Showing papers on "Generalization published in 2022"
TL;DR: In this paper , a comprehensive review of state-of-the-art robust training methods is presented, all of which are categorized into five groups according to their methodological difference, followed by a systematic comparison of six properties used to evaluate their superiority.
Abstract: Deep learning has achieved remarkable success in numerous domains with help from large amounts of big data. However, the quality of data labels is a concern because of the lack of high-quality labels in many real-world scenarios. As noisy labels severely degrade the generalization performance of deep neural networks, learning from noisy labels (robust training) is becoming an important task in modern deep learning applications. In this survey, we first describe the problem of learning with label noise from a supervised learning perspective. Next, we provide a comprehensive review of 62 state-of-the-art robust training methods, all of which are categorized into five groups according to their methodological difference, followed by a systematic comparison of six properties used to evaluate their superiority. Subsequently, we perform an in-depth analysis of noise rate estimation and summarize the typically used evaluation methodology, including public noisy datasets and evaluation metrics. Finally, we present several promising research directions that can serve as a guideline for future studies.
110 citations
TL;DR: This article proposed a robust training objective that is invariant to changes in depth range and scale, advocate the use of principled multi-objective learning to combine data from different sources, and highlight the importance of pretraining encoders on auxiliary tasks.
Abstract: The success of monocular depth estimation relies on large and diverse training sets. Due to the challenges associated with acquiring dense ground-truth depth across different environments at scale, a number of datasets with distinct characteristics and biases have emerged. We develop tools that enable mixing multiple datasets during training, even if their annotations are incompatible. In particular, we propose a robust training objective that is invariant to changes in depth range and scale, advocate the use of principled multi-objective learning to combine data from different sources, and highlight the importance of pretraining encoders on auxiliary tasks. Armed with these tools, we experiment with five diverse training datasets, including a new, massive data source: 3D films. To demonstrate the generalization power of our approach we use zero-shot cross-dataset transfer, i.e. we evaluate on datasets that were not seen during training. The experiments confirm that mixing data from complementary sources greatly improves monocular depth estimation. Our approach clearly outperforms competing methods across diverse datasets, setting a new state of the art for monocular depth estimation.
109 citations
TL;DR: Wang et al. as discussed by the authors proposed a unified dynamic deep spatio-temporal neural network model based on convolutional neural networks and long short-term memory, termed as (DHSTNet) to simultaneously predict crowd flows in every region of a city.
95 citations
TL;DR: Wang et al. as discussed by the authors proposed a unified dynamic deep spatio-temporal neural network model based on convolutional neural networks and long short-term memory, termed as (DHSTNet) to simultaneously predict crowd flows in every region of a city.
94 citations
TL;DR: Wang et al. as discussed by the authors proposed an improved SE-YOLOv5 network model for the recognition of tomato virus diseases, which used a squeeze-and-excitation module to realize the extraction of key features, using a human visual attention mechanism for reference.
82 citations
TL;DR: Wang et al. as discussed by the authors proposed a memory-augmented autoencoder approach for detecting anomalies in IoT data, which aims to use reconstruction errors to determine data anomalies. But, the traditional anomaly detection algorithm has difficulty meeting this demand, not only in complexity but also accuracy.
Abstract: With the development of the Internet of Things, it has been widely studied and deployed in industrial manufacturing, intelligent transportation, and healthcare systems. The time-series feature of the IoT makes the data density and the data dimension higher, where anomaly detection is important to ensure hardware and software security. However, the traditional anomaly detection algorithm has difficulty meeting this demand, not only in complexity but also accuracy. Sometimes the anomaly can be well reconstructed, resulting in a low reconstruction error. In this paper, we propose a memory-augmented autoencoder approach for detecting anomalies in IoT data, which aims to use reconstruction errors to determine data anomalies. First, a memory mechanism is introduced to suppress the generalization ability of the model, and a memory-augmented autoencoder TSMAE is designed for time-series data anomaly detection. Second, by adding penalties and derivable rectifier functions to loss to make the addressing vector sparse, memory modules are encouraged to extract typical normal patterns, thus inhibiting model generalization ability. Finally, through experiments on ECG and Wafer datasets, the validity of TSMAE is verified, and the rationality of hyperparameter setting is discussed through visualizing the memory module addressing vector.
71 citations
TL;DR: In this article , a Prior Guided Feature Enrichment Network (PFENet) is proposed to solve the problem of generalization ability reduction on unseen classes due to inappropriate use of high-level semantic information of training classes and spatial inconsistency between query and support targets.
Abstract: State-of-the-art semantic segmentation methods require sufficient labeled data to achieve good results and hardly work on unseen classes without fine-tuning. Few-shot segmentation is thus proposed to tackle this problem by learning a model that quickly adapts to new classes with a few labeled support samples. Theses frameworks still face the challenge of generalization ability reduction on unseen classes due to inappropriate use of high-level semantic information of training classes and spatial inconsistency between query and support targets. To alleviate these issues, we propose the Prior Guided Feature Enrichment Network (PFENet). It consists of novel designs of (1) a training-free prior mask generation method that not only retains generalization power but also improves model performance and (2) Feature Enrichment Module (FEM) that overcomes spatial inconsistency by adaptively enriching query features with support features and prior masks. Extensive experiments on PASCAL-5 i and COCO prove that the proposed prior generation method and FEM both improve the baseline method significantly. Our PFENet also outperforms state-of-the-art methods by a large margin without efficiency loss. It is surprising that our model even generalizes to cases without labeled support samples.
65 citations
TL;DR: In this article , a survey of state-of-the-art DL frameworks for hyperspectral imaging classification (HSIC) is presented. And the authors discuss some strategies to improve the generalization performance of DL strategies and provide some future guidelines.
Abstract: Hyperspectral imaging (HSI) has been extensively utilized in many real-life applications because it benefits from the detailed spectral information contained in each pixel. Notably, the complex characteristics, i.e., the nonlinear relation among the captured spectral information and the corresponding object of HSI data, make accurate classification challenging for traditional methods. In the last few years, deep learning (DL) has been substantiated as a powerful feature extractor that effectively addresses the nonlinear problems that appeared in a number of computer vision tasks. This prompts the deployment of DL for HSI classification (HSIC) which revealed good performance. This survey enlists a systematic overview of DL for HSIC and compared state-of-the-art strategies of the said topic. Primarily, we will encapsulate the main challenges of TML for HSIC and then we will acquaint the superiority of DL to address these problems. This article breaks down the state-of-the-art DL frameworks into spectral-features, spatial-features, and together spatial–spectral features to systematically analyze the achievements (future research directions as well) of these frameworks for HSIC. Moreover, we will consider the fact that DL requires a large number of labeled training examples whereas acquiring such a number for HSIC is challenging in terms of time and cost. Therefore, this survey discusses some strategies to improve the generalization performance of DL strategies which can provide some future guidelines.
63 citations
TL;DR: In this article , an artificial neural network (ANN) was trained using a decadal-long time record of hydrographic and turbulence observations in the tropical Pacific to parameterize oceanic vertical mixing processes.
Abstract: Abstract Uncertainties in ocean-mixing parameterizations are primary sources for ocean and climate modeling biases. Due to lack of process understanding, traditional physics-driven parameterizations perform unsatisfactorily in the tropics. Recent advances in the deep-learning method and the new availability of long-term turbulence measurements provide an opportunity to explore data-driven approaches to parameterizing oceanic vertical-mixing processes. Here, we describe a novel parameterization based on an artificial neural network trained using a decadal-long time record of hydrographic and turbulence observations in the tropical Pacific. This data-driven parameterization achieves higher accuracy than current parameterizations, demonstrating good generalization ability under physical constraints. When integrated into an ocean model, our parameterization facilitates improved simulations in both ocean-only and coupled modeling. As a novel application of machine learning to the geophysical fluid, these results show the feasibility of using limited observations and well-understood physical constraints to construct a physics-informed deep-learning parameterization for improved climate simulations.
60 citations
TL;DR: In this article , a dual-stage attention-based LSTM network is proposed for short-term zonal load probabilistic forecasting, where a feature attention based encoder is built to calculate the correlation of input features with electricity load at each time step, and a temporal attention based decoder is developed to mine the time dependencies.
60 citations
TL;DR: Domain generalization (DG) aims to achieve OOD generalization by using only source data for model learning as mentioned in this paper , which is a capability natural to humans yet challenging for machines to reproduce.
Abstract: Generalization to out-of-distribution (OOD) data is a capability natural to humans yet challenging for machines to reproduce. This is because most learning algorithms strongly rely on the i.i.d.~assumption on source/target data, which is often violated in practice due to domain shift. Domain generalization (DG) aims to achieve OOD generalization by using only source data for model learning. Over the last ten years, research in DG has made great progress, leading to a broad spectrum of methodologies, e.g., those based on domain alignment, meta-learning, data augmentation, or ensemble learning, to name a few; DG has also been studied in various application areas including computer vision, speech recognition, natural language processing, medical imaging, and reinforcement learning. In this paper, for the first time a comprehensive literature review in DG is provided to summarize the developments over the past decade. Specifically, we first cover the background by formally defining DG and relating it to other relevant fields like domain adaptation and transfer learning. Then, we conduct a thorough review into existing methods and theories. Finally, we conclude this survey with insights and discussions on future research directions.
TL;DR: In this article , a generalization of the regularized Ψ-Hilfer fractional derivative, called as regularized ǫ-hilfer derivative, is presented and the existence and uniqueness of its solution is analyzed.
TL;DR: A lightweight convolutional neural network was designed by incorporating different attention modules to improve the performance of the models, and nominally increased the network complexity and parameters compared to the model without attention modules, thereby producing better detection accuracy.
Abstract: Plant diseases pose a significant challenge for food production and safety. Therefore, it is indispensable to correctly identify plant diseases for timely intervention to protect crops from massive losses. The application of computer vision technology in phytopathology has increased exponentially due to automatic and accurate disease detection capability. However, a deep convolutional neural network (CNN) requires high computational resources, limiting its portability. In this study, a lightweight convolutional neural network was designed by incorporating different attention modules to improve the performance of the models. The models were trained, validated, and tested using tomato leaf disease datasets split into an 8:1:1 ratio. The efficacy of the various attention modules in plant disease classification was compared in terms of the performance and computational complexity of the models. The performance of the models was evaluated using the standard classification accuracy metrics (precision, recall, and F1 score). The results showed that CNN with attention mechanism improved the interclass precision and recall, thus increasing the overall accuracy (>1.1%). Moreover, the lightweight model significantly reduced network parameters (~16 times) and complexity (~23 times) compared to the standard ResNet50 model. However, amongst the proposed lightweight models, the model with attention mechanism nominally increased the network complexity and parameters compared to the model without attention modules, thereby producing better detection accuracy. Although all the attention modules enhanced the performance of CNN, the convolutional block attention module (CBAM) was the best (average accuracy 99.69%), followed by the self-attention (SA) mechanism (average accuracy 99.34%).
TL;DR: Zhang et al. as discussed by the authors designed a network architecture called MHF-net, which not only contains clear interpretability, but also reasonably embeds the well studied linear mapping that links the HrHS image to HrMS and LrHS images.
Abstract: Multispectral and hyperspectral image fusion (MS/HS fusion) aims to fuse a high-resolution multispectral (HrMS) and a low-resolution hyperspectral (LrHS) images to generate a high-resolution hyperspectral (HrHS) image, which has become one of the most commonly addressed problems for hyperspectral image processing. In this paper, we specifically designed a network architecture for the MS/HS fusion task, called MHF-net, which not only contains clear interpretability, but also reasonably embeds the well studied linear mapping that links the HrHS image to HrMS and LrHS images. In particular, we first construct an MS/HS fusion model which merges the generalization models of low-resolution images and the low-rankness prior knowledge of HrHS image into a concise formulation, and then we build the proposed network by unfolding the proximal gradient algorithm for solving the proposed model. As a result of the careful design for the model and algorithm, all the fundamental modules in MHF-net have clear physical meanings and are thus easily interpretable. This not only greatly facilitates an easy intuitive observation and analysis on what happens inside the network, but also leads to its good generalization capability. Based on the architecture of MHF-net, we further design two deep learning regimes for two general cases in practice: consistent MHF-net and blind MHF-net. The former is suitable in the case that spectral and spatial responses of training and testing data are consistent, just as considered in most of the pervious general supervised MS/HS fusion researches. The latter ensures a good generalization in mismatch cases of spectral and spatial responses in training and testing data, and even across different sensors, which is generally considered to be a challenging issue for general supervised MS/HS fusion methods. Experimental results on simulated and real data substantiate the superiority of our method both visually and quantitatively as compared with state-of-the-art methods along this line of research.
TL;DR: Wang et al. as discussed by the authors proposed a subdomain adaptation transfer learning network (SATLN) to reduce the marginal and conditional distribution bias in cross-domain fault diagnosis. But, the performance of SATLN is limited due to the data distribution discrepancy.
Abstract: Due to the data distribution discrepancy, fault diagnosis models, trained with labeled data in one scene, likely fails in classifying by unlabeled data acquired from the other scenes. Transfer learning is capable to generalize successful application trained in one scene to the fault diagnosis in the other scenes. However, the existing transfer methods do not pay much attention to reduce adaptively marginal and conditional distribution biases, and also ignore the degree of contribution between both biases and among network layers, which limit classification performance and generalization in reality. To overcome these weaknesses, we establish a new fault diagnosis model, called subdomain adaptation transfer learning network (SATLN). First, two convolutional building blocks were stacked to extract transferable features from raw data. Then, the pseudo label learning is amended to construct target subdomain of each class. Furthermore, a subdomain adaptation is combined with domain adaptation to reduce both marginal and conditional distribution biases simultaneously. Finally, a dynamic weight term is applied for adaptive adjustment of the contributions from both discrepancies and each network layers. The SATLN method is tested with six transfer tasks. The results demonstrate the effectiveness and superiority of the SATLN in the cross-domain fault diagnosis field.
TL;DR: In this paper , a generalization of the regularized Ψ-Hilfer fractional derivative, called as regularized ǫ-hilfer derivative, is presented and the existence and uniqueness of its solution is analyzed.
TL;DR: OpenFold as discussed by the authors is a fast, memory-efficient, and trainable implementation of AlphaFold2, and OpenProtein-Set, the largest public database of protein multiple sequence alignments.
Abstract: AlphaFold2 revolutionized structural biology with the ability to predict protein structures with exceptionally high accuracy. Its implementation, however, lacks the code and data required to train new models. These are necessary to (i) tackle new tasks, like protein-ligand complex structure prediction, (ii) investigate the process by which the model learns, which remains poorly understood, and (iii) assess the model’s generalization capacity to unseen regions of fold space. Here we report OpenFold, a fast, memory-efficient, and trainable implementation of AlphaFold2, and OpenProtein-Set, the largest public database of protein multiple sequence alignments. We use OpenProteinSet to train OpenFold from scratch, fully matching the accuracy of AlphaFold2. Having established parity, we assess OpenFold’s capacity to generalize across fold space by retraining it using carefully designed datasets. We find that OpenFold is remarkably robust at generalizing despite extreme reductions in training set size and diversity, including near-complete elisions of classes of secondary structure elements. By analyzing intermediate structures produced by OpenFold during training, we also gain surprising insights into the manner in which the model learns to fold proteins, discovering that spatial dimensions are learned sequentially. Taken together, our studies demonstrate the power and utility of OpenFold, which we believe will prove to be a crucial new resource for the protein modeling community.
TL;DR: In this paper, a systematic literature review of the state-of-the-art on emotion expression recognition from facial images is presented, where the most commonly used strategies employed to interpret and recognize facial emotion expressions, published over the past few years.
TL;DR: Domain generalization (DG) deals with a challenging setting where one or several different but related domain(s) are given, and the goal is to learn a model that can generalize to an unseen test domain this article .
Abstract: Machine learning systems generally assume that the training and testing distributions are the same. To this end, a key requirement is to develop models that can generalize to unseen distributions. Domain generalization (DG), i.e., out-of-distribution generalization, has attracted increasing interests in recent years. Domain generalization deals with a challenging setting where one or several different but related domain(s) are given, and the goal is to learn a model that can generalize to an unseen test domain. Great progress has been made in the area of domain generalization for years. This paper presents the first review of recent advances in this area. First, we provide a formal definition of domain generalization and discuss several related fields. We then thoroughly review the theories related to domain generalization and carefully analyze the theory behind generalization. We categorize recent algorithms into three classes: data manipulation, representation learning, and learning strategy, and present several popular algorithms in detail for each category. Third, we introduce the commonly used datasets, applications, and our open-sourced codebase for fair evaluation. Finally, we summarize existing literature and present some potential research topics for the future.
TL;DR: A systematic literature review of the state-of-the-art on emotion expression recognition from facial images is presented in this article , where the most commonly used strategies employed to interpret and recognize facial emotion expressions, published over the past few years.
TL;DR: In this paper , a multi-source subdomain adaptation transfer learning method is proposed to transfer diagnostic knowledge from multiple sources for cross-domain fault diagnosis, where the local maximum mean discrepancy is used for fine-grained local alignment of subdomain distributions within the same category of different domains.
Abstract: In modern industrial equipment maintenance, transfer learning is a promising tool that has been widely utilized to solve the problem of the insufficient generalization ability of diagnostic models, caused by changes in working conditions. However, owing to the single knowledge transfer source and fuzzy marginal distribution matching, the ability of traditional transfer learning methods for cross-domain fault diagnosis is not ideal. In practice, collecting multi-source data from different scenarios can provide richer generalization knowledge, and fine-grained information matching of relevant subdomains can achieve more accurate knowledge transfer, which is conducive to the improvement of the cross-domain fault diagnosis performance. To this end, a multi-source subdomain adaptation transfer learning method is proposed to transfer diagnostic knowledge from multiple sources for cross-domain fault diagnosis. This approach exploits a multi-branch network structure to match the feature spatial distributions of each source and target domain separately, where the local maximum mean discrepancy is used for fine-grained local alignment of subdomain distributions within the same category of different domains. Moreover, the weighted score of a source-specific is obtained according to its distribution distance, and multiple source classifiers are combined with the corresponding weighted scores for the joint diagnosis of the device status. Extensive experiments are conducted on three rotating machinery datasets to verify the effectiveness of the proposed model for cross-domain fault diagnosis.
TL;DR: In this paper , an additive mapping approach was developed to rapidly expand the utility of synthetic methods while generating concurrent mechanistic insight for nickel-catalyzed cross-couplings.
Abstract: Reaction generality is crucial in determining the overall impact and usefulness of synthetic methods. Typical generalization protocols require a priori mechanistic understanding and suffer when applied to complex, less understood systems. We developed an additive mapping approach that rapidly expands the utility of synthetic methods while generating concurrent mechanistic insight. Validation of this approach on the metallaphotoredox decarboxylative arylation resulted in the discovery of a phthalimide ligand additive that overcomes many lingering limitations of this reaction and has important mechanistic implications for nickel-catalyzed cross-couplings. Description Additive improvements to nickel catalysis It often takes decades of incremental optimization to apply chemical reactions beyond the small range of substrates studied at the discovery stage. Prieto Kullmer et al. sought to accelerate that optimization process by screening a large, diverse group of additives to a cooperative nickel-photoredox catalyst system. The screen revealed that phthalimides substantially expand the functional compatibility of the nickel catalyst and thus the substrate scope. The phthalimide appears to stabilize oxidative addition complexes as well as break up deactivated catalyst aggregates. —JSY An additive screen reveals that phthalimide coordination substantially stabilizes a nickel catalyst and expands its scope.
TL;DR: The MICCAI MIDOG 2021 challenge as mentioned in this paper was the first attempt to derive scanner-agnostic mitosis detection algorithms, which used a training set of 200 cases, split across four scanning systems.
TL;DR: In this paper , a deep learning network was trained to choose essential information from speech spectrograms for the classification layer, performing gender detection, achieving an accuracy of 98.57% better than the traditional ML approaches.
Abstract: Several speaker recognition algorithms failed to get the best results because of the wildly varying datasets and feature sets for classification. Gender information helps reduce this effort since categorizing the classes based on gender may help lessen the impact of gender variability on the retrieved features. This study attempted to construct a perfect classification model for language-independent gender identification utilizing the Common Voice dataset (Mozilla). Most previous studies are doing manual extracting characteristics and feeding them into a machine learning model for categorization. Deep neural networks (DNN) were the most effective strategy in our research. Nonetheless, the main goal was to take advantage of the wealth of information included in voice data without requiring significant manual intervention. We trained the deep learning network to choose essential information from speech spectrograms for the classification layer, performing gender detection. The pretrained ResNet 50 fine-tuned gender data successfully achieved an accuracy of 98.57% better than the traditional ML approaches and the previous works reported with the same dataset. Furthermore, the model performs well on additional datasets, demonstrating the approach’s generalization capacity.
TL;DR: Exhaustive simulation results on mammograms dataset, namely, MIAS, DDSM, and INbreast, as well as ultrasound datasets, depict that the suggested model outperforms the recent state-of-the-art schemes.
TL;DR: In this paper , the authors proposed an information fractal dimension of mass function for complexity analysis of time series, which was shown to be effective in complexity analysis in time series analysis.
Abstract: Fractals play an important role in nonlinear science. The most important parameter when modeling a fractal is the fractal dimension. Existing information dimension can calculate the dimension of probability distribution. However, calculating the fractal dimension given a mass function, which is the generalization of probability, is still an open problem of immense interest. The main contribution of this work is to propose an information fractal dimension of mass function. Numerical examples are given to show the effectiveness of our proposed dimension. We discover an important property in that the dimension of mass function with the maximum Deng entropy is [Formula: see text], which is the well-known fractal dimension of Sierpiski triangle. The application in complexity analysis of time series illustrates the effectiveness of our method.
20 Jun 2022
TL;DR: The authors analyzed 100 highly cited machine learning papers published at premier machine learning conferences, ICML and NeurIPS, and found that the papers most frequently justify and assess themselves based on Performance, Generalization, Quantitative evidence, Efficiency, Building on past work, and Novelty.
Abstract: Machine learning currently exerts an outsized influence on the world, increasingly affecting institutional practices and impacted communities. It is therefore critical that we question vague conceptions of the field as value-neutral or universally beneficial, and investigate what specific values the field is advancing. In this paper, we first introduce a method and annotation scheme for studying the values encoded in documents such as research papers. Applying the scheme, we analyze 100 highly cited machine learning papers published at premier machine learning conferences, ICML and NeurIPS. We annotate key features of papers which reveal their values: their justification for their choice of project, which attributes of their project they uplift, their consideration of potential negative consequences, and their institutional affiliations and funding sources. We find that few of the papers justify how their project connects to a societal need (15%) and far fewer discuss negative potential (1%). Through line-by-line content analysis, we identify 59 values that are uplifted in ML research, and, of these, we find that the papers most frequently justify and assess themselves based on Performance, Generalization, Quantitative evidence, Efficiency, Building on past work, and Novelty. We present extensive textual evidence and identify key themes in the definitions and operationalization of these values. Notably, we find systematic textual evidence that these top values are being defined and applied with assumptions and implications generally supporting the centralization of power. Finally, we find increasingly close ties between these highly cited papers and tech companies and elite universities.
TL;DR: In this article , a new intelligent fault diagnosis framework based on dynamic model and transfer learning for rolling element bearings race faults is proposed, which can learn more transferable features and reduce the feature distribution discrepancy, contributing to enhancing the fault identification performance significantly.
Abstract: Intelligent fault diagnosis of rolling element bearings gains increasing attention in recent years due to the promising development of artificial intelligent technology. Many intelligent diagnosis methods work well requiring massive historical data of the diagnosed object. However, it is hard to get sufficient fault data in advance in real diagnosis scenario and the diagnosis model constructed on such small dataset suffers from serious overfitting and losing the ability of generalization, which is described as small sample problem in this paper. Focus on the small sample problem, this paper proposes a new intelligent fault diagnosis framework based on dynamic model and transfer learning for rolling element bearings race faults. In the proposed framework, dynamic model of bearing is utilized to generate massive and various simulation data, then the diagnosis knowledge learned from simulation data is leveraged to real scenario based on convolutional neural network (CNN) and parameter transfer strategies. The effectiveness of the proposed method is verified and discussed based on three fault diagnosis cases in detail. The results show that based on the simulation data and parameter transfer strategies in CNN, the proposed method can learn more transferable features and reduce the feature distribution discrepancy, contributing to enhancing the fault identification performance significantly.
TL;DR: In this paper , the authors identify time-optimal laser pulses to implement the controlled-Z gate and its three qubit generalization, the C2Z gate, for Rydberg atoms in the blockade regime.
Abstract: We identify time-optimal laser pulses to implement the controlled-Z gate and its three qubit generalization, the C2Z gate, for Rydberg atoms in the blockade regime. Pulses are optimized using a combination of numerical and semi-analytical quantum optimal control techniques that result in smooth Ansätze with just a few variational parameters. For the CZ gate, the time-optimal implementation corresponds to a global laser pulse that does not require single site addressability of the atoms, simplifying experimental implementation of the gate. We employ quantum optimal control techniques to mitigate errors arising due to the finite lifetime of Rydberg states and finite blockade strengths, while several other types of errors affecting the gates are directly mitigated by the short gate duration. For the considered error sources, we achieve theoretical gate fidelities compatible with error correction using reasonable experimental parameters for CZ and C2Z gates.
TL;DR: LeafGAN as discussed by the authors is a novel image-to-image translation system with own attention mechanism, which generates a wide variety of diseased images via transformation from healthy images, as a data augmentation tool for improving the performance of plant disease diagnosis.
Abstract: Many applications for the automated diagnosis of plant disease have been developed based on the success of deep learning techniques. However, these applications often suffer from overfitting, and the diagnostic performance is drastically decreased when used on test data sets from new environments. In this article, we propose LeafGAN, a novel image-to-image translation system with own attention mechanism. LeafGAN generates a wide variety of diseased images via transformation from healthy images, as a data augmentation tool for improving the performance of plant disease diagnosis. Due to its own attention mechanism, our model can transform only relevant areas from images with a variety of backgrounds, thus enriching the versatility of the training images. Experiments with five-class cucumber disease classification show that data augmentation with vanilla CycleGAN cannot help to improve the generalization, i.e., disease diagnostic performance increased by only 0.7% from the baseline. In contrast, LeafGAN boosted the diagnostic performance by 7.4%. We also visually confirmed that the generated images by our LeafGAN were much better quality and more convincing than those generated by vanilla CycleGAN. The code is available publicly at https://github.com/IyatomiLab/LeafGAN . Note to Practitioners Automated plant disease diagnosis systems play an important role in the agricultural automation field. Building a practical image-based automatic plant diagnosis system requires collecting a wide variety of disease images with reliable label information. However, it is quite labor-intensive. Conventional systems have reported relatively high diagnosis performance, but most of their scores were largely biased due to the “latent similarity” between training and test images, and their true diagnosis capabilities were much lower than claimed. To address this issue, we propose LeafGAN, which generates countless diverse and high-quality training images; it works as an efficient data augmentation for the diagnosis classifier. Such generated images can be used as useful resources for improving the performance of the cucumber disease diagnosis systems.