It may be time to improve the neuron of artificial neural network

<p>Artificial neural networks (ANNs) have won numerous contests in pattern recognition, machine learning, or artificial intelligence in recent years. The neuron of ANNs was designed by the stereotypical knowledge of biological neurons 70 years ago. Artificial Neuron is expressed as f(wx+b) or f(WX). This design does not consider dendrites' information processing capacity. However, some recent studies show that biological dendrites participate in the pre-calculation of input data. Concretely, biological dendrites play a role in extracting the interaction information among inputs (features). Therefore, it may be time to improve the neuron of ANNs. According to our previous studies (DD), this paper adds the dendrites' function to artificial Neuron. The dendrite function can be expressed as W^i,i-1A^i-1 ○ A^{0|1|2|...|i-1} . The generalized new neuron can be expressed as f(W(W^i,i-1A^i-1 ○ A^{0|1|2|...|i-1})).The simplified new neuron be expressed as f(∑(WA ○ X)). After improving the neuron, there are so many networks to try. This paper shows some basic architecture for reference in the future. </p> <p>Interesting things: (1) The computational complexity of dendrite modules (W^i,i-1A^i-1 ○ A^i-1) connected in series is far lower than Horner's method. Will this speed up the calculation of basic functions in computers? (2) The range of sight of animals has a gradient, but the convolution layer does not have this characteristic. This paper proposes receptive fields with a gradient. (3) The networks using Gang neurons can delete traditional networks' Fully-connected Layer. In other words, the Fully-connected Layers' parameters are assigned to a single neuron, which reduces the parameters of a network for the same mapping capacity. </p><p>One important thing: ResDD can replace the current all ANNs' Neurons (ResDD modules+One Linear module)! ResDD has controllable precision for better generalization capability! </p><p>Gang neuron code is available at https://github.com/liugang1234567/Gang-neuron.</p>

It may be time to improve the neuron of artificial neural network

<p>Artificial neural networks (ANNs) have won numerous contests in pattern recognition, machine learning, or artificial intelligence in recent years. The neuron of ANNs was designed by the stereotypical knowledge of biological neurons 70 years ago. Artificial Neuron is expressed as f(wx+b) or f(WX). This design does not consider dendrites' information processing capacity. However, some recent studies show that biological dendrites participate in the pre-calculation of input data. Concretely, biological dendrites play a role in extracting the interaction information among inputs (features). Therefore, it may be time to improve the neuron of ANNs. According to our previous studies (DD), this paper adds the dendrites' function to artificial Neuron. The dendrite function can be expressed as W^i,i-1A^i-1 ○ A^{0|1|2|...|i-1} . The generalized new neuron can be expressed as f(W(W^i,i-1A^i-1 ○ A^{0|1|2|...|i-1})).The simplified new neuron be expressed as f(∑(WA ○ X)). After improving the neuron, there are so many networks to try. This paper shows some basic architecture for reference in the future. </p> <p>Interesting things: (1) The computational complexity of dendrite modules (W^i,i-1A^i-1 ○ A^i-1) connected in series is far lower than Horner's method. Will this speed up the calculation of basic functions in computers? (2) The range of sight of animals has a gradient, but the convolution layer does not have this characteristic. This paper proposes receptive fields with a gradient. (3) The networks using Gang neurons can delete traditional networks' Fully-connected Layer. In other words, the Fully-connected Layers' parameters are assigned to a single neuron, which reduces the parameters of a network for the same mapping capacity. </p><p>One important thing: ResDD can replace the current all ANNs' Neurons (ResDD modules+One Linear module)! ResDD has controllable precision for better generalization capability! </p><p>Gang neuron code is available at https://github.com/liugang1234567/Gang-neuron.</p>

It may be time to improve the neuron of artificial neural network

<p>Artificial neural networks (ANNs) have won numerous contests in pattern recognition, machine learning, or artificial intelligence in recent years. The neuron of ANNs was designed by the stereotypical knowledge of biological neurons 70 years ago. Artificial Neuron is expressed as f(wx+b) or f(WX). This design does not consider dendrites' information processing capacity. However, some recent studies show that biological dendrites participate in the pre-calculation of input data. Concretely, biological dendrites play a role in extracting the interaction information among inputs (features). Therefore, it may be time to improve the neuron of ANNs. According to our previous studies (DD), this paper adds the dendrites' function to artificial Neuron. The dendrite function can be expressed as W^i,i-1A^i-1 ○ A^{0|1|2|...|i-1} . The generalized new neuron can be expressed as f(W(W^i,i-1A^i-1 ○ A^{0|1|2|...|i-1})).The simplified new neuron be expressed as f(∑(WA ○ X)). After improving the neuron, there are so many networks to try. This paper shows some basic architecture for reference in the future. </p> <p>Interesting things: (1) The computational complexity of dendrite modules (W^i,i-1A^i-1 ○ A^i-1) connected in series is far lower than Horner's method. Will this speed up the calculation of basic functions in computers? (2) The range of sight of animals has a gradient, but the convolution layer does not have this characteristic. This paper proposes receptive fields with a gradient. (3) The networks using Gang neurons can delete traditional networks' Fully-connected Layer. In other words, the Fully-connected Layers' parameters are assigned to a single neuron, which reduces the parameters of a network for the same mapping capacity. </p><p>One important thing: ResDD can replace the current all ANNs' Neurons (ResDD modules+One Linear module)! ResDD has controllable precision for better generalization capability! </p><p>Gang neuron code is available at https://github.com/liugang1234567/Gang-neuron.</p>

It may be time to improve the neuron of artificial neural network

<p>Artificial neural networks (ANNs) have won numerous contests in pattern recognition, machine learning, or artificial intelligence in recent years. The neuron of ANNs was designed by the stereotypical knowledge of biological neurons 70 years ago. Artificial Neuron is expressed as f(wx+b) or f(WX). This design does not consider dendrites' information processing capacity. However, some recent studies show that biological dendrites participate in the pre-calculation of input data. Concretely, biological dendrites play a role in extracting the interaction information among inputs (features). Therefore, it may be time to improve the neuron of ANNs. According to our previous studies (DD), this paper adds the dendrites' function to artificial Neuron. The dendrite function can be expressed as W^i,i-1A^i-1 ○ A^{0|1|2|...|i-1} . The generalized new neuron can be expressed as f(W(W^i,i-1A^i-1 ○ A^{0|1|2|...|i-1})).The simplified new neuron be expressed as f(∑(WA ○ X)). After improving the neuron, there are so many networks to try. This paper shows some basic architecture for reference in the future. </p> <p>Interesting things: (1) The computational complexity of dendrite modules (W^i,i-1A^i-1 ○ A^i-1) connected in series is far lower than Horner's method. Will this speed up the calculation of basic functions in computers? (2) The range of sight of animals has a gradient, but the convolution layer does not have this characteristic. This paper proposes receptive fields with a gradient. (3) The networks using Gang neurons can delete traditional networks' Fully-connected Layer. In other words, the Fully-connected Layers' parameters are assigned to a single neuron, which reduces the parameters of a network for the same mapping capacity. </p><p>One important thing: ResDD can replace the current all ANNs' Neurons (ResDD modules+One Linear module)! ResDD has controllable precision for better generalization capability! </p><p>Gang neuron code is available at https://github.com/liugang1234567/Gang-neuron.</p>

It may be time to improve the neuron of artificial neural network

<p>Artificial neural networks (ANNs) have won numerous contests in pattern recognition, machine learning, or artificial intelligence in recent years. The neuron of ANNs was designed by the stereotypical knowledge of biological neurons 70 years ago. Artificial Neuron is expressed as f(wx+b) or f(WX). This design does not consider dendrites' information processing capacity. However, some recent studies show that biological dendrites participate in the pre-calculation of input data. Concretely, biological dendrites play a role in extracting the interaction information among inputs (features). Therefore, it may be time to improve the neuron of ANNs. According to our previous studies (DD), this paper adds the dendrites' function to artificial Neuron. The dendrite function can be expressed as W^i,i-1A^i-1 ○ A^{0|1|2|...|i-1} . The generalized new neuron can be expressed as f(W(W^i,i-1A^i-1 ○ A^{0|1|2|...|i-1})).The simplified new neuron be expressed as f(∑(WA ○ X)). After improving the neuron, there are so many networks to try. This paper shows some basic architecture for reference in the future. </p> <p>Interesting things: (1) The computational complexity of dendrite modules (W^i,i-1A^i-1 ○ A^i-1) connected in series is far lower than Horner's method. Will this speed up the calculation of basic functions in computers? (2) The range of sight of animals has a gradient, but the convolution layer does not have this characteristic. This paper proposes receptive fields with a gradient. (3) The networks using Gang neurons can delete traditional networks' Fully-connected Layer. In other words, the Fully-connected Layers' parameters are assigned to a single neuron, which reduces the parameters of a network for the same mapping capacity. </p><p>One important thing: ResDD can replace the current all ANNs' Neurons (ResDD modules+One Linear module)! ResDD has controllable precision for better generalization capability! </p><p>DD code is available at https://github.com/liugang1234567/Gang-neuron.<br></p>

Metabonomic-Transcriptome Integration Analysis on Osteoarthritis and Rheumatoid Arthritis

Purpose. This study is aimed at exploring the potential metabolite/gene biomarkers, as well as the differences between the molecular mechanisms, of osteoarthritis (OA) and rheumatoid arthritis (RA). Methods. Transcriptome dataset GSE100786 was downloaded to explore the differentially expressed genes (DEGs) between OA samples and RA samples. Meanwhile, metabolomic dataset MTBLS564 was downloaded and preprocessed to obtain metabolites. Then, the principal component analysis (PCA) and linear models were used to reveal DEG-metabolite relations. Finally, metabolic pathway enrichment analysis was performed to investigate the differences between the molecular mechanisms of OA and RA. Results. A total of 976 DEGs and 171 metabolites were explored between OA samples and RA samples. The PCA and linear module analysis investigated 186 DEG-metabolite interactions including Glycogenin 1- (GYG1-) asparagine_54, hedgehog acyltransferase- (HHAT-) glucose_70, and TNF receptor-associated factor 3- (TRAF3-) acetoacetate_35. Finally, the KEGG pathway analysis showed that these metabolites were mainly enriched in pathways like gap junction, phagosome, NF-kappa B, and IL-17 pathway. Conclusions. Genes such as HHAT, GYG1, and TRAF3, as well as metabolites including glucose, asparagine, and acetoacetate, might be implicated in the pathogenesis of OA and RA. Metabolites like ethanol and tyrosine might participate differentially in OA and RA progression via the gap junction pathway and phagosome pathway, respectively. TRAF3-acetoacetate interaction may be involved in regulating inflammation in OA and RA by the NF-kappa B and IL-17 pathway.

Setting the Stage: A Comparative Analysis of an Onstage/Offstage and a Linear Clinic Modules

Objective: The purpose of this study was to understand how two different ambulatory design modules—traditional and onstage/offstage—impact operational efficiency, patient throughput, staff collaboration, and patient privacy. Background: Delivery of healthcare is greatly shifting to ambulatory settings because of rapid advancement of medicine and technology, resulting in more day procedures and follow-up care occurring outside of hospitals. It is anticipated that outpatient services will grow roughly 15–23% within the next 10 years (Sg2, 2014). Nonetheless, there is limited research that evaluates how the built environment impacts care delivery and patient outcomes. Method: This is a cross-sectional, comparative study consisted of a mixed-method approach that included shadowing clinic staff and observing and surveying patients. The linear module had shared corridors and publicly exposed workstations, whereas the onstage/offstage module separates patient/visitors from staff with dedicated patient corridors leading to exam rooms (onstage) and enclosed staff work cores (offstage). Roughly 35 hr of clinic staff shadowing and 55 hr of patient observations occurred. A total of 269 questionnaires were completed by patients/visitors. Results: The results demonstrate that the onstage/offstage module significantly improved staff workflow, reduced travel distances, increased communication in private areas, and significantly reduced patient throughput and wait times. However, patients’ perception of privacy did not change among the two modules. Conclusion: Compared to the linear module, this study provides evidence that the onstage/offstage module could have helped to optimize operational efficiencies, staff workflow, and patient throughput.

Experimental Study of the Thermal Insulation Property of Sedum Linear Roof Module in Real Weather Condition in Guangzhou, China

In this paper, four typical building roof modules, i.e., sedum linear module, lightweight planting soil module, water-retaining board module and XPS module, were experimentally investigated by using the guarded hot box under real weather condition in Guangzhou, China. The testing results were compared and analyzed regarding to three main properties of the roof module, i.e., the top and bottom surface temperatures of the modules, air temperature inside the inner box of the guarded hot box and temperature attenuation characteristics for the thermal insulation of the modules. It was found that the sedum linear module performed well in the thermal insulation property under the typical summer weather condition. The analysis results could be used to assist in the application of the green roof module, and contribute to the energy conservation of the buildings.