Task Representation Research Articles

Learning Domain-Independent Heuristics for Grounded and Lifted Planning

We present three novel graph representations of planning tasks suitable for learning domain-independent heuristics using Graph Neural Networks (GNNs) to guide search. In particular, to mitigate the issues caused by large grounded GNNs we present the first method for learning domain-independent heuristics with only the lifted representation of a planning task. We also provide a theoretical analysis of the expressiveness of our models, showing that some are more powerful than STRIPS-HGN, the only other existing model for learning domain-independent heuristics. Our experiments show that our heuristics generalise to much larger problems than those in the training set, vastly surpassing STRIPS-HGN heuristics.

LAMM: Label Alignment for Multi-Modal Prompt Learning

With the success of pre-trained visual-language (VL) models such as CLIP in visual representation tasks, transferring pre-trained models to downstream tasks has become a crucial paradigm. Recently, the prompt tuning paradigm, which draws inspiration from natural language processing (NLP), has made significant progress in VL field. However, preceding methods mainly focus on constructing prompt templates for text and visual inputs, neglecting the gap in class label representations between the VL models and downstream tasks. To address this challenge, we introduce an innovative label alignment method named \textbf{LAMM}, which can dynamically adjust the category embeddings of downstream datasets through end-to-end training. Moreover, to achieve a more appropriate label distribution, we propose a hierarchical loss, encompassing the alignment of the parameter space, feature space, and logits space. We conduct experiments on 11 downstream vision datasets and demonstrate that our method significantly improves the performance of existing multi-modal prompt learning models in few-shot scenarios, exhibiting an average accuracy improvement of 2.31(\%) compared to the state-of-the-art methods on 16 shots. Moreover, our methodology exhibits the preeminence in continual learning compared to other prompt tuning methods. Importantly, our method is synergistic with existing prompt tuning methods and can boost the performance on top of them. Our code and dataset will be publicly available at https://github.com/gaojingsheng/LAMM.

Multi-tailed vision transformer for efficient inference

Recently, Vision Transformer (ViT) has achieved promising performance in image recognition and gradually serves as a powerful backbone in various vision tasks. To satisfy the sequential input of Transformer, the tail of ViT first splits each image into a sequence of visual tokens with a fixed length. Then, the following self-attention layers construct the global relationship between tokens to produce useful representation for the downstream tasks. Empirically, representing the image with more tokens leads to better performance, yet the quadratic computational complexity of self-attention layer to the number of tokens could seriously influence the efficiency of ViT’s inference. For computational reduction, a few pruning methods progressively prune uninformative tokens in the Transformer encoder, while leaving the number of tokens before the Transformer untouched. In fact, fewer tokens as the input for the Transformer encoder can directly reduce the following computational cost. In this spirit, we propose a Multi-Tailed Vision Transformer (MT-ViT) in the paper. MT-ViT adopts multiple tails to produce visual sequences of different lengths for the following Transformer encoder. A tail predictor is introduced to decide which tail is the most efficient for the image to produce accurate prediction. Both modules are optimized in an end-to-end fashion, with the Gumbel-Softmax trick. Experiments on ImageNet-1K demonstrate that MT-ViT can achieve a significant reduction on FLOPs with no degradation of the accuracy and outperform compared methods in both accuracy and FLOPs.

Bidirectional generation of structure and properties through a single molecular foundation model

Recent successes of foundation models in artificial intelligence have prompted the emergence of large-scale chemical pre-trained models. Despite the growing interest in large molecular pre-trained models that provide informative representations for downstream tasks, attempts for multimodal pre-training approaches on the molecule domain were limited. To address this, here we present a multimodal molecular pre-trained model that incorporates the modalities of structure and biochemical properties, drawing inspiration from recent advances in multimodal learning techniques. Our proposed model pipeline of data handling and training objectives aligns the structure/property features in a common embedding space, which enables the model to regard bidirectional information between the molecules’ structure and properties. These contributions emerge synergistic knowledge, allowing us to tackle both multimodal and unimodal downstream tasks through a single model. Through extensive experiments, we demonstrate that our model has the capabilities to solve various meaningful chemical challenges, including conditional molecule generation, property prediction, molecule classification, and reaction prediction.

Interactive simulations as a tool for logistics and maintenance of IFMIF-DONES

Pre-configured virtual reality (VR) simulations of the logistics and maintenance processes have proven to be useful for identifying potential design issues as well as planning operations during an early design phase of facility. But VR simulations can also be used to deeply explore the feasibility of these procedures in a more interactive manner, so that we can identify potential risks and difficulty levels from early stages and study different maintenance strategies to assist the maintenance worker during these procedures. This article presents a framework to design and validate logistics and maintenance procedures in complex facilities, such as the International Fusion Materials Irradiation Facility DEMO Oriented Neutron Source (IFMIF-DONES). Our framework begins with a preparatory phase where essential information about the procedures and Computer-Aided Design (CAD) models is compiled into a comprehensive Virtualization Task Document (VTD). Differently from previous work, this VTD allows representation of parallel tasks. We implement the interactive version of the virtual environment, where the different maintenance and logistics equipment, as well as a virtual maintenance worker is controlled by the user (the person executing the interactive simulation). We have validated this interactive framework with two simulations for the installation process of the Superconducting Radio Frequency Linear accelerator (SRF Linac) modules in IFMIF-DONES. In one simulation (the automatic one), the procedures are reproduced as they are planned, while in the second simulation (the interactive one) the user freely controls the movements of the moving parts of the crane, grab and release plant equipment, move platforms, etc. Based on our simulations, the interactive version allows easier detection of potential points of collisions as well as more precise assessment of the difficulty of the tasks to be performed.

Be prepared for interruptions: EEG correlates of anticipation when dealing with task interruptions and the role of aging

Dealing with task interruptions requires the flexible use of working memory and attentional control mechanisms, which are prone to age-related changes. We investigated effects of age on dealing with task interruptions and potential advantages of anticipating an interruption using EEG and a retrospective cueing (retro-cue) paradigm. Thirty-two young (18–30 years) and 28 older (55–70 years) participants performed a visual working memory task, where they had to report the orientation of a target following a retro-cue. Within blocks of 10 trials, they were always, never, or randomly interrupted with an arithmetic task before the onset of the retro-cue. The interruption-induced decline in primary task performance was more pronounced in older participants, while only these benefited from anticipation. The EEG analysis revealed reduced theta and alpha/beta response to the retro-cue following interruptions, especially for the older participants. In both groups, anticipated interruptions were associated with increased theta and alpha/beta power prior and during the interruption, and stronger beta suppression to the retro-cue. The results indicate that interruptions impede the refocusing of attention on the task-relevant representation of the primary task, especially in older people, while anticipation facilitates preparation for the interruption task and resumption of the primary task.

Latent evolutionary signatures: a general framework for analysing music and cultural evolution.

Cultural processes of change bear many resemblances to biological evolution. The underlying units of non-biological evolution have, however, remained elusive, especially in the domain of music. Here, we introduce a general framework to jointly identify underlying units and their associated evolutionary processes. We model musical styles and principles of organization in dimensions such as harmony and form as following an evolutionary process. Furthermore, we propose that such processes can be identified by extracting latent evolutionary signatures from musical corpora, analogously to identifying mutational signatures in genomics. These signatures provide a latent embedding for each song or musical piece. We develop a deep generative architecture for our model, which can be viewed as a type of variational autoencoder with an evolutionary prior constraining the latent space; specifically, the embeddings for each song are tied together via an energy-based prior, which encourages songs close in evolutionary space to share similar representations. As illustration, we analyse songs from the McGill Billboard dataset. We find frequent chord transitions and formal repetition schemes and identify latent evolutionary signatures related to these features. Finally, we show that the latent evolutionary representations learned by our model outperform non-evolutionary representations in such tasks as period and genre prediction.

Deeply Unsupervised Patch Re-Identification for Pre-Training Object Detectors.

Unsupervised pre-training aims at learning transferable features that are beneficial for downstream tasks. However, most state-of-the-art unsupervised methods concentrate on learning global representations for image-level classification tasks instead of discriminative local region representations, which limits their transferability to region-level downstream tasks, such as object detection. To improve the transferability of pre-trained features to object detection, we present Deeply Unsupervised Patch Re-ID (DUPR), a simple yet effective method for unsupervised visual representation learning. The patch Re-ID task treats individual patch as a pseudo-identity and contrastively learns its correspondence in two views, enabling us to obtain discriminative local features for object detection. Then the proposed patch Re-ID is performed in a deeply unsupervised manner, appealing to object detection, which usually requires multi-level feature maps. Extensive experiments demonstrate that DUPR outperforms state-of-the-art unsupervised pre-trainings and even the ImageNet supervised pre-training on various downstream tasks related to object detection.

Multi-Task Learning with Sequential Dependence Toward Industrial Applications: A Systematic Formulation

Multi-task learning (MTL) is widely used in the online recommendation and financial services for multi-step conversion estimation, but current works often overlook the sequential dependence among tasks. In particular, sequential dependence multi-task learning (SDMTL) faces challenges in dealing with complex task correlations and extracting valuable information in real-world scenarios, leading to negative transfer and a deterioration in the performance. Herein, a systematic learning paradigm of the SDMTL problem is established for the first time, which applies to more general multi-step conversion scenarios with longer conversion paths or various task dependence relationships. Meanwhile, an SDMTL architecture, named Task-Aware Feature Extraction (TAFE), is designed to enable the dynamic task representation learning from a sample-wise view. TAFE selectively reconstructs the implicit shared information corresponding to each sample case and performs the explicit task-specific extraction under dependence constraints, which can avoid the negative transfer, resulting in more effective information sharing and joint representation learning. Extensive experiment results demonstrate the effectiveness and applicability of the proposed theoretical and implementation frameworks. Furthermore, the online evaluations at MYbank showed that TAFE had an average increase of 9.22% and 3.76% in various scenarios on the post-view click-through & conversion rate (CTCVR) estimation task. Currently, TAFE is deployed in an online platform to provide various traffic services.

Attribute Graph Embedding Based on Multi-Order Adjacency Views and Attention Mechanisms

Graph embedding plays an important role in the analysis and study of typical non-Euclidean data, such as graphs. Graph embedding aims to transform complex graph structures into vector representations for further machine learning or data mining tasks. It helps capture relationships and similarities between nodes, providing better representations for various tasks on graphs. Different orders of neighbors have different impacts on the generation of node embedding vectors. Therefore, this paper proposes a multi-order adjacency view encoder to fuse the feature information of neighbors at different orders. We generate different node views for different orders of neighbor information, consider different orders of neighbor information through different views, and then use attention mechanisms to integrate node embeddings from different views. Finally, we evaluate the effectiveness of our model through downstream tasks on the graph. Experimental results demonstrate that our model achieves improvements in attributed graph clustering and link prediction tasks compared to existing methods, indicating that the generated embedding representations have higher expressiveness.

Multimodal meta-learning through meta-learned task representations

Few-shot meta-learning involves training a model on multiple tasks to enable it to efficiently adapt to new, previously unseen tasks with only a limited number of samples. However, current meta-learning methods assume that all tasks are closely related and belong to a common domain, whereas in practice, tasks can be highly diverse and originate from multiple domains, resulting in a multimodal task distribution. This poses a challenge for existing methods as they struggle to learn a shared representation that can be easily adapted to all tasks within the distribution. To address this challenge, we propose a meta-learning framework that can handle multimodal task distributions by conditioning the model on the current task, resulting in a faster adaptation. Our proposed method learns to encode each task and generate task embeddings that modulate the model’s activations. The resulting modulated model become specialized for the current task and leads to more effective adaptation. Our framework is designed to work in a realistic setting where the mode from which a task is sampled is unknown. Nonetheless, we also explore the possibility of incorporating auxiliary information, such as the task-mode-label, to further enhance the performance of our method if such information is available. We evaluate our proposed framework on various few-shot regression and image classification tasks, demonstrating its superiority over other state-of-the-art meta-learning methods. The results highlight the benefits of learning to embed task-specific information in the model to guide the adaptation when tasks are sampled from a multimodal distribution.

Distance profile layer for binary classification and density estimation

The research on methods trying to explain solutions based on Neural Networks (NN) is a vivid target of Machine Learning works in recent years. Such a need arose due to the black-box nature of neural models and their tendency to provide a high certainty for incorrect decisions regarding events outside the area covered by the training set. In this work, we present a novel layer building a Distance Profile of recognized samples, aiming to substitute one-hot encoding of labels. The presented approach shows promising results in classification and unsupervised distribution density estimation and maintains a consistent representation for both recognition tasks. This approach manages certainty quantification for objects outside of the scope represented by available data and could potentially contribute to the Open Set Recognition and Out of Distribution Detection research fields.

Application of Pythagorean Fuzzy Analytic Hierarchy Process for Assessing Driver Behavior Criteria Associated to Road Safety

The investigation of road safety issues has long posed a formidable challenge because of the intricate and unpredictable practice of human behavior. To address this complexity, experts in the field have turned to linguistic terms for evaluation, capitalizing on recent advancements in ordinary fuzzy sets. A promising way in Multi-criteria decision-making (MCDM) is the utilization of Pythagorean fuzzy sets (PFSs), which provide an extra flexible representation of membership tasks. This study introduces an innovative approach, the Pythagorean Fuzzy Analytic Hierarchy Process (PF-AHP), to measure and rank essential driver behavior criteria in a hierarchical model tailored for diverse driver groups in Budapest city. Our method effectively ranks the model criteria and sub-criteria based on their weighted scores. Consequently, we determine that criteria 'lapses' and 'errors' are the most pivotal factors based on the aggregated weights as compared to all other considerations. In contrast, the criterion 'disobeying speed limits' emerges as the least critical one, followed by 'disobeying overtaking rules' as the second least criterion. Our research highlights that the proposed approach yields robust and useful outcomes, well accommodating the inherent ambiguity in decision-making processes. The resilience of our findings is further affirmed through one-at-a-time sensitivity analysis.

Effects of contextual interference and differential learning on performance and mental representations in a golf putting task

AbstractIt is widely accepted that mental representations can have an important influence on motor performance. Although differences in mental representations of motor tasks have been reported between novices and experts, little is known about their development as a function of motor learning approaches. The aim of this study was to compare the effects of contextual interference (CI) and differential learning (DL) on the performance and mental representations in a golf putting task. A total of 40 participants were randomly assigned into four groups: blocked contextual interference (BCI), random contextual interference (RCI), DL, and control. First, the participant's initial mental representation level was tested by means of the structural dimensional analysis of mental representation. Then, the participant's initial performance level was tested by 12 golf‐putting trials from 2.44 m. During the acquisition phase, participants practiced golf putting according to their grouping for three consecutive days with 10 blocks of 12 trials per day. No intervention was applied for the control group. The retention‐tests were performed 72 h after the last acquisition day. In addition, a transfer test to a novel distance outside the acquired range (4 m) was performed immediately after the retention‐test. The results of the putting performance in the retention test showed that RCI and DL performed better compared to BCI and the control group (all p < 0.05). In the transfer test, BCI and RCI outperformed the control group (all p < 0.05), but both were further outperformed by the DL group (all p < 0.05). Moreover, the DL group showed a more structured mental representation than the other groups during the retention test. These results indicated that DL used a different underlying mechanism that resulted in different levels of performance during transfer and a more structured mental representation compared with CI.

Semi-supervised multi-source transfer learning for cross-subject EEG motor imagery classification.

Electroencephalogram (EEG) motor imagery (MI) classification refers to the use of EEG signals to identify and classify subjects' motor imagery activities; this task has received increasing attention with the development of brain-computer interfaces (BCIs). However, the collection of EEG data is usually time-consuming and labor-intensive, which makes it difficult to obtain sufficient labeled data from the new subject to train a new model. Moreover, the EEG signals of different individuals exhibit significant differences, leading to a significant drop in the performance of a model trained on the existing subjects when directly classifying EEG signals acquired from new subjects. Therefore, it is crucial to make full use of the EEG data of the existing subjects and the unlabeled EEG data of the new target subject to improve the MI classification performance achieved for the target subject. This research study proposes a semi-supervised multi-source transfer (SSMT) learning model to address the above problems; the model learns informative and domain-invariant representations to address cross-subject MI-EEG classification tasks. In particular, a dynamic transferred weighting schema is presented to obtain the final predictions by integrating the weighted features derived from multi-source domains. The average accuracies achieved on two publicly available EEG datasets reach 83.57 and 85.09 , respectively, validating the effectiveness of the SSMT process. The SSMT process reveals the importance of informative and domain-invariant representations in MI classification tasks, as they make full use of the domain-invariant information acquired from each subject.

HeGAE-AC: Heterogeneous graph auto-encoder for attribute completion

Heterogeneous Graphs (HGs) are widely used to model complex real-world networks that involve multiple types of nodes and relations. Many heterogeneous graph neural networks (HGNNs) based methods have been developed to learn low-dimensional representations for nodes in HGs with the assumption of complete attributes for all nodes. However, in many real applications, the attributes of certain nodes may be completely missing, leading to attribute-missing issues. Existing methods for attribute completion in HGs separate topological embedding learning from attribute completion, which can result in sub-optimal solutions. Moreover, the information of higher-order connected nodes is not fully exploited in the process of attribute completion. To address the issues mentioned above, we integrate topological embedding learning, attribute completion, and heterogeneous graph representation learning into an end-to-end framework called HeGAE-AC. Firstly, we utilize a heterogeneous graph auto-encoder (HeGAE) to incorporate the graph topology structure and attribute information of higher-order connected nodes into low-dimensional embeddings. Then, HeGAE decodes embeddings to generate attributes of all nodes. Finally, an HGNN model is employed to obtain node representations for downstream tasks using the completed HGs. We conduct extensive experiments on four real-world networks, which demonstrate that our approach achieves significant improvements over state-of-the-art methods.

Supporting AI Planning and Collaboration for Robotic Applications Using TAEMS

This paper describes a general approach to integrating higher-level reasoning mechanisms including planning and scheduling methods with lower-level robotic control processes. We adopt a domain-independent task representation language TAEMS to describe the knowledge of tasks, resources, and their interrelationships. This TAEMS representation language serves as the input of the reasoning functions, which generate a schedule of executable methods to be executed by the robot in the physical world. In the execution process of this goal-directed plan, the robot also needs to attend to basic functions. The potential interactions between the plan and these basic functions would lead to interesting challenges that will be discussed. An integrated development platform with a simulator that supports real-world physics is also presented.

Artificial intelligence as enabler of future library services: how prepared are librarians in African university libraries

PurposeThe purpose of this study is to investigate artificial intelligence (AI) as enabler of future library services, with consideration to how prepared are librarians in African university libraries.Design/methodology/approachThis study applied the interpretive content/document analysis of literature harvested from different databases of Scopus and Web of Science. AI could be used to perform daily routines in circulation, serial, reference and selective dissemination of information among others. It could also be applied to the provision of innovative services of recognition of library activities such as answering research quarries, cataloguing and classification of library materials and management of library system software of different databases within the library systems.FindingsIt could be deduced from the study that AI would continue to serve as a panacea to future library services irrespective of its geographical context. Due to the evolving nature of knowledge growth, AI having its roots in the field of engineering has been found useful to support future library services. The support accrued from library service delivery in the library profession has made librarians continue to interact with other intelligent machines that can demonstrate human behaviour even though they are not real human beings. The behaviour of machines and AI where human beings play a significant role has brought many renovations in the management of complex tasks of processing, communication, knowledge representation, decision making and suggestions, on potentials of diverse work operations.Practical implicationsThe understanding that the present paper portrays in the context of future library services is that there is no way the AI could function without a human interaction perspective when drawing an analogy from computer science, information science and information systems fields of study.Social implicationsThe interest of users across their background would be strengthen if AI advances transformed the handling complex tasks of processing, communication, knowledge representation, decision-making and giving suggestions, among other things. The possibilities of diverse work operations from empirical evidence of studies consulted in recent times gave the authors the impetus to consider AI as the enabler of future library services.Originality/valueThe increasing demands from library patrons have prompted librarians to adapt their methods of delivering services. These emerging technologies have also brought about shifts in approaches to teaching and learning. Consequently, the recent surge in digital technology-driven service innovations has ushered in a fresh paradigm for education and research. In response to these changes, librarians are actively seeking novel and innovative technologies to enhance user experiences within their libraries. They serve as catalysts for introducing modern and advanced technologies, consistently adapting to contemporary tools that enhance their offerings.

Improving cell type identification with Gaussian noise-augmented single-cell RNA-seq contrastive learning.

Cell type identification is an important task for single-cell RNA-sequencing (scRNA-seq) data analysis. Many prediction methods have recently been proposed, but the predictive accuracy of difficult cell type identification tasks is still low. In this work, we proposed a novel Gaussian noise augmentation-based scRNA-seq contrastive learning method (GsRCL) to learn a type of discriminative feature representations for cell type identification tasks. A large-scale computational evaluation suggests that GsRCL successfully outperformed other state-of-the-art predictive methods on difficult cell type identification tasks, while the conventional random genes masking augmentation-based contrastive learning method also improved the accuracy of easy cell type identification tasks in general.

Comparative Assessment of Neural Radiance Fields and Photogrammetry in Digital Heritage: Impact of Varying Image Conditions on 3D Reconstruction

This paper conducts a comparative evaluation between Neural Radiance Fields (NeRF) and photogrammetry for 3D reconstruction in the cultural heritage domain. Focusing on three case studies, of which the Terpsichore statue serves as a pilot case, the research assesses the quality, consistency, and efficiency of both methods. The results indicate that, under conditions of reduced input data or lower resolution, NeRF outperforms photogrammetry in preserving completeness and material description for the same set of input images (with known camera poses). The study recommends NeRF for scenarios requiring extensive area mapping with limited images, particularly in emergency situations. Despite NeRF’s developmental stage compared to photogrammetry, the findings demonstrate higher potential for describing material characteristics and rendering homogeneous textures with enhanced visual fidelity and accuracy; however, NeRF seems more prone to noise effects. The paper advocates for the future integration of NeRF with photogrammetry to address respective limitations, offering more comprehensive representation for cultural heritage preservation tasks. Future developments include extending applications to planar surfaces and exploring NeRF in virtual and augmented reality, as well as studying NeRF evolution in line with emerging trends in semantic segmentation and in-the-wild scene reconstruction.