Search

In this paper, the authors have compared all of the currently available Apple MacBook Pro laptops, in terms of their usability for basic machine learning research applications (text-based, vision-based, tabular). The paper presents four tests/benchmarks, comparing four Apple Macbook Pro laptop versions: Intel based (i5) and three Apple based (M1, M1 Pro and M1 Max). A script in the Swift programming language was prepared, whose goal was to conduct the training and evaluation process for four machine learning (ML) models. It used the Create ML framework—Apple's solution dedicated to ML model creation on macOS devices. The training and evaluation processes were performed three times. While running, the script performed measurements of their performance, including the time results. The results were compared with each other in tables, which allowed to compare and discuss the performance of individual devices and the benefits of the specificity of their hardware architectures. [ABSTRACT FROM AUTHOR]

People rely on task management applications and digital assistants to capture and track their tasks, and help with executing them. The burden of organizing and scheduling time for tasks continues to reside with users of these systems, despite the high cognitive load associated with these activities. Users stand to benefit greatly from a task management system capable of prioritizing their pending tasks, thus saving them time and effort. In this article, we make three main contributions. First, we propose the problem of task prioritization, formulating it as a ranking over a user’s pending tasks given a history of previous interactions with a task management system. Second, we perform an extensive analysis on the large-scale anonymized, de-identified logs of a popular task management application, deriving a dataset of grounded, real-world tasks from which to learn and evaluate our proposed system. We also identify patterns in how people record tasks as complete, which vary consistently with the nature of the task. Third, we propose a novel contextual deep learning solution capable of performing personalized task prioritization. In a battery of tests, we show that this approach outperforms several operational baselines and other sequential ranking models from previous work. Our findings have implications for understanding the ways people prioritize and manage tasks with digital tools, and in the design of support for users of task management applications. [ABSTRACT FROM AUTHOR]

Homography estimation is a basic and important step in many computer vision tasks. Traditional homography estimation methods are based on feature point matching, which are difficult to work in weak texture images. Deep learning has been applied to homography estimation to improve its robustness, but the existing methods do not consider the multi-scale problem caused by object scale differences, resulting in limited accuracy. To solve the above problems, this paper proposes a multi-scale residual network for homography estimation. The network can extract the multi-scale feature of the image, and used the Multi-Scale Feature Fusion Module to effectively fuse the features. In addition, it further reduced the difficulty of network optimization by estimating the four-corner normalized offset. Experiments on MS-COCO dataset showed that the average corner error of this method was only 0.788 pixels, which achieved sub-pixel accuracy, and can maintain high accuracy in 99% of cases. Due to the comprehensive utilization of multi-scale features and easier to optimize, this method had significantly improved accuracy and stronger robustness. [ABSTRACT FROM AUTHOR]

Supervised learning with the restriction of a few existing training samples is called Few-Shot Learning. FSL is a subarea that puts deep learning performance in a gap, as building robust deep networks requires big training data. Using transfer learning in FSL tasks is an acceptable way to avoid the challenge of building new deep models from scratch. Transfer learning methodology considers borrowing the architecture and parameters of a previously trained model on a large-scale dataset and fine-tuning it for low-data target tasks. But practically, fine-tuning pretrained models in target FSL tasks suffers from overfitting. The few existing samples are not enough to correctly adjust the pretrained model's parameters to provide the best fit for the target task. In this study, we consider mitigating the overfitting problem when applying transfer learning in few-shot Handwritten Character Recognition (HCR) tasks. A data augmentation approach based on Conditional Generative Adversarial Networks is introduced. CGAN is a generative model that can create artificial instances that appear more real and indistinguishable from the original samples. CGAN helps generate extra samples that hold the possible variations of human handwriting instead of applying traditional image transformations. These transformations are low-level, data-independent operations, and only produce augmented samples with limited diversity. The introduced approach was evaluated in fine-tuning the three pretrained models: AlexNet, VGG-16, and GoogleNet. The results show that the samples generated by CGAN can enhance transfer learning performance in few-shot HCR tasks. This is by achieving model fine-tuning with fewer epochs and by increasing the model's F 1 - s c o r e and decreasing the Generalization Error (E test) . [ABSTRACT FROM AUTHOR]

With the increasing complexity of UAV application scenarios, the performance of a single UAV cannot meet the mission requirements. Many complex tasks need the cooperation of multiple UAVs. How to coordinate UAV resources becomes the key to mission completion. In this paper, a task model including multiple UAVs and unknown obstacles is constructed, and the model is transformed into a Markov decision process (MDP). In addition, considering the influence of strategies among UAVs, a multiagent reinforcement learning algorithm based on SAC algorithm and centralized training and decentralized execution framework, MA-SAC (Multi-Agent Soft Actor-Critic), is proposed to solve the MDP. Simulation results show that the algorithm can effectively deal with the task allocation problem of multiple UAVs in this scenario, and its performance is better than other multiagent reinforcement learning algorithms. [ABSTRACT FROM AUTHOR]

With the continuous development of social networks, Weibo has become an essential platform for people to share their opinions and feelings in daily life. Analysis of users' emotional tendencies can be effectively applied to public opinion control, public opinion surveys, and product recommendations. However, the traditional deep learning algorithm often needs a large amount of data to be retained to obtain a better accuracy when faced with new work tasks. Given this situation, a multiclassification method of microblog negative sentiment based on MAML (model-agnostic metalearning) and BiLSTM (bidirectional extended short-term memory network) is proposed to represent the microblog text word vectorization and the combination of MAML and BiLSTM is constructed. The model of BiLSTM realizes the classification of negative emotions on Weibo and updates the parameters through machine gradient descent; the metalearner in MAML calculates the sum of the losses of multiple pieces of training, performs a second gradient descent, and updates the metalearner parameters. The updated metalearner can quickly iterate when faced with a new Weibo negative sentiment classification task. The experimental results show that compared with the prepopular model, on the Weibo negative sentiment dataset, the precision rate, recall rate, and F1 value are increased by 1.68 percentage points, 2.86 percentage points, and 2.27 percentage points, respectively. [ABSTRACT FROM AUTHOR]

A computer-aided diagnosis (CAD) system requires automated stages of tumor detection, segmentation, and classification that are integrated sequentially into one framework to assist the radiologists with a final diagnosis decision. In this paper, we introduce the final step of breast mass classification and diagnosis using a stacked ensemble of residual neural network (ResNet) models (i.e. ResNet50V2, ResNet101V2, and ResNet152V2). The work presents the task of classifying the detected and segmented breast masses into malignant or benign, and diagnosing the Breast Imaging Reporting and Data System (BI-RADS) assessment category with a score from 2 to 6 and the shape as oval, round, lobulated, or irregular. The proposed methodology was evaluated on two publicly available datasets, the Curated Breast Imaging Subset of Digital Database for Screening Mammography (CBIS-DDSM) and INbreast, and additionally on a private dataset. Comparative experiments were conducted on the individual models and an average ensemble of models with an XGBoost classifier. Qualitative and quantitative results show that the proposed model achieved better performance for (1) Pathology classification with an accuracy of 95.13%, 99.20%, and 95.88%; (2) BI-RADS category classification with an accuracy of 85.38%, 99%, and 96.08% respectively on CBIS-DDSM, INbreast, and the private dataset; and (3) shape classification with 90.02% on the CBIS-DDSM dataset. Our results demonstrate that our proposed integrated framework could benefit from all automated stages to outperform the latest deep learning methodologies. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

With the advent of deep learning, many dense prediction tasks, i.e., tasks that produce pixel-level predictions, have seen significant performance improvements. The typical approach is to learn these tasks in isolation, that is, a separate neural network is trained for each individual task. Yet, recent multi-task learning (MTL) techniques have shown promising results w.r.t. performance, computations and/or memory footprint, by jointly tackling multiple tasks through a learned shared representation. In this survey, we provide a well-rounded view on state-of-the-art deep learning approaches for MTL in computer vision, explicitly emphasizing on dense prediction tasks. Our contributions concern the following. First, we consider MTL from a network architecture point-of-view. We include an extensive overview and discuss the advantages/disadvantages of recent popular MTL models. Second, we examine various optimization methods to tackle the joint learning of multiple tasks. We summarize the qualitative elements of these works and explore their commonalities and differences. Finally, we provide an extensive experimental evaluation across a variety of dense prediction benchmarks to examine the pros and cons of the different methods, including both architectural and optimization based strategies. [ABSTRACT FROM AUTHOR]

• Proposes composite rotation that is composed of inner and outer rotations of image. • Composite rotation based self-supervised task to improve few-shot classification. • Achieves state-of-the-art few-shot classification performance on multiple benchmarks. • Also improves fine-grained, cross-domain and transductive few-shot classification. • Proposes composite rotation based few-shot testing. Many real-life problem settings have classes of data with very few examples for training. Deep learning networks do not perform well for such few-shot classes. In order to perform well in this setting, the networks should learn to extract highly discriminative and generic features. In this paper, we propose to use a composite rotation based self-supervised auxiliary task to improve the representation learning of the network so that it can extract such discriminative features. Our proposed composite rotation based auxiliary task rotates the image at two levels, i.e., it rotates patches inside the image (inner rotation) and also rotates the whole image (outer rotation), and assigns one out of 16 rotation classes to the transformed image. We jointly train the network on the main image classification task and the composite rotation based auxiliary task. This helps the network to learn to extract more generic and discriminative features, which in turn helps to improve its few-shot classification performance. Additionally, during the few-shot testing phase, we consolidate the predictions for images obtained by applying different composite rotation transformations to the same query image to improve the final prediction further. We perform experiments on several few-shot benchmark datasets and empirically show the efficacy of our method for various problem settings. We experimentally show that our method improves the performance of deep learning models on the few-shot classification, fine-grained few-shot classification, cross-domain few-shot classification, and transductive few-shot classification settings. We also experimentally show that models trained using our approach perform better than the baseline even when the query examples in the episode are not aligned with the support examples in the episode. We perform extensive ablation experiments to validate the different components of our approach. We also analyze the effect of our approach on the ability of the network to focus on the discriminative regions of the image. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Chongqing University of Posts & Telecommunications (Natural Science Edition) is the property of Chongqing University of Posts & Telecommunications and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The past decade witnessed a remarkable increase in deep learning (DL) workloads which require GPU resources to accelerate the training process. However, the existing coarse‐grained scheduling mechanisms are agnostic to information other than the number of GPUs or GPU memory, which results in performance degradation of DL tasks. Moreover, the common assumption held by the existing balance‐aware DL task scheduling strategies, a DL task consumes resources once it starts, fails to reduce resource contention, and further limits execution efficiency. To address these problems, this article proposes a fine‐grained and balance‐aware scheduling model (FBSM) which considers the resource consumption characteristic of the DL task. Based on FBSM, we propose customized GPU sniffer (GPU‐S) and balance‐aware scheduler (BAS) modules to construct a scheduling system called KubFBS. The experimental results demonstrate KubFBS accelerates the execution of DL tasks while improving the load balancing capability of the cluster. [ABSTRACT FROM AUTHOR]

We observe that data access and processing takes a significant amount of time in large-scale deep learning training tasks (DLTs) on image datasets. Three factors contribute to this problem: (1) the massive and recurrent accesses to large numbers of small files; (2) the repeated, expensive decoding computation on each image, and (3) the frequent communication between computation nodes and storage nodes. Existing work has addressed some aspects of these problems; however, no end-to-end solutions have been proposed. In this article, we propose DIESEL+, an all-in-one system which accelerates the entire I/O pipeline of deep learning training tasks. DIESEL+ contains several components: (1) local metadata snapshot; (2) per-task distributed caching; (3) chunk-wise shuffling; (4) GPU-assisted image decoding and (5) online region-of-interest (ROI) decoding. The metadata snapshot removes the bottleneck on metadata access in frequent reading of large numbers of files. The per-task distributed cache across the worker nodes of a DLT task to reduce the I/O pressure on the underlying storage. The chunk-based shuffle method converts small file reads into large chunk reads, so that the performance is improved without sacrificing the training accuracy. The GPU-assisted image decoding and the online ROI method minimize the image decoding workloads and reduce the cost of data movement between nodes. These techniques are seamlessly integrated into the system. In our experiments, DIESEL+ outperforms existing systems by a factor of two to three times on the overall training time. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Although deep learning algorithms have achieved significant progress in a variety of domains, they require costly annotations on huge datasets. Self-supervised learning (SSL) using unlabeled data has emerged as an alternative, as it eliminates manual annotation. To do this, SSL constructs feature representations using pretext tasks that operate without manual annotation, which allows models trained in these tasks to extract useful latent representations that later improve downstream tasks such as object classification and detection. The early methods of SSL are based on auxiliary pretext tasks as a way to learn representations using pseudo-labels, or labels that were created automatically based on the dataset's attributes. Furthermore, contrastive learning has also performed well in learning representations via SSL. To succeed, it pushes positive samples closer together, and negative ones further apart, in the latent space. This paper provides a comprehensive literature review of the top-performing SSL methods using auxiliary pretext and contrastive learning techniques. It details the motivation for this research, a general pipeline of SSL, the terminologies of the field, and provides an examination of pretext tasks and self-supervised methods. It also examines how self-supervised methods compare to supervised ones, and then discusses both further considerations and ongoing challenges faced by SSL. [ABSTRACT FROM AUTHOR]

Copyright of Journal of South China University of Technology (Natural Science Edition) is the property of South China University of Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The adoption of electronic health records (EHR) has become universal during the past decade, which has afforded in-depth data-based research. By learning from the large amount of healthcare data, various data-driven models have been built to predict future events for different medical tasks, such as auto diagnosis and heart-attack prediction. Although EHR is abundant, the population that satisfies specific criteria for learning population-specific tasks is scarce, making it challenging to train data-hungry deep learning models. This study presents the Claim Pre-Training (Claim-PT) framework, a generic pre-training model that first trains on the entire pediatric claims dataset, followed by a discriminative fine-tuning on each population-specific task. The semantic meaning of medical events can be captured in the pre-training stage, and the effective knowledge transfer is completed through the task-aware fine-tuning stage. The fine-tuning process requires minimal parameter modification without changing the model architecture, which mitigates the data scarcity issue and helps train the deep learning model adequately on small patient cohorts. We conducted experiments on a real-world pediatric dataset with more than one million patient records. Experimental results on two downstream tasks demonstrated the effectiveness of our method: our general task-agnostic pre-training framework outperformed tailored task-specific models, achieving more than 10% higher in model performance as compared to baselines. In addition, our framework showed a potential to transfer learned knowledge from one institution to another, which may pave the way for future healthcare model pre-training across institutions. [ABSTRACT FROM AUTHOR]

Unsupervised representation learning tends to produce generic and reusable latent representations. However, these representations can often miss high-level features or semantic information, since they only observe the implicit properties of the dataset. On the other hand, supervised learning frameworks learn task-oriented latent representations that may not generalise in other tasks or domains. In this article, we introduce evidence transfer, a deep learning method that incorporates the outcomes of external tasks in the unsupervised learning process of an autoencoder. External task outcomes also referred to as categorical evidence, are represented by categorical variables, and are either directly or indirectly related to the primary dataset-in the most straightforward case they are the outcome of another task on the same dataset. Evidence transfer allows the manipulation of generic latent representations in order to include domain or task-specific knowledge that will aid their effectiveness in downstream tasks. Evidence transfer is robust against evidence of low quality and effective when introduced with related, corresponding, or meaningful evidence. [ABSTRACT FROM AUTHOR]

With the diversification and individuation of user requirements as well as the rapid development of computing technology, the large-scale tasks processing for big data in edge computing environment has become a research focus nowadays. Many recent efforts for task processing are designed and implemented based on some traditional protocols and optimization methods. Therefore, it is more difficult to explore the task allocation strategy that maximizes the overall system revenue from the perspective of global load balancing. In order to overcome this problem, a large-scale tasks processing approach called Federated Learning based Optimization Methodology (FLOM) for large-scale tasks processing was presented to achieve accurate task classification and overall load balancing while satisfying task allocation requirements. FLOM performs the data aggregation and establishes the personalized models by federated learning. The deep network model is designed for deep feature learning of task requests and hosts in the substrate network. The experimental results show the capability of FLOM in terms of large-scale task classification as well as allocation. [ABSTRACT FROM AUTHOR]

The traditional object (person) retrieval (re-identification) task aims to learn a discriminative feature representation with intra-similarity and inter-dissimilarity, which supposes that the objects in an image are manually or automatically pre-cropped exactly. However, in many real-world searching scenarios (e.g., video surveillance), the objects (e.g., persons, vehicles, etc.) are seldom accurately detected or annotated. Therefore, object-level retrieval becomes intractable without bounding-box annotation, which leads to a new but challenging topic, i.e., image-level search with multi-task integration of joint detection and retrieval. In this paper, to address the image search issue, we first introduce an end-to-end Integrated Net (I-Net), which has three merits: 1) A Siamese architecture and an on-line pairing strategy for similar and dissimilar objects in the given images are designed. Benefited by the Siamese structure, I-Net learns the shared feature representation, because, on which, both object detection and classification tasks are handled. 2) A novel on-line pairing (OLP) loss is introduced with a dynamic feature dictionary, which alleviates the multi-task training stagnation problem, by automatically generating a number of negative pairs to restrict the positives. 3) A hard example priority (HEP) based softmax loss is proposed to improve the robustness of classification task by selecting hard categories. The shared feature representation of I-Net may restrict the task-specific flexibility and learning capability between detection and retrieval tasks. Therefore, with the philosophy of divide and conquer, we further propose an improved I-Net, called DC-I-Net, which makes two new contributions: 1) two modules are tailored to handle different tasks separately in the integrated framework, such that the task specification is guaranteed. 2) A class-center guided HEP loss (C $^2$ 2 HEP) by exploiting the stored class centers is proposed, such that the intra-similarity and inter-dissimilarity can be captured for ultimate retrieval. Extensive experiments on famous image-level search oriented benchmark datasets, such as CUHK-SYSU dataset and PRW dataset for person search and the large-scale WebTattoo dataset for tattoo search, demonstrate that the proposed DC-I-Net outperforms the state-of-the-art tasks-integrated and tasks-separated image search models. [ABSTRACT FROM AUTHOR]

Knowledge transfer is a promising concept to achieve real-time decision-making for autonomous vehicles. This paper constructs a transfer deep reinforcement learning (RL) framework to transform the driving tasks in the intersection environments. The driving missions at the unsignalized intersection are cast into a left turn, right turn, and running straight for automated vehicles. The goal of the autonomous ego vehicle (AEV) is to drive through the intersection situation efficiently and safely. This objective promotes the studied vehicle to increase its speed and avoid crashing other vehicles. The decision-making policy learned from one driving task is transferred through three transfer rules in another driving mission and evaluated. Simulation results reveal that the decision-making strategies related to similar tasks are transferable and have a high success rate. It indicates that the presented control framework could reduce time consumption and realize online implementation. Therefore, the transfer RL concept is helpful for establishing the real-time decision-making policy for autonomous vehicles. [ABSTRACT FROM AUTHOR]

This paper deals with the concept of multi-robot task allocation, referring to the assignment of multiple robots to tasks such that an objective function is maximized. The performance of existing meta-heuristic methods worsens as the number of robots or tasks increases. To tackle this problem, a novel Markov decision process formulation for multi-robot task allocation is presented for reinforcement learning. The proposed formulation sequentially allocates robots to tasks to minimize the total time taken to complete them. Additionally, we propose a deep reinforcement learning method to find the best allocation schedule for each problem. Our method adopts the cross-attention mechanism to compute the preference of robots to tasks. The experimental results show that the proposed method finds better solutions than meta-heuristic methods, especially when solving large-scale allocation problems. [ABSTRACT FROM AUTHOR]

In the quest to make deep learning systems more capable, a number of more complex, more computationally expensive and memory intensive algorithms have been proposed. This switchover glosses over the capabilities of many of the simpler systems or modules within them to adequately address current and future problems. This has led to some of the deep learning research being inaccessible to researchers who don't possess top-of-the-line hardware. The use of simple feed forward networks has not been explicitly explored in the current transformer-based vision-language field. In this paper, we use a series of feed-forward layers to encode image features, and caption embeddings, alleviating some of the effects of the computational complexities that accompany the use of the self-attention mechanism and limit its application in long sequence task scenarios. We demonstrate that a decoder does not require masking for conditional short sequence generation where the task is not only dependent on the previously generated sequence, but another input such as image features. We perform an empirical and qualitative analysis of the use of linear transforms in place of self-attention layers in vision-language models, and obtain competitive results on the MSCOCO dataset. Our best feed-forward model obtains average scores of over 90% of the current state-of-the-art pre-trained Oscar model in the conventional image captioning metrics. We also demonstrate that the proposed models take less time training and use less memory at larger batch sizes and longer sequence lengths. [ABSTRACT FROM AUTHOR]

Weather nowcasting consists of predicting meteorological components in the short term at high spatial resolutions. Due to its influence in many human activities, accurate nowcasting has recently gained plenty of attention. In this paper, we treat the nowcasting problem as an image-to-image translation problem using satellite imagery. We introduce Broad-UNet, a novel architecture based on the core UNet model, to efficiently address this problem. In particular, the proposed Broad-UNet is equipped with asymmetric parallel convolutions as well as Atrous Spatial Pyramid Pooling (ASPP) module. In this way, the Broad-UNet model learns more complex patterns by combining multi-scale features while using fewer parameters than the core UNet model. The proposed model is applied on two different nowcasting tasks, i.e. precipitation maps and cloud cover nowcasting. The obtained numerical results show that the introduced Broad-UNet model performs more accurate predictions compared to the other examined architectures. • The novel Broad-UNet model is proposed as an extension of the core UNet model. • The model contains parallel convolutions and Atrous Spatial Pyramid Pooling module. • The model learns more complex patterns by combining multi-scale features. • The model requires fewer parameters than the core UNet model. • The model outperforms other models in precipitation and cloud cover nowcasting tasks. [ABSTRACT FROM AUTHOR]

In this paper, a novel two-stage R-CNN network called ParallelNet is proposed for thigh fracture detection task. In the proposed method, multiple parallel backbone networks and a feature fusion connection structure are designed, which can extract features with different reception fields. Specifically, the first backbone network is denoted as main network, which adopted normal convolution to detect small fractures, the rest backbone networks are denoted as sub-networks which adopted dilated convolution to detect large fractures. We evaluated the proposed method on a thigh fracture dataset containing 3842 X-ray radiographs, 3484 of which is assigned as a training dataset and 358 as a testing dataset. The experiments compare the proposed method with other state-of-the-art deep learning frameworks, including Faster R-CNN, FPN, Cascade R-CNN and RetinaNet, especially DCFPN which focus on thighbone fracture detection task. Our framework achieved 87.8% AP50 and 49.3% AP75 which outperformed other state-of-the-art frameworks. Moreover, ablation experiments on the backbone numbers, connection styles, different dilation rates and the position of dilated convolution have been attempted, and the function of each hyperparameter is analyzed. [ABSTRACT FROM AUTHOR]

Recently, deep learning has been successfully applied to object detection and localization tasks in images. When setting up deep learning frameworks for supervised training with large datasets, strongly labeling the objects facilitates good performance; however, the complexity of the image scene and large size of the dataset make this a laborious task. Hence, it is of paramount importance that the expensive work associated with the tasks involving strong labeling, such as bounding box annotation, is reduced. In this paper, we propose a method to perform object localization tasks without bounding box annotation in the training process by means of employing a two-path activation-map-based classifier framework. In particular, we develop an activation-map-based framework to judicially control the attention map in the perception branch by adding a two-feature extractor so that better attention weights can be distributed to induce improved performance. The experimental results indicate that our method surpasses the performance of the existing deep learning models based on weakly supervised object localization. The experimental results show that the proposed method achieves the best performance, with 75.21% Top-1 classification accuracy and 55.15% Top-1 localization accuracy on the CUB-200-2011 dataset. [ABSTRACT FROM AUTHOR]

Cloud segmentation is a key task in applications of Landsat8 remote sensing images. At present, most existing deep learning based approaches usually adopt the U-type coder-encoder network structure. Although this structure can make use of spatial location information from the coding end effectively, the whole network has too many parameters and cause high computation cost. Meanwhile, the coding end only adopts simple convolution and subsampling operation, which makes it impossible to obtain high-quality contextual semantic feature information effectively. In view of the above situation, this paper proposed a new lightweight U-Net model for cloud segmentation. The whole model adopted a new U-shaped codec structure that jumped and connected the information between the shallow layer and the middle layer at the coding end, and also embedded novel optimization blocks consist of grouping convolution and attention mechanism in coding end. It also built context-semantic fusion connection to link corresponding layers in coding and decoding end. Experimental results on the common benchmark dataset 38-Cloud illustrates, compared with other mainstream cloud segmentation networks, the proposed new U-Net can achieve better segmentation precision with less model parameters. [ABSTRACT FROM AUTHOR]

• New model for scene text VQA that jointly reasons about textual and visual modalities. • Attending on multi-modal features is better than attending separately to each modality. • Grid features from an object detection backbone proves to work well in this task. • The proposed model is up to 5 times faster than current state of the art models. • Combined with a standard VQA model reaches state of the art on mixed datasets. This paper presents a new model for the task of scene text visual question answering. In this task questions about a given image can only be answered by reading and understanding scene text. Current state of the art models for this task make use of a dual attention mechanism in which one attention module attends to visual features while the other attends to textual features. A possible issue with this is that it makes difficult for the model to reason jointly about both modalities. To fix this problem we propose a new model that is based on an single attention mechanism that attends to multi-modal features conditioned to the question. The output weights of this attention module over a grid of multi-modal spatial features are interpreted as the probability that a certain spatial location of the image contains the answer text to the given question. Our experiments demonstrate competitive performance in two standard datasets with a model that is × 5 faster than previous methods at inference time. Furthermore, we also provide a novel analysis of the ST-VQA dataset based on a human performance study. Supplementary material, code, and data is made available through this link. [ABSTRACT FROM AUTHOR]

This article is concerned with the multitask multiagent allocation problem via many-objective optimization for multiagent systems (MASs). First, a novel layered MAS model is constructed to address the multitask multiagent allocation problem that includes both the original task simplification and the many-objective allocation. In the first layer of the model, the deep Q-learning method is introduced to simplify the prioritization of the original task set. In the second layer of the model, the modified shift-based density estimation (MSDE) method is put forward to improve the conventional strength Pareto evolutionary algorithm 2 (SPEA2) in order to achieve many-objective optimization on task assignments. Then, an MSDE-SPEA2-based method is proposed to tackle the many-objective optimization problem with objectives including task allocation, makespan, agent satisfaction, resource utilization, task completion, and task waiting time. As compared with the existing allocation methods, the developed method in this article exhibits an outstanding feature that the task assignment and the task scheduling are carried out simultaneously. Finally, extensive experiments are conducted to: 1) verify the validity of the proposed model and the effectiveness of two main algorithms and 2) illustrate the optimal solution for task allocation and efficient strategy for task scheduling under different scenarios. [ABSTRACT FROM AUTHOR]

• Proposed a novel semantic segmentation networks based on elements of a transformer. • Captures scale contextual information according to original contextual information. • Effectively segments objects of various scales in diverse real-world environments. • A series of ablation studies are conducted to show efficacy of our method. • Our method exhibits superior performance compared to state-of-the-art methods. Recent efforts in semantic segmentation using deep learning frameworks have made notable advances. However, capturing the existence of objects in an image at multiple scales still remains a challenge. In this paper, we address the semantic segmentation task based on transformer architecture. Unlike existing methods that capture multi-scale contextual information through infusing every single-scale piece of information from parallel paths, we propose a novel semantic segmentation network incorporating a transformer (TrSeg) to adaptively capture multi-scale information with the dependencies on original contextual information. Given the original contextual information as keys and values, the multi-scale contextual information from the multi-scale pooling module as queries is transformed by the transformer decoder. The experimental results show that TrSeg outperforms the other methods of capturing multi-scale information by large margins. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Beijing University of Technology is the property of Journal of Beijing University of Technology, Editorial Department and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The detection speed of detection algorithms based on artificial design features is generally slow, and the detection accuracy needs to be improved, which can no longer meet the needs of today's industrial production. The detection technology based on deep learning cannot be deployed on resource-constrained devices because it requires a lot of computing and storage space. In response to these problems, this paper cited a channel pruning method to realize the lightweight of the YOL0v3 detection network, and obtained the pruning model Slim YOLOv3, and proposed to use Slim YOLOv3 for real-time detection tasks in industrial scenarios. The method enhanced the channel-level sparsity of the convolutional layer by applying L, regularization to the channel scaling factor, and pruned the feature channels with less information, and obtained a lightweight network model. Compared with the original model, SlimYOLOv3 gives a 60% reduction in model size and a 50% reduction in computing operations, the detection speed is 1. 7 times of the original model. It is more suitable for real-time detection of complex targets in intelligent industrial scenes. [ABSTRACT FROM AUTHOR]

• Study of VQA systems for cross-modal retrieval tasks. • Baselines in several classification and cross-modal tasks for well know databases. • Evaluation of multiple cross-modal retrieval tasks from an end-to-end system. • Approaches that improve the search experience by using all data present in articles. • End-to-end approach to perform specialized unimodal and crossmodal retrieval. We investigate the effectiveness of a successful model in Visual-Question-Answering (VQA) problems as the core component in a cross-modal retrieval system that can accept images or text as queries, in order to retrieve relevant data from a multimodal document collection. To this end, we adapt the VQA model for deep multimodal learning to combine visual and textual representations for information search, and we call this model "Deep Multimodal Embeddings (DME)". Instead of training the model to answer questions, we supervise DME to classify semantic topics/concepts previously identified in the document collection of interest. In contrast to previous approaches, we found that this model can handle any multimodal query with a single architecture while producing improved or competitive results in all retrieval tasks. We evaluate the model performance with 3 widely known databases for cross-modal retrieval tasks: Wikipedia Retrieval Database, Pascal Sentences, and MIR-Flickr-25k. The results show that the DME model learns effective multimodal representations, resulting in strongly improved retrieval performance, specifically in the top results of the ranked list, which are the most important to users in the most-common scenarios of information retrieval. Our work represents a new baseline for a wide set of different methodologies for cross-modal retrieval. [ABSTRACT FROM AUTHOR]

To explore whether the brain contains pattern differences in the rock–paper–scissors (RPS) imagery task, this paper attempts to classify this task using fNIRS and deep learning. In this study, we designed an RPS task with a total duration of 25 min and 40 s, and recruited 22 volunteers for the experiment. We used the fNIRS acquisition device (FOIRE-3000) to record the cerebral neural activities of these participants in the RPS task. The time series classification (TSC) algorithm was introduced into the time-domain fNIRS signal classification. Experiments show that CNN-based TSC methods can achieve 97% accuracy in RPS classification. CNN-based TSC method is suitable for the classification of fNIRS signals in RPS motor imagery tasks, and may find new application directions for the development of brain–computer interfaces (BCI). [ABSTRACT FROM AUTHOR]

The rapid development of artificial intelligence technology has made deep neural networks (DNNs) widely used in various fields. DNNs have been continuously growing in order to improve the accuracy and quality of the models. Moreover, traditional data/model parallelism is hard to expand due to communication bottlenecks and hardware efficiency issues. However, pipeline parallelism trains multiple batches, reducing training overheads, so that it can achieve better acceleration effect. Considering the complexity of solving the pipeline parallel task allocation problem in heterogeneous computing resources, in this paper, a task allocation in pipeline parallelism (TAPP) based on deep reinforcement learning, is proposed. In TAPP, the predictive network is trained by a policy gradient until it obtains the optimal pipeline parallel task allocation scheme and speeds up the model training. Experimental results show that, on average, the single-step training time of TAPP is decreased by 1.37 times and the proportion of communication time is reduced by 48.92%, compared with the data parallelism, bulk synchronous parallel (BSP). [ABSTRACT FROM AUTHOR]

Copyright of Eurasian Journal of Educational Research (EJER) is the property of Eurasian Journal of Educational Research and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

With the significant development of black-box machine learning algorithms, particularly deep neural networks, the practical demand for reliability assessment is rapidly increasing. On the basis of the concept that "Bayesian deep learning knows what it does not know," the uncertainty of deep neural network outputs has been investigated as a reliabilitymeasure for classification and regression tasks. By considering an embedding task as a regression task, several existing studies have quantified the uncertainty of embedded features and improved the retrieval performance of cutting-edge models by model averaging. However, in image-caption embedding-and-retrieval tasks, well-known samples are not always easy to retrieve. This study shows that the existing method has poor performance in reliability assessment and investigates another aspect of imagecaption embedding-and-retrieval tasks. We propose posterior uncertainty by considering the retrieval task as a classification task, which can accurately assess the reliability of retrieval results. The consistent performance of the two uncertainty measures is observed with different datasets (MS-COCO and Flickr30k), different deep-learning architectures (dropout and batch normalization), and different similarity functions. To the best of our knowledge, this is the first study to perform a reliability assessment on image-caption embedding-and-retrieval tasks. [ABSTRACT FROM AUTHOR]

Purpose: Existing virtual agents (VAs) present in dialogue systems are either information retrieval based or static goal-driven. However, in real-world situations, end-users might not have a known and fixed goal beforehand for the task, i.e., they may upgrade/downgrade/update their goal components in real-time to maximize their utility values. Existing VAs are unable to handle such dynamic goal-oriented situations. Methodology: Due to the absence of any related dialogue dataset where such choice deviations are present, we have created a conversational dataset called Deviation adapted Virtual Agent(DevVA), with the manual annotation of its corresponding intents, slots, and sentiment labels. A Dynamic Goal Driven Dialogue Agent (DGDVA) has been developed by incorporating a Dynamic Goal Driven Module (GDM) on top of a deep reinforcement learning based dialogue manager. In the course of a conversation, the user sentiment provides grounded feedback about agent behavior, including goal serving action. User sentiment appears to be an appropriate indicator for goal discrepancy that guides the agent to complete the user's desired task with gratification. The negative sentiment expressed by the user about an aspect of the provided choice is treated as a discrepancy that is being resolved by the GDM depending upon the observed discrepancy and current dialogue state. The goal update capability and the VA's interactiveness trait enable end-users to accomplish their desired task satisfactorily. Findings: The obtained experimental results illustrate that DGDVA can handle dynamic goals with maximum user satisfaction and a significantly higher success rate. The interaction drives the user to decide its final goal through the latent specification of possible choices and information retrieved and provided by the dialogue agent. Through the experimental results (qualitative and quantitative), we firmly conclude that the proposed sentiment-aware VA adapts users' dynamic behavior for its goal setting with substantial efficacy in terms of primary objective i.e., task success rate (0.88). Practical implications: In real world, it can be argued that many people do not have a predefined and fixed goal for tasks such as online shopping, movie booking & restaurant booking, etc. They tend to explore the available options first which are aligned with their minimum requirements and then decide one amongst them. The DGDVA provides maximum user satisfaction as it enables them to accomplish a dynamic goal that leads to additional utilities along with the essential ones. Originality: To the best of our knowledge, this is the first effort towards the development of A Dynamic Goal Adapted Task-Oriented Dialogue Agent that can serve user goals dynamically until the user is satisfied. [ABSTRACT FROM AUTHOR]

Traditional humor computing tasks rely on artificially constructed features, which are prone to loss of features, and mainly focus on humorous judgment. The Siamese bidirectional GRU attention model based on deep learning compared pairs of humorous texts to determine which sentence was more humorous. Firstly, this paper used the text processor to train the word embedding. Secondly, it used the bidirectional GRU model to obtain the annotation of each word. Finally, it performed the humorous comparison task at the fully connected layer. This paper conducted experiments on the Semeval-2017 Task6-#HashtagWars data set, and used accuracy as an evaluation indicator. The experimental results show that the model and other related models have a significant improvement in the comparison of humorous text. [ABSTRACT FROM AUTHOR]

In this paper, we present a novel deep reinforcement learning (DRL) based method that is used to perform multi-robot task allocation (MRTA) and navigation in an end-to-end fashion. The policy operates in a decentralized manner mapping raw sensor measurements to the robot's steering commands without the need to construct a map of the environment. We also present a new metric called the Task Allocation Index ( T A I ), which measures the performance of a method that performs MRTA and navigation from end-to-end in performing MRTA. The policy was trained on a simulated gazebo environment. The centralized learning and decentralized execution paradigm was used for training the policy. The policy was evaluated quantitatively and visually. The simulation results showed the effectiveness of the proposed method deployed on multiple Turtlebot3 robots. [ABSTRACT FROM AUTHOR]

This paper describes how a simple novel Galois Power-of-Two (GPOW2) real-time embedding scheme is used to improve the performance and accuracy of downstream NLP tasks. GPOW2 computes embeddings live on the fly (real time) in the context of target NLP tasks without the need for tabulated pre-embeddings. One excellent feature of the method is the ability to capture multilevel embeddings in the same pass. It simultaneously computes character, word and sentence embeddings on the fly. GPOW2 has been derived in the context of attempts to improve the performance of the SWAM Arabic morphological engine, which is a multipurpose tool that supports segmentation, classification, POS tagging, spell checking, word embeddings, sematic search, among other tasks. SWAM is a pattern-oriented algorithm that relies on morphological patterns and POS tagging to perform NLP tasks. The paper demonstrates how GPOW2 led to improvements in the accuracy of POS tagging and pattern matching, and accordingly the performance of the whole engine. The accuracy for pattern prediction is 99.47% and is 98.80% for POS tagging. [ABSTRACT FROM AUTHOR]

Copyright of Procesamiento del Lenguaje Natural is the property of Sociedad Espanola para el Procesamiento del Lenguaje Natural and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Background: With the development of deep learning (DL), more and more methods based on deep learning are proposed and achieve state-of-the-art performance in biomedical image segmentation. However, these methods are usually complex and require the support of powerful computing resources. According to the actual situation, it is impractical that we use huge computing resources in clinical situations. Thus, it is significant to develop accurate DL based biomedical image segmentation methods which depend on resources-constraint computing. Results: A lightweight and multiscale network called PyConvU-Net is proposed to potentially work with low-resources computing. Through strictly controlled experiments, PyConvU-Net predictions have a good performance on three biomedical image segmentation tasks with the fewest parameters. Conclusions: Our experimental results preliminarily demonstrate the potential of proposed PyConvU-Net in biomedical image segmentation with resources-constraint computing. [ABSTRACT FROM AUTHOR]

Blockchain is an advanced technique to realize smart contracts, various transactions, and P2P crypto-currencies in the e-commerce society. However, the traditional blockchain does not consider a mobile environment to design a data offloading of the blockchain such that the blockchain results in high computational cost and huge data propagation delay. In this paper, to remedy the above problem, we propose a scalable blockchain and a task offloading technique based on the neural network of the mobile edge computing scenario. Experimental results show that our approach is very scalable in the mobile scenario. [ABSTRACT FROM AUTHOR]

Deep learning has been increasingly used in bioinformatics, especially in sequence-based protein prediction tasks, as large amounts of biological data are available and deep learning techniques have been developed rapidly in recent years. For sequence-based protein prediction tasks, the selection of a suitable model architecture is essential, whereas sequence data representation is a major factor in controlling model performance. Here, we summarized all the main approaches that are used to represent protein sequence data (amino acid sequence encoding or embedding), which include end-to-end embedding methods, non-contextual embedding methods and embedding methods that use transfer learning and others that are applied for some specific tasks (such as protein sequence embedding based on extracted features for protein structure predictions and graph convolutional network-based embedding for drug discovery tasks). We have also reviewed the architectures of various types of embedding models theoretically and the development of these types of sequence embedding approaches to facilitate researchers and users in selecting the model that best suits their requirements. [ABSTRACT FROM AUTHOR]

With the development of deep learning, researchers began to explore the application of deep learning to person re-identification tasks and proposed a large number of methods, which also ushered in new challenges. In order to facilitate scholars to systematically understand the research status and development trends in this field, this paper firstly introduced the person re-identification task and existing problems. Secondly, according to training methods, it discussed the research progress of the supervised learning, semi-supervised learning/weak supervision learning and unsupervised learning for the person re-identification task. It introduced the application of GAN (generative adversarial network) and attention mechanism in person re-identification following recent research hotspots. After that, it enumerated some commonly used data sets (Market-1501, CUHK03, etc.) and their performance in this field. Finally, this paper looked forward to the future research direction of person re-identification. [ABSTRACT FROM AUTHOR]