Visualization for Explanation of Deep Learning-Based Defect Detection Model Using Class Activation Map

Recently,convolutional neural network(CNN)-based visual inspec-tion has been developed to detect defects on building surfaces automatically.The CNN model demonstrates remarkable accuracy in image data analysis;however,the predicted results have uncertainty in providing accurate informa-tion to users because of the“black box”problem in the deep learning model.Therefore,this study proposes a visual explanation method to overcome the uncertainty limitation of CNN-based defect identification.The visual repre-sentative gradient-weights class activation mapping(Grad-CAM)method is adopted to provide visually explainable information.A visualizing evaluation index is proposed to quantitatively analyze visual representations;this index reflects a rough estimate of the concordance rate between the visualized heat map and intended defects.In addition,an ablation study,adopting three-branch combinations with the VGG16,is implemented to identify perfor-mance variations by visualizing predicted results.Experiments reveal that the proposed model,combined with hybrid pooling,batch normalization,and multi-attention modules,achieves the best performance with an accuracy of 97.77%,corresponding to an improvement of 2.49%compared with the baseline model.Consequently,this study demonstrates that reliable results from an automatic defect classification model can be provided to an inspector through the visual representation of the predicted results using CNN models.

Continuous gradient fusion class activation mapping: segmentation of laser-induced damage on large-aperture optics in dark-field images

Segmenting dark-field images of laser-induced damage on large-aperture optics in high-power laser facilities is challenged by complicated damage morphology, uneven illumination and stray light interference. Fully supervised semantic segmentation algorithms have achieved state-of-the-art performance but rely on a large number of pixel-level labels, which are time-consuming and labor-consuming to produce. LayerCAM, an advanced weakly supervised semantic segmentation algorithm, can generate pixel-accurate results using only image-level labels, but its scattered and partially underactivated class activation regions degrade segmentation performance. In this paper, we propose a weakly supervised semantic segmentation method, continuous gradient class activation mapping(CAM) and its nonlinear multiscale fusion(continuous gradient fusion CAM). The method redesigns backpropagating gradients and nonlinearly activates multiscale fused heatmaps to generate more fine-grained class activation maps with an appropriate activation degree for different damage site sizes. Experiments on our dataset show that the proposed method can achieve segmentation performance comparable to that of fully supervised algorithms.

Triptolide (PG-490) induces apoptosis of dendritic cells through sequential p38 MAP kinase phosphorylation and caspase 3 activation

Dendritic cells (DCs) are the most potent antigen-presen ting cells that play crucial roles in the regulation of immune response. Triptol ide, an active component purified from the medicinal plant Tripterygium wilfor dii Hook F., has been demonstrated to act as a potent immunosuppressive drug c apab le of inhibiting T cell activation and proliferation. However, little is known a bout the effects of triptolide on DCs. The present study shows that triptolide d oes not affect phenotypic differentiation and LPS-induced maturation of murine DCs. But triptolide can dramatically reduce cell recovery by inducing apoptosis of DCs at concentration as low as 10 ng/ml, as demonstrated by phosphatidylserin e exposure, mitochondria potential decrease, and nuclear DNA condensation. Tript olide induces activation of p38 in DCs, which precedes the activation of caspase 3. SB203580, a specific kinase inhibitor for p38, can block the activation of caspase 3 and inhibit the resultant apoptosis of DCs. Our results suggest that t he anti-inflammatory and immunosuppressive activities of triptolide may be due, in part, to its apoptosis-inducing effects on DCs.

Comparison of detection results of hypoxic-ischemic encephalopathy at different degrees in infant patients between brain electrical activity mapping, transcranial Doppler sonography and computer tomography examinations

BACKGROUND： It has been proved that brain electrical activity mapping （BEAM） and transcranial Doppler （TCD） detection can reflect the function of brain cell and its diseased degree of infant patients with moderate to severe hypoxic-ischemic encephalopathy （HIE）. OBJECTIVE： To observe the abnormal results of HIE at different degrees detected with BEAM and TCD in infant patients, and compare the detection results at the same time point between BEAM, TCD and computer tomography （CT） examinations. DESIGN ： Contrast observation SETTING： Departments of Neuro-electrophysiology and Pediatrics, Second Affiliated Hospital of Qiqihar Medical College. PARTICIPANTS： Totally 416 infant patients with HIE who received treatment in the Department of Newborn Infants, Second Affiliated Hospital of Qiqihar Medical College during January 2001 and December 2005. The infant patients, 278 male and 138 female, were at embryonic 37 to 42 weeks and weighing 2.0 to 4.1 kg, and they were diagnosed with CT and met the diagnostic criteria of HIE of newborn infants compiled by Department of Neonatology, Pediatric Academy, Chinese Medical Association. According to diagnostic criteria, 130 patients were mild abnormal, 196 moderate abnormal and 90 severe abnormal. The relatives of all the infant patients were informed of the experiment. METHOOS： BEAM and TCD examinations were performed in the involved 416 infant patients with HIE at different degrees with DYD2000 16-channel BEAM instrument and EME-2000 ultrasonograph before preliminary diagnosis treatment （within 1 month after birth） and 1,3,6,12 and 24 months after birth, and detected results were compared between BEAM, TCD and CT examinations. MAIN OUTCOME MEASURES： Comparison of detection results of HIE at different time points in infant patients between BEAM. TCD and CT examinations. RESULTS： All the 416 infant patients with HIE participated in the result analysis. （1） Comparison of the detected results in infant patients with mild HIE at different time points after birth between BEAM, TCD and CT examinations： BEAM examination showed that the recovery was delayed, and the abnormal rate of BEAM examination was significantly higher than that of CT examination 1 and 3 months after birth [55.4%（72/130）vs. 17.0% （22/130 ）,x^2=41.66 ;29.2% （ 38/130 ） vs. 6.2% （ 8/130 ）, x^2=23.77, P 〈 0.01 ], exceptional patients had mild abnormality and reached the normal level in about 6 months. TCD examination showed that the disease condition significantly improved and infant patients with HIE basically recovered 1 or 2 months after birth, while CT examination showed that infant patients recovered 3 or 4 months after birth. （2） Comparison of detection results of infant patients with moderate HIE at different time points between BEAM, TCD and CT examinations： The abnormal rate of BEAM examination was significantly higher than that of CT examination 1,3,6 and 12 months after birth [90.8% （178/196）,78.6% （154/196）,x^2=4.32,P 〈 0.05;64.3% （126/196）,43.9% （86/196） ,x^2=16.44 ;44.9% （88/196） ,22.4% （44/196）,x^2=22.11 ;21.4% （42/196）, 10.2% （20/196）,x^2=9.27, P 〈 0.01]. BEAM examination showed that there was still one patient who did not completely recovered in the 24^th month due to the relatives of infant patients did not combine the treatment,. TCD examination showed that the abnormal rate was 23.1%（30/196）in the 1^st month after birth, and all the patients recovered to the normal in the 3^rd month after birth, while CT examination showed that mild abnormality still existed in the 24^th month after birth （1.0% ,2/196）. （3） Comparison of detection results of infant patients with severe HIE at different time points between BEAM, TCD and CT examinations： The abnormal rate of BEAM examination was significantly higher than that of CT examination in the 1^st, 3^rd, 6^th and 12^th months after birth[86.7% （78/90）,44.4% （40/90）,x^2=35.53;62.2% （56/90）,31.1% （28/90）,x^2=17.51 ;37.8% （34/90）,6.7% （6/90）, x^2=27.14, P 〈 0.01]. BEAM examination showed that mild abnormality still existed in 4 infant patients in the 24^th month after birth. TCD examination showed that the abnormal rate was 11.1% （10/90） in the 3^rd month after birth, and all the infant patients recovered in the 6^th month after birth. CT examination showed that the abnormal rate was 6.7%（6/90） in the 12^th month after birth, and all of infant patients recovered to the normal in the 24^th month after birth.CONCLUSION ： BEAM is the direct index to detect brain function of infant patients with HIE, and positive reaction is still very sensitive in the tracking detection of convalescent period. The positive rate of morphological reaction in CT examination is superior to that in TCD examination, and the positive rate is very high in the acute period of HIE in examination.

Value Analysis among the Parties on Imported Product in Indonesia Using Process Activity Mapping： Freight Forwarder, Distributor, and Retailer

National logistics system in Indonesia can be categorized as inefficient logistics system due to current number of non-value added （NVA） activities. The unreliable National logistics system and the complexity of distribution system are the big obstacles. This study aims to analyze the value and to propose recommendation for further improvement of National distribution system for imported product. This study employed convenience sampling through in-depth interview to analyze the activity in freight forwarder （FF）, distributor and retailer. To demonstrate and analyze the activity process in each party, Process Activity Mapping （PAM） was used as a tool. The study results showed that the delivery speed in Jakarta （8.4 min/km） is slower than that of Surabaya （6.6 min/km）. The government support through creating adequate infrastructure, good bureaucracy system and good collaboration directly affects the activities of FF, distributor and retailer. Improving FF performance in a timely and reliable manner is required to reduce errors that may occur. Moreover, encouraging the practice of cold chain management is also necessary in distributing the product throughout Indonesia. The pull strategy can be chosen by retailer to eliminate storage activity. Meanwhile, the use of information technology （IT） system is essential to encourage better inventory management, database management and sharing information in distributor and retailer stage.

Breast Calcifications and Histopathological Analysis on Tumour Detection by CNN

The most salient argument that needs to be addressed universally is Early Breast Cancer Detection(EBCD),which helps people live longer lives.The Computer-Aided Detection(CADs)/Computer-Aided Diagnosis(CADx)sys-tem is indeed a software automation tool developed to assist the health profes-sions in Breast Cancer Detection and Diagnosis(BCDD)and minimise mortality by the use of medical histopathological image classification in much less time.This paper purposes of examining the accuracy of the Convolutional Neural Network(CNN),which can be used to perceive breast malignancies for initial breast cancer detection to determine which strategy is efficient for the early iden-tification of breast cell malignancies formation of masses and Breast microcalci-fications on the mammogram.When we have insufficient data for a new domain that is desired to be handled by a pre-trained Convolutional Neural Network of Residual Network(ResNet50)for Breast Cancer Detection and Diagnosis,to obtain the Discriminative Localization,Convolutional Neural Network with Class Activation Map(CAM)has also been used to perform breast microcalcifications detection tofind a specific class in the Histopathological image.The test results indicate that this method performed almost 225.15%better at determining the exact location of disease(Discriminative Localization)through breast microcalci-fications images.ResNet50 seems to have the highest level of accuracy for images of Benign Tumour(BT)/Malignant Tumour(MT)cases at 97.11%.ResNet50’s average accuracy for pre-trained Convolutional Neural Network is 94.17%.

PLDMLT:Multi-Task Learning of Diabetic Retinopathy Using the Pixel-Level Labeled Fundus Images

In the field of medical images,pixel-level labels are time-consuming and expensive to acquire,while image-level labels are relatively easier to obtain.Therefore,it makes sense to learn more information(knowledge)from a small number of hard-to-get pixel-level annotated images to apply to different tasks to maximize their usefulness and save time and training costs.In this paper,using Pixel-Level Labeled Images forMulti-Task Learning(PLDMLT),we focus on grading the severity of fundus images for Diabetic Retinopathy(DR).This is because,for the segmentation task,there is a finely labeled mask,while the severity grading task is without classification labels.To this end,we propose a two-stage multi-label learning weakly supervised algorithm,which generates initial classification pseudo labels in the first stage and visualizes heat maps at all levels of severity using Grad-Cam to further provide medical interpretability for the classification task.A multitask model framework with U-net as the baseline is proposed in the second stage.A label update network is designed to alleviate the gradient balance between the classification and segmentation tasks.Extensive experimental results show that our PLDMLTmethod significantly outperforms other stateof-the-art methods in DR segmentation on two public datasets,achieving up to 98.897%segmentation accuracy.In addition,our method achieves comparable competitiveness with single-task fully supervised learning in the DR severity grading task.

Global and Graph Encoded Local Discriminative Region Representation for Scene Recognition

Scene recognition is a fundamental task in computer vision,which generally includes three vital stages,namely feature extraction,feature transformation and classification.Early research mainly focuses on feature extraction,but with the rise of Convolutional Neural Networks(CNNs),more and more feature transformation methods are proposed based on CNN features.In this work,a novel feature transformation algorithm called Graph Encoded Local Discriminative Region Representation(GEDRR)is proposed to find discriminative local representations for scene images and explore the relationship between the discriminative regions.In addition,we propose a method using the multi-head attention module to enhance and fuse convolutional feature maps.Combining the two methods and the global representation,a scene recognition framework called Global and Graph Encoded Local Discriminative Region Representation(G2ELDR2)is proposed.The experimental results on three scene datasets demonstrate the effectiveness of our model,which outperforms many state-of-the-arts.

Deep Stacked Ensemble Learning Model for COVID-19 Classification

COVID-19 is a growing problem worldwide with a high mortality rate.As a result,the World Health Organization(WHO)declared it a pandemic.In order to limit the spread of the disease,a fast and accurate diagnosis is required.A reverse transcript polymerase chain reaction(RT-PCR)test is often used to detect the disease.However,since this test is time-consuming,a chest computed tomography(CT)or plain chest X-ray(CXR)is sometimes indicated.The value of automated diagnosis is that it saves time and money by minimizing human effort.Three significant contributions are made by our research.Its initial purpose is to use the essential finetuning methodology to test the action and efficiency of a variety of vision models,ranging from Inception to Neural Architecture Search(NAS)networks.Second,by plotting class activationmaps(CAMs)for individual networks and assessing classification efficiency with AUC-ROC curves,the behavior of these models is visually analyzed.Finally,stacked ensembles techniques were used to provide greater generalization by combining finetuned models with six ensemble neural networks.Using stacked ensembles,the generalization of the models improved.Furthermore,the ensemble model created by combining all of the finetuned networks obtained a state-of-the-art COVID-19 accuracy detection score of 99.17%.The precision and recall rates were 99.99%and 89.79%,respectively,highlighting the robustness of stacked ensembles.The proposed ensemble approach performed well in the classification of the COVID-19 lesions on CXR according to the experimental results.

Quaternary tectonic control on channel morphology over sedimentary low land:A case study in the Ajay-Damodar interfluve of Eastern India

The style of active tectonic on the deformation and characterization of fluvial landscape has been investigated on three typical skrike-slip fault zones of the Ajay-Damodar Interfluve（ADI） in Eastern India through field mapping,structural analysis and examination of digital topography（ASTER-30 m）,multispectral imageries,and Google Earth images,Channel morphology in Quaternary sediment is more deformed than Cenozoic lateritic tract and igneous rock system by the neotectonic activities,The structural and lithological controls on the river system in ADI region are reflected by distinct drainage patterns,abrupt change in flow direction,offset river channels,straight river lines,ponded river channel,marshy lands,sag ponds,palaeo-channels,alluvial fans,meander cutoffs,multi-terrace river valley,incised compressed meander,convexity of channel bed slope and knick points in longitudinal profile,Seven morphotectonic indices have been used to infer the role of neotectonic on the modification of channel morphology,A tectonic index map for the ADI region has been prepared by the integration of used morphotectonic indices,which is also calibrated by Bouguer gravity anomaly data and field investigation.

Axonal remodeling in the corticospinal tract after stroke： how does rehabilitative training modulate it？

Stroke causes long-term disability, and rehabilitative training is commonly used to improve the consecutive functional recovery. Following brain damage, surviving neurons undergo morphological alterations to reconstruct the remaining neural network. In the motor system, such neural network remodeling is observed as a motor map reorganization. Because of its significant correlation with functional recovery, motor map reorganization has been regarded as a key phenomenon for functional recovery after stroke. Although the mechanism underlying motor map reorganization remains unclear, increasing evidence has shown a critical role for axonal remodeling in the corticospinal tract. In this study, we review previous studies investigating axonal remodeling in the corticospinal tract after stroke and discuss which mechanisms may underlie the stimulatory effect of rehabilitative training. Axonal remodeling in the corticospinal tract can be classified into three types based on the location and the original targets of corticospinal neurons, and it seems that all the surviving corticospinal neurons in both ipsilesional and contralesional hemisphere can participate in axonal remodeling and motor map reorganization. Through axonal remodeling, corticospinal neurons alter their output selectivity from a single to multiple areas to compensate for the lost function. The remodeling of the corticospinal axon is influenced by the extent of tissue destruction and promoted by various therapeutic interventions, including rehabilitative training. Although the precise molecular mechanism underlying rehabilitation-promoted axonal remodeling remains elusive, previous data suggest that rehabilitative training promotes axonal remodeling by upregulating growth-promoting and downregulating growth-inhibiting signals.

Surrounding rock classification from onsite images with deep transfer learning based on EfficientNet

This paper proposes an accurate,efficient and explainable method for the classification of the surrounding rock based on a convolutional neural network(CNN).The state-of-the-art robust CNN model(EfficientNet)is applied to tunnel wall image recognition.Gaussian filtering,data augmentation and other data pre-processing techniques are used to improve the data quality and quantity.Combined with transfer learning,the generality,accuracy and efficiency of the deep learning(DL)model are further improved,and finally we achieve 89.96%accuracy.Compared with other state-of-the-art CNN architectures,such as ResNet and Inception-ResNet-V2(IRV2),the presented deep transfer learning model is more stable,accurate and efficient.To reveal the rock classification mechanism of the proposed model,Gradient-weight Class Activation Map(Grad-CAM)visualizations are integrated into the model to enable its explainability and accountability.The developed deep transfer learning model has been applied to support the tunneling of the Xingyi City Bypass in the high mountain area of Guizhou,China,with great results.

Variation of activation energy determined by a modified Arrhenius approach:Roles of dynamic recrystallization on the hot deformation of Ni-based superalloy

The hot deformation behaviors of Ni18 Cr9 Co9 Fe5 Nb3 Mo superalloy were explored in the formation temperature range free ofγ’phase with various strain rates applied.The hot deformation behaviors are initially modeled with Arrhenius equation which gives an average activation energy of 581.1 kJ mol^(-1).A modified Arrhenius approach,including the updated Zener-Hollomon parameter is proposed to consider the change of activation ene rgy under different deformation conditions which turns out a relatively accurate computation for activation energy of hot deformation,i.e.,the standard variance for modified model calculated in the covered deformation condition is just 35.4%of that for Arrhenius equation.The modified model also proposes a map for activation ene rgy which ranges from 571.5-589.0 kJ mol^(-1)for various deformation conditions.Microstructural features of the representative superalloy specimens were characterized by electron backscattered diffraction(EBSD)techniques in order to clarify the influence of activation energy on the microstructural formation.It is found that the Ni-based superalloy samples with higher activation energy are promoted by the degree of dynamic recrystallization which suggests that the rise in activation energy gives either a better recrystallization rate or finer grains.

Mapping Domain- and Age-Specific Functional Brain Activity for Children’s Cognitive and Affective Development

The human brain undergoes rapid development during childhood,with significant improvement in a wide spectrum of cognitive and affective functions.Mapping domain-and age-specific brain activity patterns has important implications for characterizing the development of children’s cognitive and affective functions.The current mainstay of brain templates is primarily derived from structural magnetic resonance imaging(MRI),and thus is not ideal for mapping children’s cognitive and affective brain development.By integrating task-dependent functional MRI data from a large sample of 250 children(aged 7 to 12)across multiple domains and the latest easy-to-use and transparent preprocessing workflow,we here created a set of age-specific brain functional activity maps across four domains:attention,executive function,emotion,and risky decision-making.Moreover,we developed a toolbox named Developmental Brain Functional Activity maps across multiple domains that enables researchers to visualize and download domain-and age-specific brain activity maps for various needs.This toolbox and maps have been released on the Neuroimaging Informatics Tools and Resources Clearinghouse website(http://www.nitrc.org/projects/dbfa).Our study provides domain-and age-specific brain activity maps for future developmental neuroimaging studies in both healthy and clinical populations.

On brain activity mapping: Insights and lessons from Brain Decoding Project to map memory patterns in the hippocampus

The BRAIN project recently announced by the president Obama is the reflection of unrelenting human quest for cracking the brain code, the patterns of neuronal activity that define who we are and what we are. While the Brain Activity Mapping proposal has rightly emphasized on the need to develop new technologies for measuring every spike from every neuron, it might be helpful to consider both the theoretical and experimental aspects that would accelerate our search for the organizing principles of the brain code. Here we share several insights and lessons from the similar proposal, namely, Brain Decoding Project that we initiated since 2007. We provide a specific example in our initial mapping of real-time memory traces from one part of the memory circuit, namely, the CA1 region of the mouse hippocampus. We show how innovative behavioral tasks and appropriate mathematical analyses of large datasets can play equally, if not more, important roles in uncovering the specific-to-general feature-coding cell assembly mechanism by which episodic memory, semantic knowledge, and imagination are generated and organized. Our own experiences suggest that the bottleneck of the Brain Project is not only at merely developing additional new technologies, but also the lack of efficient avenues to disseminate cutting edge platforms and decoding expertise to neuroscience community. Therefore, we propose that in order to harness unique insights and extensive knowledge from various investigators working in diverse neuroscience subfields, ranging from perception and emotion to memory and social behaviors, the BRAIN project should create a set of International and National Brain Decoding Centers at which cutting-edge recording technologies and expertise on analyzing large datasets analyses can be made readily available to the entire community of neuroscientists who can apply and schedule to perform cutting-edge research.

Expression Analysis Based on Face Regions in Real-world Conditions

Facial emotion recognition is an essential and important aspect of the field of human-machine interaction.Past research on facial emotion recognition focuses on the laboratory environment.However,it faces many challenges in real-world conditions,i.e.,illumination changes,large pose variations and partial or full occlusions.Those challenges lead to different face areas with different degrees of sharpness and completeness.Inspired by this fact,we focus on the authenticity of predictions generated by different<emotion,region>pairs.For example,if only the mouth areas are available and the emotion classifier predicts happiness,then there is a question of how to judge the authenticity of predictions.This problem can be converted into the contribution of different face areas to different emotions.In this paper,we divide the whole face into six areas:nose areas,mouth areas,eyes areas,nose to mouth areas,nose to eyes areas and mouth to eyes areas.To obtain more convincing results,our experiments are conducted on three different databases:facial expression recognition+(FER+),real-world affective faces database(RAF-DB)and expression in-the-wild(ExpW)dataset.Through analysis of the classification accuracy,the confusion matrix and the class activation map(CAM),we can establish convincing results.To sum up,the contributions of this paper lie in two areas:1)We visualize concerned areas of human faces in emotion recognition;2)We analyze the contribution of different face areas to different emotions in real-world conditions through experimental analysis.Our findings can be combined with findings in psychology to promote the understanding of emotional expressions.

A convolutional neural network based approach to sea clutter suppression for small boat detection

Current methods for radar target detection usually work on the basis of high signal-to-clutter ratios.In this paper we propose a novel convolutional neural network based dual-activated clutter suppression algorithm,to solve the problem caused by low signal-to-clutter ratios in actual situations on the sea surface.Dual activation has two steps.First,we multiply the activated weights of the last dense layer with the activated feature maps from the upsample layer.Through this,we can obtain the class activation maps(CAMs),which correspond to the positive region of the sea clutter.Second,we obtain the suppression coefficients by mapping the CAM inversely to the sea clutter spectrum.Then,we obtain the activated range-Doppler maps by multiplying the coefficients with the raw range-Doppler maps.In addition,we propose a sampling-based data augmentation method and an effective multiclass coding method to improve the prediction accuracy.Measurement on real datasets verified the effectiveness of the proposed method.

Biatrial Macroreentry Atrial Tachycardia after Atria Fibrillation Ablation

Catheter ablation is an important therapy for atrial fibrillation （AF） in the last decade. In parallel, atrial tachycardia （AT） has become the most common type of arrhythmia after AF ablation, especially after extensive left atrial （LA） substrate modification,t^j The occurrence of AT after AF is due to the conduction gaps of ablation lines and the conduction obstacle caused by the ablation lesions？-~1 Most of these ATs locate in LA, and here, we described a biatrial macroreentry AT （MAT） after AF ablation.

Deep learning and machine learning neural network approaches for multi class leather texture defect classification and segmentation

Modern leather industries are focused on producing high quality leather products for sustaining the market com-petitiveness. However, various leather defects are introduced during various stages of manufacturing process such as material handling, tanning and dyeing. Manual inspection of leather surfaces is subjective and inconsistent in nature;hence machine vision systems have been widely adopted for the automated inspection of leather defects. It is neces-sary develop suitable image processing algorithms for localize leather defects such as folding marks, growth marks, grain off, loose grain, and pinhole due to the ambiguous texture pattern and tiny nature in the localized regions of the leather. This paper presents deep learning neural network-based approach for automatic localization and classifica-tion of leather defects using a machine vision system. In this work, popular convolutional neural networks are trained using leather images of different leather defects and a class activation mapping technique is followed to locate the region of interest for the class of leather defect. Convolution neural networks such as Google net, Squeeze-net, RestNet are found to provide better accuracy of classification as compared with the state-of-the-art neural network architectures and the results are presented.

Nanoscience and the nano-bioelectronics frontier

This review describes work presented in the 2014 inaugural Tsinghua University Press-Springer Nano Research Award lecture, as well as current and future opportunities for nanoscience research at the interface with brain science. First, we briefly summarize some of the considerations and the research journey that has led to our focus on bottom-up nanoscale science and technology. Second, we recapitulate the motivation for and our seminal contributions to nanowire- based nanoscience and technology, including the rational design and synthesis of increasingly complex nanowire structures, and the corresponding broad range of ＂applications＂ enabled by the capability to control structure, com- position and size from the atomic level upwards. Third, we describe in more detail nanowire-based electronic devices as revolutionary tools for brain science, including （i） motivation for nanoelectronics in brain science, （ii） demonstration of nanowire nanoelectronic arrays for high-spatial/high-temporal resolution extracellular recording, （iii） the development of fundamentally-new intracellular nanoelectronic devices that approach the sizes of single ion channels, （iv） the introduction and demonstration of a new paradigm for innervating cell networks with addressable nanoelectronic arrays in three-dimensions. Last, we conclude with a brief discussion of the exciting and potentially transformative advances expected to come from work at the nanoelectronics-brain interface.