about 4.8% of impedance bandwidth (480 MHz) only. In addition to the enhanced bandwidth characteristics, the proposed antenna, also reveals a little increase in the gain throughout the operating frequency band. For the experimental validation, a set of antenna prototype has been fabricated using the commercially available dielectric substrate. The measured result is very closely agreed with the simulated predictions.

Novel Low-Cost, Low-Fidelity Hemorrhoidectomy Task Trainers

Develop and describe a set of low-cost hemorrhoidectomy task trainer prototypes in the setting of inadequate junior resident surgical skill preparation for anorectal cases. This is a study comparing expert and novice performance and opinions. Three task trainers were developed to simulate dissecting, knot-tying, and suturing in a confined space, like the anus. Participants were asked to dissect the peel off of an orange, tie seven 2-handed knots on a weight, and close a defect in a piece of felt with a running stitch. An 8-oz mason jar was used to simulate the confined space. Participants were asked to fill out a 5-point Likert-based evaluation regarding the skills. The primary outcome was time to complete each task in seconds. Secondary outcome measures were number of errors associated with each task, subjective achievability of tasks, and utility of tasks for improving surgical skills. General surgery residency program at a safety-net academic center. Forty subjects participated in this study. There were 20 experts (7 attending surgeons, 13 PGY-1-PGY-5 surgical residents) and 20 novices (11 third- and 9 fourth-year medical students). Experts knot-tied (59s vs 140s, p < 0.001) and sutured (219s vs 295s, p < 0.001) faster than novices. Experts were able to tie 7 knots in fewer attempts than novices (p < 0.001). There was no significant difference in speed of orange dissection between groups. There were no significant differences in the number or frequency of other errors. All participants felt the tasks were achievable (4.90/5) and would be useful in improving skills (4.93/5). This study demonstrated that a set of low-cost, low-fidelity prototypical hemorrhoidectomy task trainers can discriminate between experts and novices. Simulation models such as these can offer useful practice opportunities for junior general surgery trainees.

Online semi-supervised learning with learning vector quantization

Online semi-supervised learning (OSSL) is a learning paradigm simulating human learning, in which the data appear in a sequential manner with a mixture of both labeled and unlabeled samples. Despite the recent advances, there are still many unsolved problems in this area. In this paper, we propose a novel OSSL method based on learning vector quantization (LVQ). LVQ classifiers, which represent the data of each class by a set of prototypes, have found their usage in a wide range of pattern recognition problems and can be naturally adapted to the online scenario by updating the prototypes with stochastic gradient optimization. However, most of the existing LVQ algorithms were designed for supervised classification. To extract useful information from unlabeled data, we propose two simple and computationally efficient methods based on clustering assumption. To be specific, we use the maximum conditional likelihood criterion for updating prototypes when data sample is labeled, and the Gaussian mixture clustering criterion or neural gas clustering criterion for adjusting prototypes when data sample is unlabeled. These two criteria are utilized alternatively according to the availability of label information to make full use of both supervised and unsupervised data to boost the performance. By extensive experiments, we show that the proposed method exhibits higher accuracy compared with the baseline methods and graph-based methods and is much more efficient than graph-based methods in both training and test time.

Local wisdom value’s-based literacy education learning model in elementary school

The lack of a model for promoting productive learning environments and a clear major value of choice to guide practice has resulted in the needs for this research. Values are important aspects of learning, yet they are not used to their fullest potential in educational practices in Indonesia. The primary objective for this study was to propose an appropriate model with application of local wisdom value in elementary school, and synthesized strategies for challenging the value for sustainable use in schools. This paper discusses and describe possible guidelines targeting the design of learning local wisdom value’s-based learning model with creative learning environments that influence student literacy skills. This research was an Educational Design Research. To develop the local wisdom value’s-based learning model we need, the following five steps will be taken 1) putting several available models side by side (literature research), 2) comparing and contrasting the models (plenary study), 3) deciding most essential model components (ethnography analysing), 4) creating theoretical model and prototype (setting by Focus Group Discussion/FGD), 5) testing the prototype models with limited empirical verification sample (setting by workshop). At present, this research has been in the stage of compiling a prototype of the local wisdom value’s-based literacy learning model. The next stage will be consolidated through a workshop. The outcomes of the focus group interview and future workshop activities will be synthesized into five design guidelines providing orientation for the design of a local wisdom value’s-based literacy learning model. The results of this research recommend that the learning with the local wisdom value can be used as an approach to learning literacy. The model also promotes a positive classroom climate, a community of learners, developing learner responsibility, and maximizing opportunities for learning.

Abstract 19: Default Mode Network Alterations in Cerebral Amyloid Angiopathy: A Resting State FMRI Study

Introduction: Cerebral amyloid angiopathy (CAA) is an established cause of intracerebral hemorrhage and vascular dysfunction leading to ischemia. Functional connectivity analysis using MRI is becoming an important tool to analyze the brain activity during resting state, the default mode network (DMN) representing the prototypical set of connections. As CAA pathology has a posterior predominance, we sought to characterize the functional connectivity of the posterior DMN at resting state in patients with CAA. Methods: Patients with probable CAA diagnosed using Boston Criteria and healthy controls (HC) were prospectively enrolled and received high resolution 3T MRI scans including dedicated resting-state fMRI sequences. Functional seed-to-seed analyses were done using the default processing pipeline in the CONN Toolbox. Correlation maps between the established DMN and specific regions of the posterior DMN, the precuneus and posterior cingulate, were averaged within groups and compared in an ANCOVA model. Results: Study participants consisted of 60 patients with probable CAA and 20 healthy controls [aged 69 ± 7.5 vs 72.3 ± 8 years, P = 0.108]. Seed-to-seed analysis revealed a significantly lower strength of DMN connectivity in CAA when compared to controls in the precuneus [ P = 0.009] and posterior cingulate [ P = 0.003] adjusted for age and sex (Fig 1). Conclusion: Patients with CAA exhibited significant loss of connectivity in the posterior regions of the DMN when compared to controls. The precuneus and posterior cingulate are core regions of the DMN with reportedly high metabolic rates at rest. Disruption of these posterior DMN regions might occur due to vascular amyloid pathology that shows a predominantly posterior distribution.