Showing papers on "Task analysis published in 2022"

OPAT: Optimized Allocation of Time-Dependent Tasks for Mobile Crowdsensing

Abstract: Mobile crowdsensing (MCS) is an emerging paradigm that leverages pervasive smart terminals equipped with various embedded sensors to collect sensory data for wide applications. As the sensing scale increases in MCS, the design of efficient task allocation becomes crucial. However, many prior task allocation schemes, which ignore the time for task-performing, are not applicable to the scenario where mobile users with limited time budgets are able to undertake multiple sensing tasks. In this article, we focus on the task allocation in time dependent crowdsensing systems and formulate the time dependent task allocation problem, in which both the sensing duration and the user's sensing capacity are considered. We prove that the task allocation problem is NP-hard and propose an efficient task allocation algorithm called optimized allocation scheme of time-dependent tasks (OPAT), which can maximize the sensing capacity of each mobile user. The extensive simulations are conducted to demonstrate the effectiveness of the proposed OPAT scheme.

Integrated Scheduling of Real-Time and Interactive Tasks for Configurable Industrial Systems

Abstract: With the recent advances in Internet of Things and cyber-physical systems technologies, smart industrial systems support configurable processes consisting of human interactions as well as hard real-time functions. This implies that irregularly arriving interactive tasks and traditional hard real-time tasks coexist. As the characteristics of the tasks are heterogeneous, it is not an easy matter to schedule them all at once. To cope with this situation, this article presents a new task scheduling policy that uses the notion of “virtual real-time task” and two-phase scheduling. As hard real-time tasks must keep their deadlines, we perform offline scheduling based on genetic algorithms beforehand. This determines the processor's voltage level and memory location of each task and also reserves the virtual real-time tasks for interactive tasks. When interactive tasks arrive during the execution, online scheduling is performed on the time slot of the virtual real-time tasks. As interactive workloads evolve over time, we monitor them and periodically update the offline scheduling. Experimental results show that the proposed policy reduces the energy consumption by 66.8% on average without deadline misses and also supports the waiting time of less than 3 (s) for interactive tasks.

OCR-RCNN: An Accurate and Efficient Framework for Elevator Button Recognition

Abstract: Autonomous elevator operation is considered a promising solution for mobile navigation in office buildings. As a fundamental function, elevator button recognition remains unsolved due to the challenging image conditions and severe data imbalance problem. In this article, we propose an accurate and efficient framework, named OCR-RCNN, for elevator button recognition. The framework is comprised of an region-based convolutional neural network (R-CNN)-based button detector and an attention-RNN-based character recognizer. Leveraging the two components, we further propose an end-to-end architecture and a cascaded architecture to explore the most effective network design for the framework. Moreover, a perspective distortion removal algorithm is also developed to enhance the inference performance of OCR-RCNN. Another key contribution of this work is that we release the first large-scale elevator panel dataset with 2005 images and 21 767 button labels. Extensive experiments are conducted on the released dataset and other two publicly available datasets. The proposed framework achieves an F1 score of 0.94, 1.00, and 1.00 in detection task, and an accuracy of 79.6% 96.5%, and 96.4% in character recognition task. The results demonstrate the advantages of our method, outperforming alternative strategies and other state-of-the-art methods in the literature. The data and code are available on the project webpage https://github.com/zhudelong/ocr-rcnn-v2 .

Task-Based Language Teaching

Abstract: This Element is a guide to task-based language teaching (TBLT), for language instructors, teacher educators, and other interested parties. The work first provides clear definitions and principles related to communication task design. It then explains how tasks can inform all stages of curriculum development. Diverse, localized cases demonstrate the scope of task-based approaches. Recent research illustrates the impact of task design (complexity, mode) and task implementation (preparation, interaction, repetition) on various second language outcomes. The Element also describes particular challenges and opportunities for teachers using tasks. The epilogue considers the potential of TBLT to transform classrooms, institutions, and society.

Quality-Driven Online Task-Bundling-Based Incentive Mechanism for Mobile Crowdsensing

Abstract: With the advancement of wireless communication technologies and rich embedded sensors in smart mobile devices, mobile crowdsensing has become an attractive paradigm for collecting sensing data from surrounding environments. Effective incentive mechanisms often play an important role in guaranteeing high quality of services and wide involvement of workers. Much work has been carried out in this aspect to encourage workers to strenuously and truthfully provide high quality data. However, task quality does not only depend on workers’ subjective effort or truthfulness, but also workers’ distribution and task quality growth law. In this paper, we consider a scenario where worker arrivals are unevenly distributed and the growth of task quality conforms to the law of diminishing margin of workers’ efforts. We propose a Quality-driven Online Task-Bundling-based incentive mechanism (QOTB). The design objective is to maximize the social welfare while maximally satisfying the task quality requirements. In QOTB, we introduce Mental accounting Theory to build accounts for task execution profit and bonus, respectively, which are then used to derive the participation willingness of workers. We adopt task bundling to stimulate workers to change their original travel schedules for balancing the task participations according to the popularities of task locations and also the traveling cost. We present the detailed mechanism design of QOTB. We prove that the QOTB mechanism has desired properties of willingness truthfulness, individual rationality, and computation efficiency. We conduct extensive simulations and the results show that QOTB can effectively improve the social welfare while satisfying the task quality requirements.

Spacing Effects in Task Repetition Research

Abstract: This article provides a conceptual review of the principles of input spacing as they might relate specifically to oral task repetition research and presents some of the common methodological considerations from the broader input spacing literature. The specific considerations discussed include the interaction between intersession intervals and retention intervals, the manipulation of posttests as a between-participants variable, the number of task repetitions, absolute versus relative spacing, the criterion of learning, task type versus exact task repetition, and blocked versus interleaved practice. Each of these considerations is discussed with links, as appropriate, to the relevant empirical input spacing and task repetition literature. The purpose of this review is to highlight how, in many cases, these methodological considerations have been overlooked by task repetition researchers, including in studies where input spacing has and has not been a direct focus, and to suggest ways of addressing these methodological shortcomings in future research.

Towards Transferring Human Preferences from Canonical to Actual Assembly Tasks

Abstract: To assist human users according to their individual preference in assembly tasks, robots typically require user demonstrations in the given task. However, providing demonstrations in actual assembly tasks can be tedious and time-consuming. Our thesis is that we can learn the preference of users in actual assembly tasks from their demonstrations in a representative canonical task. Inspired by prior work in economy of human movement, we propose to represent user preferences as a linear reward function over abstract task-agnostic features, such as movement and physical and mental effort required by the user. For each user, we learn the weights of the reward function from their demonstrations in a canonical task and use the learned weights to anticipate their actions in the actual assembly task; without any user demonstrations in the actual task. We evaluate our proposed method in a model-airplane assembly study and show that preferences can be effectively transferred from canonical to actual assembly tasks, enabling robots to anticipate user actions.

Task complexity, task repetition, and L2 writing complexity: exploring interactions in the TBLT domain

Abstract: Abstract Considering the increasing application of task-based frameworks to second language (L2) writing research, there has been a pressing need to examine TBLT views on the interactions between task conceptualization, task performance, and L2 writing outcomes. To address this need, the present study was designed to explore the synergistic effects of task complexity and task repetition on different facets of syntactic complexity as a key construct of proficiency and development in L2 writing. In doing so, 96 ESL learners performed written argumentative task versions with varying cognitive complexity in a counterbalanced fashion and then complete a task difficulty questionnaire. Afterward, they repeated the tasks in the opposite order at a one-week interval. The essays were analyzed using fine-grained syntactic complexity measures (reported by Lu’s L2 syntactic complexity analyzer, 2010). Linear mixed-effects models indicated significant main effects of task complexity and task repetition on different facets of syntactic complexity with robust effect sizes. However, no significant interaction effect between task complexity and task repetition was observed. These findings provide a reliable and accurate understanding of how syntactic complexity plays a role in the current accounts of connections between task features, implementation variables, and L2 writing task performance.

Online Stable Task Assignment in Opportunistic Mobile Crowdsensing With Uncertain Trajectories

Abstract: In opportunistic mobile crowdsensing, participants (workers) accept to carry out the requested sensing tasks only if they are already close to or within the regions of interest. Thus, the existence of an assignment opportunity between a worker-task pair strictly depends on whether or not the worker will visit the task region. However, when worker trajectories are uncertain and hence not known in advance, existing solutions fail to produce an effective task assignment. Besides, a satisfactory task assignment should respect the preferences and capacity constraints of workers and task requesters, which are generally neglected in the literature. In this study, we address all of these issues together and propose novel task assignment algorithms for different settings, which we prove to be optimal in terms of preference awareness (or stability). Extensive simulations performed on both synthetic and real data sets validate our theoretical results, and demonstrate that the proposed algorithms significantly outperform the existing solutions in terms of preference awareness and average quality of sensing attained in the final task assignment in almost all scenarios.

TIP: Task-Informed Motion Prediction for Intelligent Vehicles

Abstract: When predicting trajectories of road agents, motion predictors often approximate the future distribution by a limited number of samples. This constraint requires the predictors to generate samples that best support the task given task specifications. However, existing predictors are often optimized and evaluated via task-agnostic measures without accounting for the use of predictions in downstream tasks, and thus could result in sub-optimal task performance. In this paper, we propose a task-informed motion prediction model that better supports the tasks through its predictions by jointly reasoning about prediction accuracy and the utility of the downstream tasks during training. The task utility function is commonly used to evaluate task performance. It does not require the full task information, but rather a specification of the utility of the task, resulting in predictors that are tailored to different downstream tasks. We demonstrate our approach on two use cases of common decision making tasks and their utility functions, in the context of autonomous driving and parallel autonomy. Experiment results show that our predictor produces accurate predictions that improve the task performance by a large margin in both tasks when compared to task-agnostic baselines on the Waymo Open Motion dataset.

Elephants Don’t Pack Groceries: Robot Task Planning for Low Entropy Belief States

Abstract: Recent advances in computational perception have significantly improved the ability of autonomous robots to perform state estimation with low entropy. Such advances motivate a reconsideration of robot decision-making under uncertainty. Current approaches to solving sequential decision-making problems model states as inhabiting the extremes of the perceptual entropy spectrum. As such, these methods are either incapable of overcoming perceptual errors or asymptotically inefficient in solving problems with low perceptual entropy. With low entropy perception in mind, we aim to explore a happier medium that balances computational efficiency with the forms of uncertainty we now observe from modern robot perception. We propose an approach for efficient task planning for goal-directed robot reasoning. Our approach combines belief space representation with the fast, goal-directed features of classical planning to efficiently plan for low entropy goal-directed reasoning tasks. We compare our approach with current classical planning and belief space planning approaches by solving low entropy goal-directed grocery packing tasks in simulation. Our approach outperforms these approaches in planning time, execution time, and task success rate in our simulation experiments. We also demonstrate our approach on a real world grocery packing task with physical robot. A video summary of this letter can be found at this url: https://youtu.be/im6tve9-9A0

Investigating the impact of task complexity on uptake and noticing of corrective feedback recasts

Abstract: Abstract This study investigated the relationship between task complexity, second language (L2) learners’ response and awareness of corrective feedback provided in the form of recasts during teacher–student interaction. Drawing on Robinson’s Triadic Componential Framework, the study examined how degrees of task complexity created by two specific task characteristics had an impact on learners’ responses (referred to as uptake), and their reported noticing of grammatical and lexical recasts. Data documenting learners’ uptake, operationalized as changes in response to feedback during interactions and noticing of recasts, as indicated in students’ self reports of detection and attention to recasts, were collected during one-on-one interaction sessions and stimulated recall sessions with ESL learners in Canada. Frequency analysis and Cochran’s Q analysis with multiple McNemar post hoc tests were carried out to compare the uptake and noticing of recasts across different tasks. The results revealed that tasks with different degrees of complexity impacted uptake and noticing of recasts differently. The results also showed that linguistic target, i.e., lexical or grammatical features, modulated the relationship between task complexity and recast uptake and noticing. The study calls for a more nuanced approach to investigating task complexity in research, and for practitioners to consider task complexity in decision making related to the use of corrective feedback and the design of classroom-based tasks.

Task Publication Time Recommendation in Spatial Crowdsourcing

Abstract: The increasing proliferation of networked and geo-positioned mobile devices brings about increased opportunities for Spatial Crowdsourcing (SC), which aims to enable effective location-based task assignment. We propose and study a novel SC framework, namely Task Assignment with Task Publication Time Recommendation. The framework consists of two phases, task publication time recommendation and task assignment. More specifically, the task publication time recommendation phase hybrids different learning models to recommend the suitable publication time for each task to ensure the timely task assignment and completion while reducing the waiting time of the task requester at the SC platform. We use a cross-graph neural network to learn the representations of task requesters by integrating the obtained representations from two semantic spaces and utilize the self-attention mechanism to learn the representations of task-publishing sequences from multiple perspectives. Then a fully connected layer is used to predict suitable task publication time based on the obtained representations. In the task assignment phase, we propose a greedy and a minimum cost maximum flow algorithm to achieve the efficient and the optimal task assignment, respectively. An extensive empirical study demonstrates the effectiveness and efficiency of our framework.

Research on Task Reassignment Method of Heterogeneous UAV in Dynamic Environment

Abstract: When the traditional CBBA algorithm solves the task allocation problem, it has a large amount of calculation and does not have the ability of dynamic task allocation. In order to solve the above problems, this paper improves the traditional CBBA algorithm from three aspects: introducing the concept of time window, proposing a target redistribution model, and improving the target function. Combined with the task reassignment problem of multiple UAVs in dynamic scenarios, the simulation verification of task reassignment is carried out in this paper. The simulation results show that the CBBA algorithm proposed in this paper can solve the problem of UAV task reassignment in a dynamic environment.

Joint Task Offloading and Resource Allocation for Cooperative Mobile-Edge Computing Under Sequential Task Dependency

Abstract: The emergence of mobile-edge computing (MEC) makes it possible to run intelligent applications on Internet of Things (IoT) devices. However, due to blockage or deep fading, one IoT device may not have direct link with the edge server. In this case, many surrounding wireless devices can serve as a cooperative node. In this article, we study a cooperative MEC system running sequential task, which is composed of a series of subtasks and can support many intelligent applications. To minimize the energy consumption of the IoT device and cooperative node, a task offloading policy together with the allocation of communication and computation resources is designed jointly. The cases when the cooperative node has no/has private task to complete are investigated, which are denoted as cases I and II, respectively. Although both cases involve the optimization of integer variables, their optimal solutions are achieved. For the first case, the associated problem is simplified equivalently and then decomposed into two levels, with the upper level dealing with integer variables and the lower level handling continuous variables. Bisection search is employed to reach optimality in the lower level and the searching space is compressed in the upper level. For the second case, the associated problem is subdivided into three subproblems. To solve every subproblem optimally, a similar operation like case I is followed, with a semiclosed form solution derived in the lower level. Numerical results verify the effectiveness of our proposed methods compared with benchmark methods and our effort on reducing computation complexity.

In-BoXBART: Get Instructions into Biomedical Multi-Task Learning

Abstract: Single-task models have proven pivotal in solving specific tasks; however, they have limitations in real-world applications where multi-tasking is necessary and domain shifts are exhibited. Recently, instructional prompts have shown significant improvement towards multi-task generalization; however, the effect of instructional prompts and Multi-Task Learning (MTL) has not been systematically studied in the biomedical domain. Motivated by this, this paper explores the impact of instructional prompts for biomedical MTL. We introduce the BoX, a collection of 32 instruction tasks for Biomedical NLP across (X) various categories. Using this meta-dataset, we propose a unified model termed as In-BoXBART, that can jointly learn all tasks of the BoX without any task-specific modules. To the best of our knowledge, this is the first attempt to propose a unified model in the biomedical domain and use instructions to achieve generalization across several biomedical tasks. Experimental results indicate that the proposed model: 1) outperforms single-task baseline by ~3% and multi-task (without instruction) baseline by ~18% on an average, and 2) shows ~23% improvement compared to single-task baseline in few-shot learning (i.e., 32 instances per task) on an average. Our analysis indicates that there is significant room for improvement across tasks in the BoX, implying the scope for future research direction.

An Online Learning Approach towards Far-sighted Emergency Relief Planning under Intentional Attacks in Conflict Areas

Abstract: A large number of emergency humanitarian rescue demands in conflict areas around the world are accompanied by intentional, persistent and unpredictable attacks on rescuers and supplies. Unfortunately, existing work on humanitarian relief planning mostly ignores this challenge in reality resulting a parlous and short-sighted relief distribution plan to a large extent. To address this, we first propose an offline multi-stage optimization problem of emergency relief planning under intentional attacks, in which all parameters in the game between the rescuer and attacker are supposed to be known or predictable. Then, an online version of this problem is introduced to meet the need of online and irrevocable decision making when those parameters are revealed in an online fashion. To achieve a far-sighted emergency relief planning under attacks, we design an online learning approach which is proven to obtain a near-optimal solution of the offline problem when those online reveled parameters are i.i.d. sampled from an unknown distribution. Finally, extensive experiments on a real anti-Ebola relief planning case based on the data of Ebola outbreak and armed attacks in DRC Congo show the scalability and effectiveness of our approach.

DeTTO: Dependency-Aware Trustworthy Task Offloading in Vehicular IoT

Abstract: 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.

A Self-training Multi-task Attention Method for NILM

Abstract: The key task of non-invasive load monitoring (NILM) is to know the power consumption of all household appliances, from which the power consumption of individual household appliance can be disaggregated. The power consumption and on/off state of household appliances are expected to be obtained, the multi-task learning model is used. One task is used to train the power comsumpution of household appliances, the other task is used to train the on/off state of household appliances, and then these two results are combined as the final result. In this paper, a self-training multi-task learning model is proposed. In the model, a parallel structure is used to deal with two different tasks, and the outputs of two branches are directly combined as the final output. The model only needs one loss function and is only trained once. In addition, we also introduce attention mechanism into the proposed model. Finally, two public data sets are simulated to verify the effectiveness and superiority of the proposed method.

Vision-Language Pre-Training for Multimodal Aspect-Based Sentiment Analysis

Abstract: As an important task in sentiment analysis, Multimodal Aspect-Based Sentiment Analysis (MABSA) has attracted increasing attention inrecent years. However, previous approaches either (i) use separately pre-trained visual and textual models, which ignore the crossmodalalignment or (ii) use vision-language models pre-trained with general pre-training tasks, which are inadequate to identify fine-grainedaspects, opinions, and their alignments across modalities. To tackle these limitations, we propose a task-specific Vision-LanguagePre-training framework for MABSA (VLP-MABSA), which is a unified multimodal encoder-decoder architecture for all the pretrainingand downstream tasks. We further design three types of task-specific pre-training tasks from the language, vision, and multimodalmodalities, respectively. Experimental results show that our approach generally outperforms the state-of-the-art approaches on three MABSA subtasks. Further analysis demonstrates the effectiveness of each pre-training task. The source code is publicly released at https://github.com/NUSTM/VLP-MABSA.

Attentive Task Interaction Network for Multi-Task Learning

Abstract: Multitask learning (MTL) has recently gained a lot of popularity as a learning paradigm that can lead to improved per-task performance while also using fewer per-task model parameters compared to single task learning. One of the biggest challenges regarding MTL networks involves how to share features across tasks. To address this challenge, we propose the Attentive Task Interaction Network (ATI-Net). ATI-Net employs knowledge distillation of the latent features for each task, then combines the feature maps to provide improved contextualized information to the decoder. This novel approach to introducing knowledge distillation into an attention based multitask network outperforms state of the art MTL baselines such as the standalone MTAN and PAD-Net, with roughly the same number of model parameters.

A Tale of Two Tasks: Automated Issue Priority Prediction with Deep Multi-task Learning

Abstract: Background. Issues are prevalent, and identifying the correct priority of the reported issues is crucial to reduce the maintenance effort and ensure higher software quality. There are several approaches for the automatic priority prediction, yet they do not fully utilize the related information that might influence the priority assignment. Our observation reveals that there are noticeable correlations between an issue’s priority and its category, e.g., an issue of bug category tends to be assigned with higher priority than an issue of document category. This correlation motivates us to employ multi-task learning to share the knowledge about issue’s category prediction and facilitating priority prediction. Aims. This paper aims at providing an automatic approach for effective issue’s priority prediction, to reduce the burden of the project members and better manage the issues. Method. We propose issue priority prediction approach PRIMA with deep multi-task learning, which takes the issue category prediction as another task to facilitate the information sharing and learning. It consists of three main phases: 1) data preparation and augmentation phase, which allows data sharing beyond single task learning; 2) model construction phase, which designs shared layers to encode the semantics of textual descriptions, and task-specific layers to model two tasks in parallel; it also includes the indicative attributes to better capture an issue’s inherent meaning; 3) model training phase, which enables eavesdropping by shared loss function between two tasks. Results. Evaluations with four large-scale open-source projects show that PRIMA outperforms commonly-used and state-of-the-art baselines, with 32% -55% higher precision, and 28% - 56% higher recall. Compared with single task learning, the performance improvement reaches 18% in precision and 19% in recall. Results from our user study further prove its potential practical value. Conclusions. The proposed approach provides a novel and effective way for issue priority prediction, and sheds light on jointly exploring other issue-management tasks.

Design requirements to improve laparoscopy via XR

Abstract: Laparoscopic surgery has the advantage of avoiding large open in-cisions and thereby decreasing blood loss, pain, and discomfort to patients. However, on the other side, it is hampered by restricted workspace, ambiguous communication, and surgeon fatigue caused by non-ergonomic head positioning. We aimed to identify critical problems and suggest design requirements and solutions. We used user and task analysis methods to learn about practices performed in an operating room by observing surgeons in their working environment to understand how they performed tasks and achieved their intended goals. Drawing on observations and analysis from recorded laparoscopic surgeries, we have identified several constraints and design requirements to propose potential solutions to address the issues. Surgeons operate in a dimly lit environment, surrounded by monitors, and communicate through verbal commands and pointing gestures. Therefore, performing user and task analysis allowed us to understand the existing problems in laparoscopy better while identifying several communication constraints and design requirements, which a solution has to follow to address those problems. Our contributions include identifying design requirements for laparoscopy surgery through a user and task analysis. These requirements propose design solutions towards improved surgeons' comfort and make the surgical procedure less laborious.

Dynamic Human-Robot Role Allocation based on Human Ergonomics Risk Prediction and Robot Actions Adaptation

Abstract: Even though cobots have high potential in bringing several benefits in the manufacturing and logistic processes, their rapid (re-)deployment in changing environments is still limited. To enable fast adaptation to new product demands and to boost the fitness of the human workers to the allocated tasks, we propose a novel method that optimizes assembly strategies and distributes the effort among the workers in human-robot cooperative tasks. The cooperation model exploits AND/OR Graphs that we adapted to solve also the role allocation problem. The allocation algorithm considers quantitative measurements that are computed online to describe human operators' ergonomic status and task properties. We conducted preliminary experiments to demonstrate that the proposed approach succeeds in controlling the task allocation process to ensure safe and ergonomic conditions for the human worker.

Lessons Learned and Future Direction of a Teaming Interface for Unmanned Vehicle Tasks

Abstract: Having multiple operators requires aspects critical to teaming, such as coordination and team awareness, to be considered during system design. A Task Manager interface was developed that supports shared awareness across team members by summarizing the relative priority, recency, assignment, and completion status of mission tasks. While the original design provided information essential to the operator, evaluation results indicated that critical information needed to be more accessible. Primarily, important unmanned vehicle (UV) task details should be available at the higher level without the need to “drill down” into the task. Evaluation results informed a Task Manager redesign that does not remove any functionality but altered how information is represented. The goal of these modifications is to improve awareness for the UV operators and support more efficient teaming between operator/autonomy teammates. This new design will be evaluated in future research, and those results will then inform future designs using an iterative design and evaluation process.

EGCN: An Ensemble-based Learning Framework for Exploring Effective Skeleton-based Rehabilitation Exercise Assessment

Abstract: Recently, some skeleton-based physical therapy systems have been attempted to automatically evaluate the correctness or quality of an exercise performed by rehabilitation subjects. However, in terms of algorithms and evaluation criteria, the task remains not fully explored regarding making full use of different skeleton features. To advance the prior work, we propose a learning framework called Ensemble-based Graph Convolutional Network (EGCN) for skeleton-based rehabilitation exercise assessment. As far as we know, this is the first attempt that utilizes both two skeleton feature groups and investigates different ensemble strategies for the task. We also examine the properness of existing evaluation criteria and focus on evaluating the prediction ability of our proposed method. We then conduct extensive cross-validation experiments on two latest public datasets: UI-PRMD and KIMORE. Results indicate that the model-level ensemble scheme of our EGCN achieves better performance than existing methods. Code is available: https://github.com/bruceyo/EGCN.

Task Detection in Continual Learning via Familiarity Autoencoders

Abstract: Continual learning requires the ability to reliably transfer previously learned knowledge to new tasks without disrupting established competencies. Methods such as Progressive Neural Network [1] accomplish high-quality transfer learning while nullifying the insidious problem of catastrophic forgetting. However, most module-based continual learning systems require task labels during operation – a constraint that limits their application in many real-world conditions where task indicators are opaque. This paper proposes a task detector neural algorithm to acquire task information while maintaining immunity to forgetting. Our proposed task detector allows progressive neural networks (and many similar systems) to operate without task labels during test time. Our task detector is built from familiarity autoencoders which recognize the nature of the required task from input data. We demonstrate the generality and effectiveness of our approach through experiments in video game playing and automated image repair. Our results show near-perfect task recognition in all domains (>.99 F1), rewards above published single-task scores in MinAtar, and realistic image repairs on damaged human face pictures. The performance of our integrated method is nearly identical to the progressive systems equipped with ground-truth task labels ¹

Abstraction in Data-Sparse Task Transfer (Extended Abstract)

Abstract: When a robot adapts a learned task for a novel environment, any changes to objects in the novel environment have an unknown effect on its task execution. For example, replacing an object in a pick-and-place task affects where the robot should target its actions, but does not necessarily affect the underlying action model. In contrast, replacing a tool that the robot will use to complete a task will effectively alter its end-effector pose with respect to the robot's base coordinate system, and thus the robot's motion must be replanned accordingly. These examples highlight the relationship among (i) differences between the source and target environments, (ii) the level of abstraction at which a robot's task model should be represented to enable transfer to the target environment, and (iii) the information needed to ground the abstracted task representation in the target environment. In this abstract, summarizing our full article [Fitzgerald et al., 2021], we present our taxonomy of transfer problems based on this relationship. We also describe a knowledge representation called the Tiered Task Abstraction (TTA) and demonstrate its applicability to a variety of transfer problems in the taxonomy. Our experimental results indicate a trade-off between the generality and data requirements of a task representation, and reinforce the need for multiple transfer methods that operate at different levels of abstraction.