Multiple Subtasks Research Articles

Straggler-Exploiting Fully Private Distributed Matrix Multiplication With Chebyshev Polynomials

In this paper, we consider coded computation for matrix multiplication tasks in distributed computing to mitigate straggler effects. We assume that the stragglers’ computation results can be leveraged at the master by assigning multiple sub-tasks to the workers. We propose a new coded computation scheme, namely Chebyshev coded fully private matrix multiplication (CFP), to preserve the privacy of a master in a scenario where a master wants to obtain a matrix multiplication result from the libraries which are shared by the workers, while concealing both of the two indices of the desired matrices from each worker. The key idea of CFP is to introduce Chebyshev polynomials, which have commutative property, in queries sent to workers to allocate sub-tasks. We also extend CFP to keep the privacy of a master from colluding workers. In conclusion, we show that CFP can preserve the privacy of a master from each worker and efficiently mitigate straggler effects compared to existing schemes.

Cooperative modular reinforcement learning for large discrete action space problem

Deep reinforcement learning (DRL) has achieved remarkable results on high-dimension state tasks. However, it suffers in hard convergence and low sample efficiency when solving large discrete action space problems. To meet these challenges, we develop a cooperative modular reinforcement learning (CMRL) method to distributedly solve the problems with a large discrete action space. A general yet effective task decomposition method is proposed to decompose the complex decision task in a large action space into multiple decision sub-tasks in small action subsets, using a rule-based action division method. The CMRL method consisting of multiple Critic networks is proposed to settle the multiple sub-tasks, where each Critic network learns a decomposed value function to obtain the local optimal action in a sub-task. The global optimal action is cooperatively chosen by all local optimal actions. Moreover, we propose a new parallel training mechanism, which trains multiple Critic networks with different models and multi-data in parallel. Mathematical properties are proposed to analyze the rationality and superiority of CMRL. Four different simulation experiments are conducted to verify the generality and effectiveness of CMRL for large action space problems. The results show that CMRL has superior performance on training efficiency compared with classical and latest DRL methods while maintaining the accuracy of the solution.

Multi-task Joint Learning to Enhance Named Entity Recognition

Named Entity Recognition (NER) models have achieved good performance in recent years but also have some shortcomings. Existing models regard NER as a sequence labeling task for label prediction, without considering the impact of different stages in the entity recognition process on the final result. NER can be viewed as two separate subtasks: boundary detection task and type prediction task. The two subtasks can transmit information and cooperate in the process of entity recognition, so the synergy of the two subtasks is beneficial to NER. In this paper, we propose a method to split the NER task into multiple subtasks and use the information of each subtask to enhance the NER task. According to the characteristics of different subtasks, we use different feature extraction method models to extract structural information useful for this task effectively. Using the information extracted from each subtask or the final results of the subtasks, the performance of NER was enhanced through the gating network. We conducted extensive experiments on the CONLL2003 dataset, and the experimental results show that our proposed multi-task joint learning enhances the effectiveness of the named entity recognition model.

Differences in the difficulty of subtasks comprising the toileting task among patients with subacute stroke: A cohort study

IntroductionToileting comprises multiple subtasks, and the difficulty of each is critical to determining the target and priority of intervention. The study aimed to examine the difficulty of subtasks that comprise toileting upon admission and the reacquisition of skills of subtasks during hospitalization. Materials and methodsThis was a single-center prospective cohort study. We enrolled 101 consecutive stroke patients (mean age: 69.3 years) admitted to subacute rehabilitation wards. The independence in each of the 24 toileting subtasks was assessed using the Toileting Tasks Assessment Form (TTAF) every two or four weeks. The number of patients who were independent upon admission, as well as those who were not independent upon admission but became independent during hospitalization, was examined in each subtask. ResultsThe most difficult subtask upon admission was “Lock the wheelchair brakes” (16.8% of patients were independent), followed by “Turn while standing (before urination/defecation)” (17.8%), “Pull the lower garments down” (18.0%), “Turn while standing (after urination/defecation)” (18.8%), “Pull the lower garments up and adjust them” (18.8%), and “Maintain a standing position (before urination/defecation)” (18.8%). The most difficult subtask for those who were not independent but became independent was “Dispose of incontinence pad/sanitary items” (19.3%), followed by “Press the nurse call button (after urination/defecation)” (28.3%), “Take the foot off the footrest and place it on the ground” (28.6%), and “Clean up after urination/defecation” (29.0%). ConclusionsThe difficult subtasks upon admission and those for reacquired skills were different. The most difficult subtasks upon admission were main tasks, and the difficult subtasks in reacquiring skills were preparatory tasks.

Evaluation of Stoffenmanager® and ART for Estimating Occupational Inhalation Exposures to Volatile Liquids.

In practice, workers often handle the same chemical(s) of interest under different control measures (e.g. local ventilation, enclosed system) during a full shift. Stoffenmanager® allows users to predict either task-based or full-shift exposures. However, most previous studies evaluated the tool by comparing task-based exposures with measured exposures. Also, limited evaluation studies of the Advanced REACH Tool (ART) with the Bayesian approach (ART+B) are available, requiring additional evaluation studies. The performance of Stoffenmanager® and ART with and without the Bayesian approach was evaluated with measured full-shift exposures to volatile liquids in terms of accuracy, precision, and conservatism. Forty-two exposure situation scenarios (including 251 exposures), developed based on job tasks and chemicals handled during tasks from workplaces, were used to generate full-shift estimates. The estimates were then compared with measured exposures using various comparison methods. Overall, Stoffenmanager® appeared to be the most accurate among the testing tools, while ART+B was the most precise. The percentage of measured exposures exceeding the tools' 90th percentile estimates (%M>T) demonstrated that Stoffenmanager® (16%M>T) and ART+B (13%M>T) were more conservative than ART (41%M>T). When the 90% upper confidence limit of the 90th percentile estimate was considered, the level of conservatism changed from low (41%M>T) to medium (17%M>T) for ART and from medium (13%M>T) to high (0.8%M>T) for ART+B. The findings of this study indicate that no single tool would work for all ESs. Thus, it is recommended that users select a tool based on the performance results of three components (i.e. accuracy, precision, and conservatism), not depending on one or two components. The strength of this study is that the required tools' input parameters were obtained during the sample collection to minimize assumptions for many input parameters. In addition, unlike other previous studies, multiple subtasks, which happen often in workplaces, were incorporated in this study. Nevertheless, the present study did not cover all activities listed in the tools and was limited to volatile liquids, suggesting further studies cover other exposure categories (e.g. solid, metal) and diverse activities.

Task-Offloading and Resource Allocation Strategy in Multidomain Cooperation for IIoT

This study proposes a task-offloading and resource allocation strategy in multidomain cooperation (TARMC) for the industrial Internet of Things (IIoT) to resolve the problem of the non-uniform distribution of task computation among various cluster domain networks in the IIoT and the solidification of traditional industrial wireless network architecture, which produces low efficiency of static resource allocation and high delay in closed-loop data processing. Based on the closed-loop process of task interaction of intelligent terminals in wireless networks, the proposed strategy constructs a network model of multidomain collaborative task-offloading and resource allocation in IIoT for flexible and dynamic resource allocation among intelligent terminals, edge servers, and cluster networks. Considering the partial offloading mechanism, various tasks were segmented into multiple subtasks marked at bit-level per demand, which enabled local and edge servers to process all subtasks in parallel. Moreover, this study established a utility function for the closed-loop delay and terminal energy consumption of task processing, which transformed the process of multidomain collaborative task-offloading and resource allocation into the problem of task computing revenue. Furthermore, an improved Cuckoo Search algorithm was developed to derive the optimal offloading position and resource allocation decision through an alternating iterative method. The simulation results revealed that TARMC performed better than strategies.

Modularized Mutuality Network for Emotion-Cause Pair Extraction

Emotion-cause pair extraction (ECPE) is an emerging task born out of Emotion cause extraction (ECE), which aims to extract the emotion clause and the corresponding cause clause simultaneously. Previous methods decompose ECPE into multiple sub-tasks, namely emotion clause extraction, cause clause extraction, and emotion-cause pair extraction, and employ different modules to address them separately. However, these methods fail to effectively capture the mutuality within the three sub-tasks, which may hinder the information interaction between emotion and cause. In this paper, we revisit and analyze the mutuality between emotion and cause clauses from a linguistic perspective and further propose a novel Modularized Mutuality Network (MMN) to capture the mutuality explicitly. Specifically, the mutuality can be divided into the following categories, including position bias, sentiment consistency, and natural duality. To this end, we design three modules wrapped with various simple but effective mechanisms to address the mutuality, respectively. Extensive experiments demonstrate that MMN achieves state-of-the-art performances on the ECPE task and detailed analyzed the effect of the three modules for capturing the mutuality within sub-tasks.

A General Global and Local Pre-Training Framework for 3D Medical Image Segmentation.

Accurate target segmentation from computed tomography (CT) scans is crucial for surgical robots to perform clinical surgeries successfully. However, the lack of medical image data and annotations has been the biggest obstacle to learning robust medical image segmentation models. Self-supervised learning can effectively address this problem by providing a strategy to pre-train a model with unlabeled data, and then fine-tune downstream tasks with limited labeled data. Existing self-supervised methods fail to simultaneously utilize the abundant global anatomical structure information and local feature differences in medical imaging. In this work, we propose a new strategy for the pre-training framework, which uses the three-dimensional anatomical structure of medical images and specific task and background cues to segment volumetric medical images with limited annotations. Specifically, we propose (1) learning intrinsic patterns of volumetric medical image structures through multiple sub-tasks, and (2) designing a multi-level background cube contrastive learning strategy to enhance the target feature representation by exploiting the differences between the specific target and background. We conduct extensive evaluations on two publicly available datasets. Under limited annotation settings, the proposed method yields significant improvements compared to other self-supervised learning techniques. The proposed method achieves within 6% of the baseline performance using only five labeled CT volumes for training. Once the paper is online, the code and dataset will be available at https://github.com/PinkGhost0812/SGL.

P&T-Inf: A Result Inference Method for Context-Sensitive Tasks in Crowdsourcing

Context-Sensitive Task (CST) is a complex task type in crowdsourcing, such as handwriting recognition, route plan, and audio transcription. The current result inference algorithms can perform well in simple crowdsourcing tasks, but cannot obtain high-quality inference results for CSTs. The conventional method to solve CSTs is to divide a CST into multiple independent simple subtasks for crowdsourcing, but this method ignores the context correlation among subtasks and reduces the quality of result inference. To solve this problem, we propose a result inference algorithm based on the Partially ordered set and Tree augmented naive Bayes Infer (P&T-Inf) for CSTs. Firstly, we screen the candidate results of context-sensitive tasks based on the partially ordered set. If there are parallel candidate sets, the conditional mutual information among subtasks containing context information in external knowledge (such as Google n-gram corpus, American Contemporary English corpus, etc.) will be calculated. Combined with the tree augmented naive (TAN) Bayes model, the maximum weighted spanning tree is used to model the dependencies among subtasks in each CST. We collect two crowdsourcing datasets of handwriting recognition tasks and audio transcription tasks from the real crowdsourcing platform. The experimental results show that our approach improves the quality of result inference in CSTs and reduces the time cost compared with the latest methods.

A Memetic Algorithm for Cooperative Complex Task Offloading in Heterogeneous Vehicular Networks

With the booming of intelligent connected vehicles as well as the emergence of edge computing paradigm, complex task offloading becomes a critical yet promising issue in vehicular networks to enable various real-time and scalable future intelligent systems. This article makes the first effort on proposing an end-edge-cloud cooperation architecture together with a tailored memetic algorithm for complex task offloading in heterogeneous vehicular networks. Specifically, we consider the scenario where a complex task consists of multiple subtasks, which require different amount of resources for being processed, and a task is completed only when all of its subtasks have been processed. Then, we formulate a Cooperative Complex Task Offloading (CCTO) problem by considering heterogeneous computation, communication and memory capacities of nodes in vehicular networks, as well as task dependency and mobility of vehicles, targeting at minimizing average service delay of the system. We prove that CCTO is NP-hard by constructing a polynomial time reduction from the parallel machines scheduling problem (PMSP). Further, we propose a memetic computing based algorithm named MDMA (Meme Dependency-aware Memetic Algorithm), which consists of a meme dependency based encoding solution, a mix-strategy for initialization, a dedicated offspring generation scheme, a meme recombined strategy for local search, and a task feature driven method for repairing infeasible solutions. Finally, we build a simulation model and give a comprehensive performance evaluation. The results demonstrate the superiority of MDMA on minimizing the service delay by best exploiting heterogeneous resources in vehicular networks.

Software-Defined Heterogeneous Edge Computing Network Resource Scheduling Based on Reinforcement Learning

With the rapid development of wireless networks, wireless edge computing networks have been widely considered. The heterogeneous characteristics of the 6G edge computing network bring new challenges to network resource scheduling. In this work, we consider a heterogeneous edge computing network with heterogeneous edge computing nodes and task requirements. We design a software-defined heterogeneous edge computing network architecture to separate the control layer and the data layer. According to different requirements, the tasks in heterogeneous edge computing networks are decomposed into multiple subtasks at the control layer, and the edge computing node alliance responding to the tasks is established to perform the decomposed subtasks. In order to optimize both network energy consumption and network load balancing, we model the resource scheduling problem as a Markov Decision Process (MDP), and design a Proximal Policy Optimization (PPO) resource scheduling algorithm based on deep reinforcement learning. Simulation analysis shows that the proposed PPO resource scheduling can achieve low energy consumption and ideal load balancing.

Multi-target assignment hunting strategy of UAV swarm based on improved K-means algorithm and shortest time mechanism

Multi-target hunting of UAV swarm is an important tactical means. This paper proposes a hunting strategy based on improved K-means and the shortest time mechanism. The large-scale task assignment problem is complex in structure and difficult to solve. To obtain higher hunting efficiency and reduce the amount of calculation on the single UAV, the hybrid architecture is used to decompose the complex multi-target hunting problem into a set of tasks that the UAV need to perform, which reduces the coupling of the system and the complexity of problem. Firstly, the multi-target hunting problem is stratified by the improved K-means algorithm to form multiple independent single target hunting subsystems. In the subsystem, the single target hunting task is decomposed into multiple subtasks that are easy to be executed by UAVs, and a one-to-one matching relationship between subtasks and UAVs is established by using the shortest time mechanism. UAV swarm can achieve multi-target hunting only by executing subtasks. The simulation results show that the UAV swarm can effectively allocate the multi-target hunting problem, which proves the effectiveness of the allocation strategy is proved.

DeTTO: Dependency-Aware Trustworthy Task Offloading in Vehicular IoT

In this paper, we investigate the dependency-aware trustworthy task offloading problem (DeTTO), especially in an IoT-enabled vehicular network, where a large computation-intensive task offloaded from a vehicle is fragmented into multiple subtasks and then offloaded to multiple trusted nodes. First, we formulate the task offloading problem as a graph optimization problem intending to find an optimal set of trustworthy nodes for offloading the subtasks. We aim to minimize the task completion delay and energy consumption, while satisfying the dependency relations between the subtasks and the trust requirements of the tasks. We consider three types of dependency structures – fully independent task, fully dependent task, and partially dependent task. For a fully independent task with no dependency between the subtasks, we propose a greedy algorithm to get the optimal set of nodes for task offloading. After showing the NP-hardness of solving the dependent task offloading problem, we propose a two-fold efficient heuristic approach for the tasks with all dependent subtasks. We adopt the solution approaches used by the first two types of tasks for a partially dependent task. Through simulation experiments, we analyze the performance of the proposed algorithms for three types of intra-task dependencies. The experimental results show that the proposed algorithms significantly reduce the delay and energy consumption, when compared to the benchmark schemes.

Tracking vision transformer with class and regression tokens

Aim of this work is to introduce a novel visual object tracking model based on siamese network and vision transformer. Tracking is performed by multiple tokens exploiting the learning and memorization capabilities of the vision transformers. Therefore, the tracking problem is divided into multiple sub-tasks and experiments by using multiple tokens for learning each individual sub-task. This makes possible to learn a robust characterization of the problem with an explainable architecture, understanding the motivation of the choice that the neural network does. This is due to the attention in the transformer that uses the representational capacity of tokens that allows one to identify, simply with respect to different architectures and methodologies, where all the interest is focused. Several experiments are performed on benchmark data proving to be among the most performing trackers compared with the state of the art in explainability, precision, robustness and speed.

Learning and Generalizing Cooperative Manipulation Skills Using Parametric Dynamic Movement Primitives

This paper presents an approach that generates the overall trajectory of mobile manipulators for a complex mission consisting of several sub-tasks. Parametric dynamic movement primitives (PDMPs) can quickly generalize the online motion of robot manipulation by learning multiple demonstrations in offline. However, regarding complex missions consisting of multiple sub-tasks, a large number of demonstrations are required for full generalization, which is impractical. In this paper, we propose a framework that reduces the number of demonstrations for a complex mission. In the proposed method, complex demonstrations are segmented into multiple unit motions representing sub-tasks, and one PDMP is formed per each segment, resulting in multiple PDMPs. The phase decision process determines which sub-task and associated PDMPs to be executed online, allowing multiple PDMPs to be autonomously configured within an integrated framework. In order to generalize the execution time and regional goal in each phase, the Gaussian process regression (GPR) is applied. Simulation results from two different scenarios confirm that the proposed framework not only effectively reduces the number of demonstrations but also improves generalization performance. The actual experiments also demonstrate that the mobile manipulators effectively perform complex missions through the proposed framework. Note to Practitioners—This paper presents an approach of learning from demonstration (LfD) to generalize complex movements of robots. Parametric dynamic movement primitives (PDMPs) compute styles of movements from multiple demonstrations. However, the complexity of the PDMP increases as the mission involves more sub-tasks. In this paper, we resolve this issue by segmenting the complex mission into multiple sub-tasks and configuring multiple PDMPs. This work effectively reduces the number of required demonstrations for PDMPs, moderates the complexity of the algorithm. Also, the proposed approach allows flexible sub-task sequencing. It enables the mission in an unlearned sequence or a new combination of sub-tasks. The proposed approach is validated in both simulation and experimental results. Our approach is applicable for complex missions whose sub-tasks are clearly identified

Difficulty of the subtasks comprising bed-wheelchair transfer in patients with subacute strokes: A cohort study

ObjectiveBed-wheelchair transfer comprises multiple subtasks; hence, it is important to know the difficulty of each subtask to identify and prioritize subtasks that must be practiced. This study aimed to investigate the difficulty of the subtasks that comprise bed-wheelchair transfer upon admission and reacquiring subtask skills during hospitalization. Materials and methodsThis was a single-center prospective cohort study. We enrolled 137 consecutive stroke patients (mean age: 69.8 years) admitted to subacute rehabilitation wards who used wheelchairs upon admission. The degree of independence in each of the 25 subtasks that comprised transferring was assessed using the Bed-wheelchair transfer Tasks Assessment Form every 2 weeks. The number of patients who were independent in the subtasks upon admission and those who were not but became independent during hospitalization were examined. ResultsThe most difficult subtask for independent patients upon admission was "Manipulate the handrail for the bed" (18.3%), followed by "Ready the wheelchair for transfer" (19.3%), "Maneuver the wheelchair toward the appropriate place for transfer to the bed" (20.6%), "Wear shoes/brace" (24.8%), and "Turn while standing" (25.5%). The most difficult subtask for those who were not independent but became independent was "Ready the wheelchair for transfer" (32.1%), followed by "Manipulate the handrail for the bed" (32.9%), "Press the nurse call button" (36.4%), "Press the nurse call button (wheelchair-to-bed)" (36.7%), and "Lock the wheelchair brakes" (37.3%). ConclusionsSubtasks related to preparation for transfer were difficult upon admission, and this tendency became more pronounced during the skill acquisition process.

Speech GAU: A Single Head Attention for Mandarin Speech Recognition for Air Traffic Control

The rise of end-to-end (E2E) speech recognition technology in recent years has overturned the design pattern of cascading multiple subtasks in classical speech recognition and achieved direct mapping of speech input signals to text labels. In this study, a new E2E framework, ResNet–GAU–CTC, is proposed to implement Mandarin speech recognition for air traffic control (ATC). A deep residual network (ResNet) utilizes the translation invariance and local correlation of a convolutional neural network (CNN) to extract the time-frequency domain information of speech signals. A gated attention unit (GAU) utilizes a gated single-head attention mechanism to better capture the long-range dependencies of sequences, thus attaining a larger receptive field and contextual information, as well as a faster training convergence rate. The connectionist temporal classification (CTC) criterion eliminates the need for forced frame-level alignments. To address the problems of scarce data resources and unique pronunciation norms and contexts in the ATC field, transfer learning and data augmentation techniques were applied to enhance the robustness of the network and improve the generalization ability of the model. The character error rate (CER) of our model was 11.1% on the expanded Aishell corpus, and it decreased to 8.0% on the ATC corpus.

Intelligent machinery health prognostics under variable operation conditions with limited and variable-length data

High accuracy health prognostics are significant to machinery intelligent operation and maintenance. Current data-driven prognostics achieve great success that benefits from amply learning samples. In fact, data scarcity challenge widely exists in machinery prognostics and health management, especially for high-end equipment. This study aims to solve this dilemma and proposes a novel meta learning algorithm reconstructed by classic variable-length prediction mode and attention mechanism, namely meta attention recurrent neural network (MARNN). Specifically, we first develop the encoder-decoder with attention mechanism (EDA) cell to perform episodic learning for the subtask-level upgrade. Then multiple subtasks with EDA as prediction models are aggregated to accomplish meta-level upgrade, thus mining the general degradation knowledge from historical datasets. Finally, cross-domain prognostics tasks can be easily realized through fine-tuning tricks, and three rotating machinery run-to-failed experiments are conducted to prove the generalizations of MARNN, which can obtain desired results even when the on-site adaptation data is reduced to one-twentieth.

Learning Parameterized Task Structure for Generalization to Unseen Entities

Real world tasks are hierarchical and compositional. Tasks can be composed of multiple subtasks (or sub-goals) that are dependent on each other. These subtasks are defined in terms of entities (e.g., "apple", "pear") that can be recombined to form new subtasks (e.g., "pickup apple", and "pickup pear"). To solve these tasks efficiently, an agent must infer subtask dependencies (e.g. an agent must execute "pickup apple" before "place apple in pot"), and generalize the inferred dependencies to new subtasks (e.g. "place apple in pot" is similar to "place apple in pan"). Moreover, an agent may also need to solve unseen tasks, which can involve unseen entities. To this end, we formulate parameterized subtask graph inference (PSGI), a method for modeling subtask dependencies using first-order logic with factored entities. To facilitate this, we learn parameter attributes in a zero-shot manner, which are used as quantifiers (e.g. is_pickable(X)) for the factored subtask graph. We show this approach accurately learns the latent structure on hierarchical and compositional tasks more efficiently than prior work, and show PSGI can generalize by modelling structure on subtasks unseen during adaptation.

Schedulability Performance Improvement via Task Split in Real-Time Systems

In this paper, we focus on splitting a task into multiple sub-tasks to improve schedulability performance for a set of tasks subject to timing constraints. Targeting FPS (Fixed Priority Scheduling) as a scheduling algorithm and RTA (Response Time Analysis) as a schedulability analysis on a multiprocessor platform, we develop a systematic method to utilize the task split, which addresses (i) how to apply the task split without violating the timing requirement of the original tasks and (ii) how to apply each task so as to make an unschedulable task set to be schedulable. Our simulation results demonstrate that RTA with the method finds 16.1%–19.4% (on average) additional task sets that were not proven schedulable by the vanilla RTA.