School of Robotics

1. Facile preparation of Co3O4 nanoparticles incorporating with highly conductive MXene nanosheets as high-performance anodes for lithium-ion batteries

Author:Zhao, YC;Liu, CG;Yi, RW;Li, ZQ;Chen, YB;Li, YQ;Mitrovic, I;Taylor, S;Chalker, P;Yang, L;Zhao, CZ


Abstract:There is considerable scientific interest in the newly available family of MXenes material. An analog as graphene, this two-dimensional (2D) layered material with the structure of transition metal carbides or nitrides exhibits superior electronic conductivity, large interlayer spacing for highly efficient lithium ions diffusion pathways and environmental benignity, making it useful as energy storage material. However, the inferior capability to store lithium ions impedes its wide application in lithium-ion batteries. Therefore, a facile strategy for preparing Co3O4 nanoparticles incorporated with MXene nanosheets on Ni foams has been developed. Small-size Co3O4 nanoparticles are uniformly distributed within the MXene nanosheets leading to the highly efficient lithium ions and electrons transmission, as well as the prevention for the restacking of MXene nanosheets and huge volume change of the Co3O4 nanoparticles. Under the cooperative effect of Co3O4 nanoparticles and MXene nanosheets, the Co3O4/MXene composite electrode with the mass ratio of Co3O4/MXene = 1:1 exhibits an excellent reversible capacity of 1005 mAh g(-1) after 300 cycles at the current density of 1 C, which significantly exceeds that of pristine Co3O4 electrode. Though the current density climbs to 5 C, the composite electrode remains a stable capacity of 307 mAh g(-1) after 1000 cycles. It is demonstrated that Co3O4/MXene composite electrode has the potential as an anode for the high-performance lithium-ion batteries. (c) 2020 Elsevier Ltd. All rights reserved.
2. Comparisons of switching characteristics between Ti/Al2O3/Pt and TiN/Al2O3/Pt RRAM devices with various compliance currents

Author:Qi, YF;Zhao, CZ;Liu, CG;Fang, YX;He, JH;Luo, T;Yang, L;Zhao, C


Abstract:In this study, the influence of the Ti and TiN top electrodes on the switching behaviors of the Al2O3/Pt resistive random access memory devices with various compliance currents (CCs, 1-15 mA) has been compared. Based on the similar statistical results of the resistive switching (RS) parameters such as V-set/V-reset, R-HRS/R-LRS (measured at 0.10 V) and resistance ratio with various CCs for both devices, the Ti/Al2O3/Pt device differs from the TiN/Al2O3/Pt device mainly in the forming process rather than in the following switching cycles. Apart from the initial isolated state, the Ti/Al2O3/Pt device has the initial intermediate state as well. In addition, its forming voltage is relatively lower. The conduction mechanisms of the ON and OFF state for both devices are demonstrated as ohmic conduction and Frenkel-Poole emission, respectively. Therefore, with the combined modulations of the CCs and the stop voltages, the TiN/Al2O3/Pt device is more stable for nonvolatile memory applications to further improve the RS performance.
3. Field Support Vector Machines

Author:Huang, KZ;Jiang, HC;Zhang, XY


Abstract:The identically and independently distributed (i.i.d.) condition required by conventional machine learning approaches may sometimes be violated when patterns occur as groups (where each group shares a homogeneous style, called a field). By breaking it, we extend in this paper the famous Support Vector Machine (SVM) to a novel framework named Field Support Vector Machine (F-SVM), in which the training and predicting a group of patterns (i.e., a field pattern) are performed simultaneously. Specifically, the proposed F-SVM is learned by optimizing simultaneously both the classifier and the Style Normalization Transformation (SNT) for each group of data, even feasible in the high-dimensional kernel space. The SNT transform the original style-discriminative patterns to style-free ones, satisfying the i.i.d. assumption required by the conventional SVM learning and implementation. An efficient optimization algorithm is further developed with the convergence guaranteed theoretically. More importantly, by appropriately exploring the style consistency in each field, the proposed F-SVM model is able to significantly improve the classification accuracy. A series of experiments are conducted to verify the effectiveness and confirmed improvement on the performance of the F-SVM model. Empirical results show that the proposed F-SVM outperforms other relevant baselines in two different benchmark data sets.
4. Fast and straightforward analysis approach of charge transport data in single molecule junctions

Author:Zhang, Q;Liu, CG;Tao, SH;Yi, RW;Su, WT;Zhao, CZ;Zhao, C;Dappe, YJ;Nichols, RJ;Yang, L


Abstract:In this study, we introduce an efficient data sorting algorithm, including filters for noisy signals, conductance mapping for analyzing the most dominant conductance group and sub-population groups. The capacity of our data analysis process has also been corroborated on real experimental data sets of Au-1,6-hexanedithiol-Au and Au-1,8-octanedithiol-Au molecular junctions. The fully automated and unsupervised program requires less than one minute on a standard PC to sort the data and generate histograms. The resulting one-dimensional and two-dimensional log histograms give conductance values in good agreement with previous studies. Our algorithm is a straightforward, fast and user-friendly tool for single molecule charge transport data analysis. We also analyze the data in a form of a conductance map which can offer evidence for diversity in molecular conductance. The code for automatic data analysis is openly available, well-documented and ready to use, thereby offering a useful new tool for single molecule electronics.
5. Semi-supervised Pathology Segmentation with Disentangled Representations

Author:Jiang,Haochuan;Chartsias,Agisilaos;Zhang,Xinheng;Papanastasiou,Giorgos;Semple,Scott;Dweck,Mark;Semple,David;Dharmakumar,Rohan;Tsaftaris,Sotirios A.

Source:Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics),2020,Vol.12444 LNCS

Abstract:Automated pathology segmentation remains a valuable diagnostic tool in clinical practice. However, collecting training data is challenging. Semi-supervised approaches by combining labelled and unlabelled data can offer a solution to data scarcity. An approach to semi-supervised learning relies on reconstruction objectives (as self-supervision objectives) that learns in a joint fashion suitable representations for the task. Here, we propose Anatomy-Pathology Disentanglement Network (APD-Net), a pathology segmentation model that attempts to learn jointly for the first time: disentanglement of anatomy, modality, and pathology. The model is trained in a semi-supervised fashion with new reconstruction losses directly aiming to improve pathology segmentation with limited annotations. In addition, a joint optimization strategy is proposed to fully take advantage of the available annotations. We evaluate our methods with two private cardiac infarction segmentation datasets with LGE-MRI scans. APD-Net can perform pathology segmentation with few annotations, maintain performance with different amounts of supervision, and outperform related deep learning methods.
6. Fabrication of a Light-Weight Dual-Function Modified Separator towards High-Performance Lithium-Sulfur Batteries

Author:Yi, RW;Lin, XF;Zhao, YC;Liu, CG;Li, YQ;Hardwick, LJ;Yang, L;Zhao, CZ;Geng, XW;Zhang, Q


Abstract:A light-weight dual-functional modified separator for lithium-sulfur batteries is prepared through a physical blend and blade-coating approach. The separator is coated with carbon black/poly(3,4-ethylene dioxythiophene):poly(styrene sulfonate) (CB/PEDOT:PSS), remarkably improving the utilization of sulfur by serving as the co-current collector. Moreover, the PEDOT:PSS effectively inhibits the diffusion of polysulfides and promotes the migration of lithium ions by providing chemical absorption and cation transport acceleration. When assembling this modified separator into the coin cell, an initial specific capacity of 1315 mAh g(-1) at 0.2 C is achieved with a capacity of 956 mAh g(-1) after 100 cycles, showing a superior performance compared to the cell without an interlayer. Meanwhile, the cell exhibits a rate capability with a discharge capacity of 699 mAh g(-1) at a current density of 2 C. Notably, the areal density of CB/PEDOT:PSS coating is as low as 0.604 mg cm(-2), bringing a specific electrode capacity of 522 mAh g(-1) at 1 C.
7. Maximum Power Point Estimation for Photovoltaic Strings Subjected to Partial Shading Scenarios

Author:Ma, JM;Jiang, HC;Bi, ZQ;Huang, KZ;Li, XS;Wen, HQ


Abstract:Partial shading is an unavoidable complication in the field of photovoltaic (PV) generation. Bypass diodes have become a standard feature of solar cell arrays to improve array performance under partial shading scenarios (PSS). However, the current-voltage and power-voltage characteristic data vary with different shading patterns. In this paper, a shading pattern detection algorithm is first proposed to estimate the number of shaded modules in a PV string. The result is then fed to a field-support vector regression (F-SVR) model, in which the measured features are transferred into a style-free high-dimension space prior to data training. The process of style-normalized transformation enables the features to be independent and identically distributed. Both simulations and experiments are conducted to evaluate the F-SVR model's ability to estimate the voltage at maximum power points. The results show that the proposed maximum power point estimation method can evidently reduce prediction errors.
8. Self-training field pattern prediction based on kernel methods

Author:Jiang,Haochuan;Huang,Kaizhu;Zhang,Xu Yao;Zhang,Rui

Source:Semi-Supervised Learning: Background, Applications and Future Directions,2018,Vol.

Abstract:Conventional predictors often regard input samples as identically and independently distributed (i.i.d.). Such an assumption does not always hold in many real scenarios, especially when patterns occur as groups, where each group shares a homogeneous style. These tasks are named as the field prediction, which can be divided into the field classification and the field regression. Traditional i.i.d.-based machine learning models would always face degraded performance. By breaking the i.i.d. assump- tion, one novel framework called Field SupportVector Machine (F-SVM) with both classification (F-SVC) and regression (F-SVR) purposes is in- troduced in this chapter. To be specific, the proposed F-SVM predictor is investigated by learning simultaneously both the predictor and the Style Normalization Transformation (SNT) for each group of data (called field). Such joint learning is proved to be even feasible in the high-dimensional kernel space. An efficient alternative optimization algorithm is further designed with the final convergence guaranteed theoretically and experimentally. More importantly, a self-training based kernelized algorithm is also developed to incorporate the F-SVM model with the unknown field during the training phase by learning the transductive SNT to transfer the trained field information to this unknown style data. A series of experiments are conducted to verify the effectiveness of the F-SVM model with both classification and regression tasks by promoting the classification accuracy and declining regression error. Empirical results demonstrate that the proposed F-SVM achieves in several benchmark datasets the best performance so far, significantly better than those state-of-the-art predictors.
9. Alloyed Cu/Si core-shell nanoflowers on the three-dimensional graphene foam as an anode for lithium-ion batteries

Author:Liu, CG;Zhao, YC;Yi, RW;Sun, Y;Li, YQ;Yang, L;Mitrovic, I;Taylor, S;Chalker, P;Zhao, CZ


Abstract:In this study, we demonstrate a facile method to fabricate a flexible alloyed copper/silicon core-shell nanoflowers structure anchored on the three-dimensional graphene foam as a current collector. This combination provides flexible and free-standing structure and three-dimensional conductive network, allowing unique properties for current collection and transmission. The copper oxide nanoflowers are synthesized on the three-dimensional graphene foam by a simple electrodeposition and etching, which serves as an outstanding template to retard the stress effects during the lithiation/delithiation of silicon. After the silicon coating uniformly deposited on the copper oxide nanoflowers, a simple hydrogen annealing was applied to reduce copper oxide nanoflowers and form the copper/silicon alloy, remarkably enhancing the conductivity of silicon. Moreover, this structure can be directly assembled without any conductive additive or binder. In electrochemical testing, the resulting copper/silicon core-shell nanoflowered electrode demonstrates a high initial capacity of 1869 mAh g(-1) at 1.6 A g(-1), with a high retention rate of 66.6%% after 500 cycles. More importantly, at a high current density of 10 A g(-1), this anode still remains a high capacity retention >63%% (compared with the highest capacity 679 mAh g(-1)), offering enormous potential for energy storage applications. (c) 2019 Elsevier Ltd. All rights reserved.
10. Field Support Vector Regression

Author:Jiang, HC;Huang, KZ;Zhang, R


Abstract:In regression tasks for static data, existing methods often assume that they were generated from an identical and independent distribution (i.i.d.). However, violation can be found when input samples may form groups, each affected by a certain different domain. In this case, style consistency exists within a same group, leading to degraded performance when conventional machine learning models were applied due to the violation of the i.i.d. assumption. In this paper, we propose one novel regression model named Field Support Vector Regression (F-SVR) without i.i.d. assumption. Specifically, we perform a style normalization transformation learning and the regression model learning simultaneously. An alternative optimization with final convergence guaranteed is designed, as well as a transductive learning algorithm, enabling extension on unseen styles during the testing phase. Experiments are conducted on two synthetic as well as two real benchmark data sets. Results show that the proposed F-SVR significantly outperforms many other state-of-the-art regression models in all the used data sets.
11. A light-weight free-standing graphene foam-based interlayer towards improved Li-S cells

Author:Yi, RW;Liu, CG;Zhao, YC;Hardwick, LJ;Li, YQ;Geng, XW;Zhang, Q;Yang, L;Zhao, CZ


Abstract:A light-weight and free-standing graphene foam interlayer placed between sulfur cathode and separator is investigated to improve the electrochemical performance of lithium-sulfur batteries. The highly conductive and light-weight porous graphene foam not only increases the electron pathway of cathode, but also adsorbs the dissolved high-order lithium polysulfides during cycles, thus the loss of active materials is greatly avoided with only minimum mass addition approximately 0.3 mg cm(-2) on cathodic side. Additionally, the atomic layer deposition method is applied to deposit the zinc oxide nano-scale coating on graphene foam interlayer in order to chemically trap the polysulfides with minimized deterioration on conductivity of graphene foam. Among all the graphene foam, graphene foam@zinc oxide and graphene foam/graphene foam@zinc oxide interlayers, the graphene foam/graphene foam@zinc oxide exhibits the best electrochemical performance, delivering an initial specific capacity of 1051 mAh g(-1) at 0.5 C and retaining a reversible capacity of 672 mAh g(-1) after 100 cycles, while the cell without interlayer only shows 346 mAh g(-1). These results demonstrate the strategy of including a zinc oxide modified graphene foam interlayer as an effective light-weight interlayer for improving Li-S cell performance. (C) 2019 Elsevier Ltd. All rights reserved.
12. Style-Neutralized Pattern Classification Based on Adversarially Trained Upgraded U-Net

Author:Jiang, Haochuan ; Huang, Kaizhu ; Zhang, Rui ; Hussain, Amir

Source:Cognitive Computation,2021,Vol.13

Abstract:Traditional machine learning approaches usually hold the assumption that data for model training and in real applications are created following the identical and independent distribution (i.i.d.). However, several relevant research topics have demonstrated that such condition may not always describe the real scenarios. One particular case is that the patterns are equipped with diverse and changeable style information. In this paper, a novel classification framework named Style Neutralization Generative Adversarial Classifier (SN-GAC), based on an upgraded U-Net architecture, and trained adversarially with the Generative Adversarial Network (GAN) framework, is introduced to accomplish the classification in such disparate and inconsistent data information case. The generative model in SN-GAC neutralizes style information from the original style-discriminative patterns (style-source) by building the mapping function from them to their style-free counterparts (corresponding standard examples, standard-target). A well-learned generator in the SN-GAC framework is capable of producing the targeted style-neutralized data (generated-target), satisfying the i.i.d. condition. Additionally, SN-GAC is trained adversarially, where an independent discriminator is used to surveil and supervise the training progress of the above-mentioned generator by distinguishing between the real and the generated. Simultaneously, an auxiliary classifier is also embedded in the discriminator to assign the correct class label of both the real and generated data. This process proves effective to aid the generator to produce high-quality human-readable style-neutralized patterns. It will then be further fine-tuned for the sake of promoting the final classification performance. Extensive experiments have adequately demonstrated the effectiveness of the proposed SN-GAC framework it outperforms several relevant state-of-the-art baselines on two empirical data sets in the non-i.i.d. data classification task. © 2019, Springer Science+Business Media, LLC, part of Springer Nature.
13. Enhanced electrochemical performance by GeOx-Coated MXene nanosheet anode in lithium-ion batteries

Author:Liu, CG;Zhao, YC;Yi, RW;Wu, H;Yang, WB;Li, YQ;Mitrovic, I;Taylor, S;Chalker, P;Liu, R;Yang, L;Zhao, CZ


Abstract:Here, we demonstrate a facile method to synthesize an amorphous GeOx-coated MXene nanosheet structure as the anode in lithium-ion batteries. By using the GeO32- as the precursor, NaBH4 as the reduction agent, we performed a one-pot in situ synthesis to prepare a composite of GeOx nanoparticles coated on MXene nanosheet. The size of the GeOx nanoparticles is approximately 50 nm, which offers abundant contact surface between active materials with the electrolyte, as well as fast pathways for Li-ion interaction. Moreover, the unique two-dimensional MXene nanosheet serves as the excellent conductive additives to improve the electrochemical stability and electrical conductivity of composite when used in LIBs. The results indicate that the GeOx/MXene nanosheet structure significantly improves the stability during the lithiation/delithiation processes, with the enhanced capacity through an improved kinetic process. Another attractive element of this novel anode is the flexibility to tune the electrochemical properties by using different combination of binder and solvent when the slurry is prepared for the electrode fabrication. The electrode prepared with polyvinylidene fluoride binder and N-methyl pyrrolidinone solvent exhibits an excellent sustainable capacity of 381 mAh gel at 15 A g(-1). By contrast, the electrode with lithium polyacrylate and de-ionized water delivers a reversible capacity of 950 mAh g(-1) at 0.5 A g(-1) after 100 cycles. These interesting results are ascribed to the inner characteristic structure of the two types of electrodes, which have been verified by electrochemical kinetics and scanning electron microscopic images. It also reveals that the different dispersion state is responsible to the difference of electrochemical properties, which highlights the importance of the electrode design for high-performance lithium-ion batteries. (C) 2020 Elsevier Ltd. All rights reserved.
14. Field Support Vector Machines

Author:Huang, KZ;Jiang, HC;Zhang, XY


Abstract:Conventional classifiers often regard input samples as identically and independently distributed (i.i.d.). This is however not true in many real applications, especially when patterns occur as groups (where each group shares a homogeneous style). Such tasks are also called field classification. By breaking the i.i.d. assumption, one novel framework called Field Support Vector Machine (F-SVM) is proposed in this paper. The distinction lies that it is capable of training and predicting a group of patterns (i.e., a field pattern) simultaneously. Specifically, the proposed F-SVM classifier is investigated by learning simultaneously both the classifier and the Style Normalization Transformation for each group of data (called field). Such joint learning proves even feasible in the high-dimensional kernel space. An efficient optimization algorithm is further developed with the convergence guaranteed. More importantly, by appropriately exploring the style consistency in each field, the F-SVM is able to significantly improve the classification accuracy. A series of experiments are conducted to verify the effectiveness of the F-SVM model. Empirical results show that the proposed F-SVM achieves in three different benchmark data sets the hest performance so far, significantly better than those state-of-the-art classifiers.
15. Novel Field-Support Vector Regression-Based Soft Sensor for Accurate Estimation of Solar Irradiance

Author:Ma, JM;Jiang, HC;Huang, KZ;Bi, ZQ;Man, KL


Abstract:An accurate measurement of the solar irradiance is of significance for evaluating and developing of solar renewable energy systems. Soft sensors are used to provide feasible and economical alternatives to costly physical measurement instruments (e.g., pyranometers and pyrheliometers). Conventional soft-sensing methods assume that input data are identically and independently distributed (i.i.d.) and calculate an estimate of the solar irradiance via a regression model. However, different ambient temperatures result in various current-voltage characteristics, meaning that the i.i.d. assumption is violated. To improve the estimation accuracy, a field-support vector regression soft sensor is proposed to estimate the irradiance levels from photovoltaic (PV) electrical characteristics. The soft sensing system groups its input data into several fields in accordance with ambient temperatures. By transforming the original data into a style-free and i.i.d. space, the soft sensing model achieves better estimation performance. The proposed soft sensor can be easily implemented through a PV module, a thermometer, a current sensor, and a DSP development board. It is validated by simulations and experimental prototyping using real outdoor measurements.
16. Charge transport in hybrid platinum/molecule/graphene single molecule junctions

Author:He, CH;Zhang, Q;Gao, TW;Liu, CG;Chen, ZY;Zhao, CZ;Zhao, C;Nichols, RJ;Dappe, YJ;Yang, L


Abstract:The single molecule conductance of hybrid platinum/alkanedithiol/graphene junctions has been investigated with a focus on understanding the influence of employing two very different contact types. We call this an "anti-symmetric" configuration, with the two different contacts here being platinum and graphene, which respectively provide very different electronic coupling to the alkanedithiol bridge. The conductance of these junctions is experimentally investigated by using a non-contact scanning tunneling microscopy (STM) based method called theI(s) technique. These experimental determinations are supported by density functional theory (DFT) calculations. These alkanedithiol bridging molecules conduct electric current through the highest occupied molecular orbital (HOMO), and junctions formed with Pt/graphene electrode pairs are slightly more conductive than those formed with Au/graphene electrodes which we previously investigated. This is consistent with the lower work function of gold than that of platinum. The measured conductance decays exponentially with the length of the molecular bridge with a low tunneling decay constant, which has a similar value for Pt/graphene and Au/graphene electrode pairs, respectively. These new results underline the importance of the coupling asymmetry between the two electrodes, more than the type of the metal electrode itself. Importantly, the tunneling decay constant is much lower than that of alkanedithiols with the symmetrical equivalent,i.e.identical metal electrodes. We attribute this difference to the relatively weak van der Waals coupling at the graphene interface and the strong bond dipole at the Pt-S interface, resulting in a decrease in the potential barrier at the interface.
17. Isothermal sulfur condensation into carbon nanotube/nitrogen-doped graphene composite for high performance lithium-sulfur batteries

Author:Geng, XW;Yi, RW;Yu, ZM;Zhao, CZ;Li, YQ;Wei, QP;Liu, CG;Zhao, YC;Lu, B;Yang, L


Abstract:Nitrogen-doped graphene (NG) is a promising material for fabricating high-performance lithium-sulfur batteries. Here a facile hydrothermal method was used to synthesize the NG and then the composite of NG and SWCNT (NG/SWCNT) was obtained by mixing with single-walled carbon nanotubes (SWCNT) via a simple ultrasonic method. Finally, the NG/SWCNT-sulfur composite (NG/SWCNT-S) is synthesized via an isothermal method that enables rapid vapor infiltration of sulfur into carbon nanotubes. The resulting sulfur-containing cathode shows a good capacity performance, reaching high initial capacities of 1199.6 mAh g(-1) at 0.1 C and 725.2 mAh g(-1) at 1 C. The optimized electrochemical performance can be attributed to the NG addition which leads to an effective improvement of sulfur utilization and seizing polysulfides during cycling. Moreover, we show that the vapor infiltration method based on the thermodynamics of capillary condensation on nanoscale surfaces offers a new idea for assembling cathode, compared to the traditional melt infiltration method.
18. W-Net: One-Shot Arbitrary-Style Chinese Character Generation with Deep Neural Networks

Author:Jiang, HC;Yang, GY;Huang, KZ;Zhang, R


Abstract:Due to the huge category number, the sophisticated combinations of various strokes and radicals, and the free writing or printing styles, generating Chinese characters with diverse styles is always considered as a difficult task. In this paper, an efficient and generalized deep framework, namely, the W-Net, is introduced for the one-shot arbitrary-style Chinese character generation task. Specifically, given a single character (one-shot) with a specific style (e.g., a printed font or hand-writing style), the proposed W-Net model is capable of learning and generating any arbitrary characters sharing the style similar to the given single character. Such appealing property was rarely seen in the literature. We have compared the proposed W-Net framework to many other competitive methods. Experimental results showed the proposed method is significantly superior in the one-shot setting.
19. 3D-structured multi-walled carbon nanotubes/copper nanowires composite as a porous current collector for the enhanced silicon-based anode

Author:Zhao, YC;Liu, CG;Sun, Y;Yi, RW;Cai, YT;Li, YQ;Mitrovic, I;Taylor, S;Chalker, P;Yang, L;Zhao, CZ


Abstract:In this study, multi-walled carbon nanotubes/Cu nanowires-coated on the copper foil used as a three-dimensional porous current collector for Si electrode has been developed to tackle the problems of silicon-based lithium-ion batteries. The highly conductive Cu nanowires cooperated with robust multi-walled carbon nanotubes not only improve the inferior electric conductivity of the Si anode but also strengthen the total frame stability. Furthermore, the three-dimensional structure creates numerous voids on the surface of Cu foils. Such porous structure of the modified current collector offers the flexible volume expansion during lithiation/delithiation process. Meanwhile, the core-shell structure of multi-walled carbon nanotubes/Si and Cu nanowires/Si minimizes the deformation strain and greatly improves the long-term cycling performance in a real battery. As a result, a high specific capacity of 1845 mAh g(-1) in a half cell at a current density of 3.5 A g(-1) after 180 cycles with a capacity retention of 85.1%% has been achieved without any conductive additives or binder. The demonstrated three-dimensional current collector coupled with Si anode might inspire new material development on high-performance of lithium-ion batteries. (C) 2019 Elsevier B.V. All rights reserved.
20. Style Neutralization Generative Adversarial Classifier


Source:Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics),2018,Vol.10989 LNAI

Abstract:Breathtaking improvement has been seen with the recently proposed deep Generative Adversarial Network (GAN). Purposes of most existing GAN-based models majorly concentrate on generating realistic and vivid patterns by a pattern generator with the aid of the binary discriminator. However, few study were related to the promotion of classification performance with merits of those generated ones. In this paper, a novel and generalized classification framework called Style Neutralization Generative Adversarial Classifier (SN-GAC), based on the GAN framework, is introduced to enhance the classification accuracy by neutralizing possible inconsistent style information existing in the original data. In the proposed model, the generator of SN-GAC is trained by mapping the original patterns with certain styles (source) to their style-neutralized or standard counterparts (standard-target), capable of generating the targeted style-neutralized one (generated-target). On the other hand, pairs of both standard (source + standard-target) and generated (source + generated-target) patterns are fed into the discriminator, optimized by not only distinguishing between real and fake, but also classifying the input pairs with correct class label assignment. Empirical experiments fully demonstrate the effectiveness of the proposed SN-GAC framework by achieving so-far the highest accuracy on two benchmark classification databases including the face and the Chinese handwriting character, outperforming several relevant state-of-the-art baseline approaches.
Total 23 results found
Copyright 2006-2020 © Xi'an Jiaotong-Liverpool University 苏ICP备07016150号-1 京公网安备 11010102002019号