model features
Recently Published Documents


TOTAL DOCUMENTS

320
(FIVE YEARS 124)

H-INDEX

25
(FIVE YEARS 4)

2022 ◽  
Vol 2022 ◽  
pp. 1-15
Author(s):  
Yan Zeng ◽  
Xin Wang ◽  
Junfeng Yuan ◽  
Jilin Zhang ◽  
Jian Wan

Federated learning is a new framework of machine learning, it trains models locally on multiple clients and then uploads local models to the server for model aggregation iteratively until the model converges. In most cases, the local epochs of all clients are set to the same value in federated learning. In practice, the clients are usually heterogeneous, which leads to the inconsistent training speed of clients. The faster clients will remain idle for a long time to wait for the slower clients, which prolongs the model training time. As the time cost of clients’ local training can reflect the clients’ training speed, and it can be used to guide the dynamic setting of local epochs, we propose a method based on deep learning to predict the training time of models on heterogeneous clients. First, a neural network is designed to extract the influence of different model features on training time. Second, we propose a dimensionality reduction rule to extract the key features which have a great impact on training time based on the influence of model features. Finally, we use the key features extracted by the dimensionality reduction rule to train the time prediction model. Our experiments show that, compared with the current prediction method, our method reduces 30% of model features and 25% of training data for the convolutional layer, 20% of model features and 20% of training data for the dense layer, while maintaining the same level of prediction error.


BioMed ◽  
2022 ◽  
Vol 2 (1) ◽  
pp. 13-26
Author(s):  
Avishek Chatterjee ◽  
Guus Wilmink ◽  
Henry Woodruff ◽  
Philippe Lambin

We conducted a systematic survey of COVID-19 endpoint prediction literature to: (a) identify publications that include data that adhere to FAIR (findability, accessibility, interoperability, and reusability) principles and (b) develop and reuse mortality prediction models that best generalize to these datasets. The largest such cohort data we knew of was used for model development. The associated published prediction model was subjected to recursive feature elimination to find a minimal logistic regression model which had statistically and clinically indistinguishable predictive performance. This model could still not be applied to the four external validation sets that were identified, due to complete absence of needed model features in some external sets. Thus, a generalizable model (GM) was built which could be applied to all four external validation sets. An age-only model was used as a benchmark, as it is the simplest, effective, and robust predictor of mortality currently known in COVID-19 literature. While the GM surpassed the age-only model in three external cohorts, for the fourth external cohort, there was no statistically significant difference. This study underscores: (1) the paucity of FAIR data being shared by researchers despite the glut of COVID-19 prediction models and (2) the difficulty of creating any model that consistently outperforms an age-only model due to the cohort diversity of available datasets.


Author(s):  
Vladimir Konyakhin ◽  
Marina Prokhorova ◽  
Anton Petrovsky

The authors study socio-psychological determinants of extremist criminal behavior of young people in Krasnodar Region within the framework of socio-economic, national and geographical specifics of the territory. The main goal was to identify and specify subjective (inner) causes as an aggregate of personal psychological features, needs, emotions, motives, specifics of conscience and volition that shape the intent and determine the qualitative side of extremist crimes. The authors used both general scientific (analysis, synthesis, induction, deduction, etc.) and special (statistical and specific-sociological) methods of cognition. The dominant source of information was results of a questionnaire survey of 146 young people who were residents of Krasnodar Region aged 18–24 with the same level of education; they were university students (57 %), and students of vocational schools and colleges (28 %). The obtained data were used for SPSS (Statistical Package for the Social Sciences) analysis, which identified regularities in the formation of the public opinion typical for young people in Krasnodar Region. Besides, the authors established a factor commonality which served as a logical proof that psychological patterns typical of the youth environment, stereotypes and models of behavior act as determinants of extremist actions. All of these allowed the authors to state that there are a number of negative trends, such as the mental acceptance of some extremist actions by young people, especially actions against people of a different race, nationality, religion; this acceptance is common for 20 % of people aged 18 to 24; young people do not know about 5 out of 13 types of extremist activities included in the federal legislation; extremist information is easily available on the Internet. To eliminate these trends, the authors suggest a number of measures: activization of legal information campaign; identification of students who are highly likely to commit extremist actions; creation of a system of model features of extremist behavior to be used in the preventive work in educational establishments of Krasnodar Region; designing and teaching, on the regional level, a special subject of preventive nature to high school, college and university students.


Sensors ◽  
2021 ◽  
Vol 21 (23) ◽  
pp. 8069
Author(s):  
Qinghua Sheng ◽  
Haixiang Sheng ◽  
Peng Gao ◽  
Zhu Li ◽  
Haibing Yin

Currently, the target detection based on convolutional neural network plays an important role in image recognition, speech recognition and other fields. However, the current network model features a complex structure, a huge number of parameters and resources. These conditions make it difficult to apply in embedded devices with limited computational capabilities and extreme sensitivity to power consumption. In this regard, the application scenarios of deep learning are limited. This paper proposes a real-time detection scheme for cook assistant overalls based on the Hi3559A embedded processor. With YOLOv3 as the benchmark network, this scheme fully mobilizes the hardware acceleration resources through the network model optimization and the parallel processing technology of the processor, and improves the network reasoning speed, so that the embedded device can complete the task of real-time detection on the local device. The experimental results show that through the purposeful cropping, segmentation and in-depth optimization of the neural network according to the specific processor, the neural network can recognize the image accurately. In an application environment where the power consumption is only 5.5 W, the recognition speed of the neural network on the embedded end is increased to about 28 frames (the design requirement was to achieve a recognition speed of 25 frames or more), so that the optimized network can be effectively applied in the back kitchen overalls identification scene.


2021 ◽  
Vol 13 (23) ◽  
pp. 13059
Author(s):  
Wenliang Zhou ◽  
Mehdi Oldache

In order to improve train operation planning from the two perspectives of enterprise operating costs and passengers’ travel time, this paper proposes an integrated optimization model of three sub-problems, namely line planning, timetabling and rolling stock allocation for urban railway transit lines based on passengers’ travelling demands and the constraints of the urban rail line. The model features dwelling time at stations, turnaround operations at terminal stations, entering/exiting depot operations and an assignment for passengers’ travelling flow. We propose a solution method based on a metaheuristic method that simulates annealing to generate an optimal solution for the overall problem using MATLAB. Finally, we use the example of Xi’an metro line one to demonstrate the performance of the model.


2021 ◽  
Vol 12 ◽  
Author(s):  
Christian Panitz ◽  
Dominik Endres ◽  
Merle Buchholz ◽  
Zahra Khosrowtaj ◽  
Matthias F. J. Sperl ◽  
...  

Expectations are probabilistic beliefs about the future that shape and influence our perception, affect, cognition, and behavior in many contexts. This makes expectations a highly relevant concept across basic and applied psychological disciplines. When expectations are confirmed or violated, individuals can respond by either updating or maintaining their prior expectations in light of the new evidence. Moreover, proactive and reactive behavior can change the probability with which individuals encounter expectation confirmations or violations. The investigation of predictors and mechanisms underlying expectation update and maintenance has been approached from many research perspectives. However, in many instances there has been little exchange between different research fields. To further advance research on expectations and expectation violations, collaborative efforts across different disciplines in psychology, cognitive (neuro)science, and other life sciences are warranted. For fostering and facilitating such efforts, we introduce the ViolEx 2.0 model, a revised framework for interdisciplinary research on cognitive and behavioral mechanisms of expectation update and maintenance in the context of expectation violations. To support different goals and stages in interdisciplinary exchange, the ViolEx 2.0 model features three model levels with varying degrees of specificity in order to address questions about the research synopsis, central concepts, or functional processes and relationships, respectively. The framework can be applied to different research fields and has high potential for guiding collaborative research efforts in expectation research.


2021 ◽  
Author(s):  
◽  
Nicolas Eduard Alype Brikke

<p>The three-dimensional (3D) evolution of the Australian-Pacifi c late boundary in the central South Island of New Zealand is investigated by analysing seismic data from the South Island GeopHysical Transect (SIGHT) project and by using a novel 3D tomography inversion method, FMTOMO. A 380 km-long, 350 km-wide and 56 km-deep 3D tomography image of the P-wave velocity structure and interface geometry of the crust and upper-mantle is constructed by inverting for 164,048 traveltime picks. The picks are both coincident (in-line) and oblique (cross-line) to the survey geometry. The traveltime picks and station elevations were static corrected and reduced to basement level, respectively, to eliminate the highly variable sedimentary component of the inversion process. Synthetic testing of the model space was carried out to help the interpretation of the solution model features. Some model features are consistent with previous results. Usual crustal velocities (5.5 km/s close to the surface and 6.3 km/s at the bottom of the crust) are found at distal ends of the collision zone. Lower velocities (5.7 km/s) intrude the mid-crust of the Australian plate to depths of about 20 km, which is consistent with the downward  flexure of the Australian plate. A low velocity zone (5.9 - 6.1 km/s) is situated to the southeast of the Alpine fault, which is consistent with the Alpine fault low velocity zone. Furthermore, a high-velocity body (6.3 km/s) is observed in the top 10 km of the upper-crust immediately above the thickened crust between the west coast of the South Island and the Main Divide of the Southern Alps. This body is interpreted as a drier, more rigid body of schist. A zone of low velocity (5.8 km/s reaching 8 km depth) is observed immediately to the southeast of the aforementioned high velocity body. The feature is interpreted as a back-shearing faulting structure through which fluid escape towards the surface. A flexural analysis of an apparent  flexure profile of the Australian Plate along SIGHT line 01 yielded a  flexural parameter, a, of 89 km, an elastic thickness, Te, of 14 km and a  flexural rigidity, D, of 1.5 : 10^(23) N.m. These results are consistent with results of a  flexural analysis of SIGHT line 02W [Harrison 1999]. The following features are derived from the solution model. An apparent gradient in uppermantle anisotropy is observed with seismic velocities increasing towards the south of the model. Also, the geometry of the Mohorovicic discontinuity is apparently smooth between the two main SIGHT transects. The tomography method used in this project proves to be complementary to other coarser-scale and finer-scale seismic studies of the region in that it brings out features that were not seen by them. Notwithstanding that the interface inversion process remains to be perfected in the software, the velocity inversion produced a satisfactory solution model.</p>


2021 ◽  
Author(s):  
◽  
Nicolas Eduard Alype Brikke

<p>The three-dimensional (3D) evolution of the Australian-Pacifi c late boundary in the central South Island of New Zealand is investigated by analysing seismic data from the South Island GeopHysical Transect (SIGHT) project and by using a novel 3D tomography inversion method, FMTOMO. A 380 km-long, 350 km-wide and 56 km-deep 3D tomography image of the P-wave velocity structure and interface geometry of the crust and upper-mantle is constructed by inverting for 164,048 traveltime picks. The picks are both coincident (in-line) and oblique (cross-line) to the survey geometry. The traveltime picks and station elevations were static corrected and reduced to basement level, respectively, to eliminate the highly variable sedimentary component of the inversion process. Synthetic testing of the model space was carried out to help the interpretation of the solution model features. Some model features are consistent with previous results. Usual crustal velocities (5.5 km/s close to the surface and 6.3 km/s at the bottom of the crust) are found at distal ends of the collision zone. Lower velocities (5.7 km/s) intrude the mid-crust of the Australian plate to depths of about 20 km, which is consistent with the downward  flexure of the Australian plate. A low velocity zone (5.9 - 6.1 km/s) is situated to the southeast of the Alpine fault, which is consistent with the Alpine fault low velocity zone. Furthermore, a high-velocity body (6.3 km/s) is observed in the top 10 km of the upper-crust immediately above the thickened crust between the west coast of the South Island and the Main Divide of the Southern Alps. This body is interpreted as a drier, more rigid body of schist. A zone of low velocity (5.8 km/s reaching 8 km depth) is observed immediately to the southeast of the aforementioned high velocity body. The feature is interpreted as a back-shearing faulting structure through which fluid escape towards the surface. A flexural analysis of an apparent  flexure profile of the Australian Plate along SIGHT line 01 yielded a  flexural parameter, a, of 89 km, an elastic thickness, Te, of 14 km and a  flexural rigidity, D, of 1.5 : 10^(23) N.m. These results are consistent with results of a  flexural analysis of SIGHT line 02W [Harrison 1999]. The following features are derived from the solution model. An apparent gradient in uppermantle anisotropy is observed with seismic velocities increasing towards the south of the model. Also, the geometry of the Mohorovicic discontinuity is apparently smooth between the two main SIGHT transects. The tomography method used in this project proves to be complementary to other coarser-scale and finer-scale seismic studies of the region in that it brings out features that were not seen by them. Notwithstanding that the interface inversion process remains to be perfected in the software, the velocity inversion produced a satisfactory solution model.</p>


2021 ◽  
Vol 15 ◽  
Author(s):  
Carli L. Poisson ◽  
Liv Engel ◽  
Benjamin T. Saunders

Addiction is a complex disease that impacts millions of people around the world. Clinically, addiction is formalized as substance use disorder (SUD), with three primary symptom categories: exaggerated substance use, social or lifestyle impairment, and risky substance use. Considerable efforts have been made to model features of these criteria in non-human animal research subjects, for insight into the underlying neurobiological mechanisms. Here we review evidence from rodent models of SUD-inspired criteria, focusing on the role of the striatal dopamine system. We identify distinct mesostriatal and nigrostriatal dopamine circuit functions in behavioral outcomes that are relevant to addictions and SUDs. This work suggests that striatal dopamine is essential for not only positive symptom features of SUDs, such as elevated intake and craving, but also for impairments in decision making that underlie compulsive behavior, reduced sociality, and risk taking. Understanding the functional heterogeneity of the dopamine system and related networks can offer insight into this complex symptomatology and may lead to more targeted treatments.


Author(s):  
G. Purna Chandar Rao ◽  
V. B. Narasimha

A social media adoption is important to provide content authenticity and awareness for the unknown news that might be fake. Therefore, a Natural Language Processing (NLP) model is required to identify the content properties for language-driven feature generation. The present research work utilizes language-driven features that extract the grammatical, sentimental, syntactic, readable features. The feature from the particular news content is extracted to deal with the dimensional problem as the language level features are quite complex. Thus, the Dropout layer-based Long Short Term Network Model (LSTM) for sequential learning achieved better results during fake news detection. The results obtained validate the important features extracted linguistic model features and are combined to achieve better classification accuracy. The proposed Drop out based LSTM model obtained accuracy of 95.3% for fake news classification and detection when compared to the sequential neural model for fake news detection.


Sign in / Sign up

Export Citation Format

Share Document