Task Representation Research Articles

Semantic-Aware Dynamic Generation Networks for Few-Shot Human-Object Interaction Recognition.

Recognizing human-object interaction (HOI) aims at inferring various relationships between actions and objects. Although great progress in HOI has been made, the long-tail problem and combinatorial explosion problem are still practical challenges. To this end, we formulate HOI as a few-shot task to tackle both challenges and design a novel dynamic generation method to address this task. The proposed approach is called semantic-aware dynamic generation networks (SADG-Nets). Specifically, SADG-Net first assigns semantic-aware task representations for different batches of data, which further generates dynamic parameters. It obtains the features that highlight intercategory discriminability and intracategory commonality adaptively. In addition, we also design a dual semantic-aware encoder module (DSAE-Module), that is, verb-aware and noun-aware branches, to yield both action and object prototypes of HOI for each task space, which generalizes to novel combinations by transferring similarities among interactions. Extensive experimental results on two benchmark datasets, that is, humans interacting with common objects (HICO)-FS and trento universal HOI (TUHOI)-FS, illustrate that our SADG-Net achieves superior performance over state-of-the-art approaches, which proves its impressive effectiveness on few-shot HOI recognition.

Reducing age differences in the retrieval of verbatim and gist representations: Encoding manipulations.

Aging causes changes in episodic memory. While the ability to remember the general idea or gist of past episodes is relatively preserved from the effects of aging, the ability to remember the verbatim details of these episodes declines. The aim of the present study was to examine whether age-related differences in the quality of episodic representations could be reduced by manipulations of information encoding. Two experiments were conducted with younger adults (NExp.1 = 32, NExp.2 = 31, 18-27 years old) and older adults (NExp.1 = 31, NExp.2 = 30, 54-81 years old) in which either the retrieval of gist representations of studied items was facilitated by presenting items that could be grouped within the same category (Exp. 1) or the retrieval of verbatim details of studied items was facilitated by presenting the same items repeatedly (Exp. 2). Both manipulations proved effective in increasing retrieval of either gist or verbatim representations in a recognition task in younger and older adults. Increasing gist retrieval improved correct recognition performance for both younger and older adults but also led the latter to make more recognition errors. Increasing verbatim retrieval improved correct recognition performance for both younger and older adults, and the gain was greater for older adults. These findings support the notion that age-related changes in episodic memory can be explained by changes in the specificity of representations and suggest that simple encoding manipulations could improve episodic memory in older adults.

Domain generalization via causal fine-grained feature decomposition and learning

Domain generalization aims to accurately predict unknown data using models trained by known domain data. Learning domain-invariant representations based on causal inference is one of the popular directions in domain generalization. However, existing domain generalization models based on causal inference cannot correctly determine the invariance conditions of the data. It is because of the interdependent relationship between domain-invariant features and confounding factors that ultimately leads to difficulty in obtaining truly invariant representations by the model. In this paper, we propose a novel domain generalization method called causal fine-grained feature decomposition and learning (CFFDL), which aims to eliminate latent confounding factors and learn causal domain-invariant representations in image classification tasks. Specifically, we design a feature decomposition module based on mutual information, which can decompose deep features into fine-grained feature factors and achieve factor independence between different dimensions, thereby helping to eliminate confounding factors. Furthermore, we introduce a causal representations learning module that can effectively filter and extract relevant causal features of the prediction task while eliminating the influence of confounding factors, improving the model’s performance on domain generalization. Extensive experiments on three domain generalization datasets VLCS, PACS and Office–Home show that our method outperforms the current state-of-the-art models, proving its effectiveness and superiority on domain generalization tasks of image classification.

DPGS: Cross-cooperation guided dynamic points generation for scene text spotting

End-to-end text spotting aims to combine scene text detection and recognition into a unified framework. Dealing with the relationship between the two sub-tasks plays a pivotal role in designing effective spotters. While polygon or segmentation-based methods eliminate heuristic post-processing, they still face challenges such as background noise and high computational burden. In this study, we introduce DPGS, a coarse-to-fine learning framework that lets Dynamic Points Generation for text Spotting. DPGS simultaneously learns character representations for both detection and recognition tasks. Specifically, for each text instance, we represent the character sequence as ordered points and model them with learnable point queries. This approach progressively selects appropriate key points covering character and leverages group attention to associate similar information from different positions, improving detection accuracy. After passing through a single decoder, the point queries encode text semantics and locations, facilitating decoding to central line, boundary, script, and confidence of text through simple prediction heads. Additionally, we introduce an adaptive cooperative criterion to combine more useful feature knowledge, enhancing training efficiency. Extensive experiments show the superiority of our DPGS when handling scene text detection and recognition tasks. Compared to their respective top-1 methods, DPGS has significantly improved the average recognition accuracy by 3.7%, 1.9%, and 0.7% on the Total-Text, ICDAR15, and CTW1500 datasets, respectively.

Enhancing clustering representations with positive proximity and cluster dispersion learning

Contemporary deep clustering approaches often rely on contrastive or non-contrastive techniques to acquire effective representations for clustering tasks. Contrastive methods apply negative pairs to achieve homogenous representations but can introduce class collision problems, potentially compromising clustering performance. In contrast, non-contrastive techniques prevent class collisions but may produce non-uniform representations that lead to clustering collapse. In this work, we propose a novel end-to-end deep clustering approach named PIPCDR, designed to harness the strengths of both approaches while mitigating their limitations. PIPCDR incorporates a positive instance proximity loss and a cluster dispersion regularizer. The positive instance proximity loss ensures alignment between augmented views of instances and their sampled neighbors, enhancing within-cluster compactness by selecting genuinely positive pairs within the embedding space. Meanwhile, the cluster dispersion regularizer maximizes inter-cluster distances while minimizing within-cluster compactness, promoting uniformity in the learned representations. PIPCDR excels in producing well-separated clusters, generating uniform representations, avoiding class collision issues, and enhancing within-cluster compactness. We extensively validate the effectiveness of PIPCDR within an end-to-end Majorize-Minimization framework, demonstrating its competitive performance on moderate-scale clustering benchmark datasets and establishing new state-of-the-art results on large-scale datasets.

Active mutual conjoint estimation of multiple contrast sensitivity functions.

Recent advances in nonparametric contrast sensitivity function (CSF) estimation have yielded a new tradeoff between accuracy and efficiency not available to classical parametric estimators. An additional advantage of this new framework is the ability to independently tune multiple aspects of the estimator to seek further improvements. Machine learning CSF estimation with Gaussian processes allows for design optimization in the kernel, acquisition function, and underlying task representation, to name a few. This article describes a novel kernel for CSF estimation that is more flexible than a kernel based on strictly functional forms. Despite being more flexible, it can result in a more efficient estimator. Further, trial selection for data acquisition that is generalized beyond pure information gain can also improve estimator quality. Finally, introducing latent variable representations underlying general CSF shapes can enable simultaneous estimation of multiple CSFs, such as from different eyes, eccentricities, or luminances. The conditions under which the new procedures perform better than previous nonparametric estimation procedures are presented and quantified.

Inhibition of cued but not executed task sets depends on cue-task compatibility and practice

In task switching, processing a task cue is thought to activate the corresponding task representation (“task set”), thereby allowing for advance task preparation. However, the contribution of preparatory processes to the emergence of n−2 repetition costs as index of task set inhibition processes is debated. The present study investigated whether merely preparing for a task activates a corresponding task set, which needs to be inhibited in order to switch to a different task. To this end, we presented so-called task cue-only trials in trial n−2 and assessed subsequent n−2 repetition costs. The results revealed n−2 repetition costs following a task cue-only, but only for compatible cues with a transparent cue-task relation and only at the beginning of the experiment. In contrast, n−2 repetition costs following task execution in trial n−2 were absent. In a second experiment, we sought to rule out that the presence of n−2 repetition costs following a task cue-only and the corresponding absence following task execution were the consequence of a decay of task sets. This second experiment replicated the result pattern of the first experiment, with n−2 repetition costs following a task cue-only being present only at the beginning of the experiment and only for compatible cues. Hence, cue-induced task set inhibition effects depended on cue-task compatibility and practice. Furthermore, merely prepared task sets were more likely inhibited than executed task sets.

What do voters want from their legislators? evidence from ghana

Abstract Legislators make trade-offs when allocating their time and resources to their multiple tasks of representation, legislation, executive oversight, and constituency service. Furthermore, they must decide how much effort to exert or the balance to strike when undertaking a specific function. Existing research provides limited insights into citizens’ preferences over these officeholder multifaceted decisions in sub-Saharan Africa. I offer novel insights into citizens’ preferences using a conjoint survey experiment of Ghanaians to address this knowledge gap. My findings are threefold. First, I find that citizens put more ‘weight’ on constituency-related activities than parliamentary work. Second, in the constituency, citizens value political representation activities more than constituency services. Third, they weigh public-good-oriented constituency services higher than private ones. The research contributes to our understanding of citizen–legislator accountability relationships in sub-Saharan Africa.

Task switch costs scale with dissimilarity between task rules.

Cognitive flexibility enables humans to voluntarily switch tasks. Task switching requires replacing the previously active task representation with a new one, an operation that typically results in a switch cost. Thus, understanding cognitive flexibility requires understanding how tasks are represented in the brain. We hypothesize that task representations are cognitive map-like, such that the magnitude of the difference between task representations reflects their conceptual differences: The greater the distinction between the two task representations, the more updating is required. This hypothesis predicts that switch costs should increase with between task dissimilarity. To test this hypothesis, we use an experimental design that parametrically manipulates the similarity between task rules. We observe that response time scales with the dissimilarity between the task rules. The findings shed light on the organizational principles of task representations and extend the conventional binary task-switch effect (task repeat vs. switch) to a theoretical framework with parametric task switches. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

A cognition-driven framework for few-shot class-incremental learning

Few-Shot Class-Incremental Learning (FSCIL) aims at integrating new concepts from a minimal set of instances while preserving previously acquired knowledge. This study explores the potential of Vision Transformers (ViTs) in addressing the challenges inherent in the FSCIL paradigm, such as catastrophic forgetting and overfitting problems. Drawing insights from cognitive neuroscience, we propose a Cognition-Driven Framework (CoDF) for FSCIL, leveraging Vision Transformers to emulate human cognitive processes from the aspects of Intuitive Acquisition and Structured Cognition. On the one hand, we employ self-supervised learning techniques to imbue the representations with richer information, thereby facilitating a more intuitive acquisition of essential information to solve FSCIL tasks. On the other hand, we propose to structure the learned representations by introducing biases into the prior distribution of the latent factors, which involves a multivariate Gaussian Mixture Model (GMM) and an intra-class distribution assumption. Furthermore, we apply an innovative extended warm-up strategy to effectively harness the acquired representations for downstream tasks. Comprehensive experiments on three public datasets substantiate the efficacy of our proposed framework.

Source-free cross-domain fault diagnosis of rotating machinery using the Siamese framework

Despite deep learning based intelligent diagnosis has become an essential means of perceiving the health status of rotating machinery, existing diagnostic models struggle to efficiently extract domain-invariant representations for cross-domain tasks in the absence of source domain data due to the limitations of imbalanced machine fault data, the various scenarios, as well as privacy protection in industrial applications. In order to solve these issues, we innovatively propose a source-free cross-domain fault diagnosis algorithm for rotating machinery using the Siamese framework (SCSF). Specifically, a deep residual shrinkage network with parallel spatial and channel-wise attention mechanisms (DRSN-SCW) is used for feature extraction to suppress background noise. In addition, pre-trained feature extraction networks and classifiers use only source domains based on the Siamese framework for end-to-end representation learning to address the problem of fault diagnosis of category-imbalanced data at the feature representation level. Finally, for cross-domain scenarios where the target domain does not have access to source domain data during training, we use a prototype-based pseudo-labeling strategy as well as impose consistency and diversity constraints on the classifier outputs to improve the diagnostic performance of unlabeled target domains. Extensive experiments conducted on the Spectra Quest mechanical fault dataset and the Case Western Reserve University rolling bearing dataset validate the effectiveness of the SCSF. The code is available at .

A Semi-supervised Gaussian Mixture Variational Autoencoder method for few-shot fine-grained fault diagnosis

In practical engineering, obtaining labeled high-quality fault samples poses challenges. Conventional fault diagnosis methods based on deep learning struggle to discern the underlying causes of mechanical faults from a fine-grained perspective, due to the scarcity of annotated data. To tackle those issue, we propose a novel semi-supervised Gaussian Mixed Variational Autoencoder method, SeGMVAE, aimed at acquiring unsupervised representations that can be transferred across fine-grained fault diagnostic tasks, enabling the identification of previously unseen faults using only the small number of labeled samples. Initially, Gaussian mixtures are introduced as a multimodal prior distribution for the Variational Autoencoder. This distribution is dynamically optimized for each task through an expectation–maximization (EM) algorithm, constructing a latent representation of the bridging task and unlabeled samples. Subsequently, a set variational posterior approach is presented to encode each task sample into the latent space, facilitating meta-learning. Finally, semi-supervised EM integrates the posterior of labeled data by acquiring task-specific parameters for diagnosing unseen faults. Results from two experiments demonstrate that SeGMVAE excels in identifying new fine-grained faults and exhibits outstanding performance in cross-domain fault diagnosis across different machines. Our code is available at https://github.com/zhiqan/SeGMVAE.

A puzzle questions form training for self-supervised skeleton-based action recognition

This paper proposed a novel pretext task to address the skeleton-based video representation for self-supervised action recognition tasks. Instead of exploiting only the whole body, various levels of the skeleton structure (e.g., upper body, lower body, left arm, left leg, right arm, right leg, and torso) are employed to extract essential coarser-grained characteristics. This involves computing statistical representations like motion, orientation, trajectory, and magnitude shift from unlabeled skeleton configurations. Then a learning model is built and trained to yield these statistical representations given the sequence configuration as the input. Our approach is question-driven, where each question acts as a puzzle piece contributing to a deeper understanding of the skeleton joint configuration. It's inspired by the ability of the cognitive system observed in individuals to hypothesize unseen actions. This is accomplished by posing pertinent questions and envisioning plausible scenarios to recognize the actions taking place. The answers to these devised questions are derived from the statistical representation of skeleton configurations. To this end, we made 44 questions designed to encompass the broadest overview to the finest detail. Our experiments on the NTU RGB-D, NW-UCLA, and PKU-MMD datasets demonstrate outstanding results in action recognition, proving the superiority of our approach in learning discriminative characteristics.

Remote Sensing Image Classification Based on Canny Operator Enhanced Edge Features.

Remote sensing image classification plays a crucial role in the field of remote sensing interpretation. With the exponential growth of multi-source remote sensing data, accurately extracting target features and comprehending target attributes from complex images significantly impacts classification accuracy. To address these challenges, we propose a Canny edge-enhanced multi-level attention feature fusion network (CAF) for remote sensing image classification. The original image is specifically inputted into a convolutional network for the extraction of global features, while increasing the depth of the convolutional layer facilitates feature extraction at various levels. Additionally, to emphasize detailed target features, we employ the Canny operator for edge information extraction and utilize a convolution layer to capture deep edge features. Finally, by leveraging the Attentional Feature Fusion (AFF) network, we fuse global and detailed features to obtain more discriminative representations for scene classification tasks. The performance of our proposed method (CAF) is evaluated through experiments conducted across three openly accessible datasets for classifying scenes in remote sensing images: NWPU-RESISC45, UCM, and MSTAR. The experimental findings indicate that our approach based on incorporating edge detail information outperforms methods relying solely on global feature-based classifications.

Standards Update: Robot Task Representation Standard and User Guide Development [Standards

Standards Update: Robot Task Representation Standard and User Guide Development [Standards

Surprise-minimization as a solution to the structural credit assignment problem.

The structural credit assignment problem arises when the causal structure between actions and subsequent outcomes is hidden from direct observation. To solve this problem and enable goal-directed behavior, an agent has to infer structure and form a representation thereof. In the scope of this study, we investigate a possible solution in the human brain. We recorded behavioral and electrophysiological data from human participants in a novel variant of the bandit task, where multiple actions lead to multiple outcomes. Crucially, the mapping between actions and outcomes was hidden and not instructed to the participants. Human choice behavior revealed clear hallmarks of credit assignment and learning. Moreover, a computational model which formalizes action selection as the competition between multiple representations of the hidden structure was fit to account for participants data. Starting in a state of uncertainty about the correct representation, the central mechanism of this model is the arbitration of action control towards the representation which minimizes surprise about outcomes. Crucially, single-trial latent-variable analysis reveals that the neural patterns clearly support central quantitative predictions of this surprise minimization model. The results suggest that neural activity is not only related to reinforcement learning under correct as well as incorrect task representations but also reflects central mechanisms of credit assignment and behavioral arbitration.

Interpretable molecular encodings and representations for machine learning tasks

Molecular encodings and their usage in machine learning models have demonstrated significant breakthroughs in biomedical applications, particularly in the classification of peptides and proteins. To this end, we propose a new encoding method: Interpretable Carbon-based Array of Neighborhoods (iCAN). Designed to address machine learning models' need for more structured and less flexible input, it captures the neighborhoods of carbon atoms in a counting array and improves the utility of the resulting encodings for machine learning models. The iCAN method provides interpretable molecular encodings and representations, enabling the comparison of molecular neighborhoods, identification of repeating patterns, and visualization of relevance heat maps for a given data set. When reproducing a large biomedical peptide classification study, it outperforms its predecessor encoding. When extended to proteins, it outperforms a lead structure-based encoding on 71% of the data sets. Our method offers interpretable encodings that can be applied to all organic molecules, including exotic amino acids, cyclic peptides, and larger proteins, making it highly versatile across various domains and data sets. This work establishes a promising new direction for machine learning in peptide and protein classification in biomedicine and healthcare, potentially accelerating advances in drug discovery and disease diagnosis.

Sailfish optimizer based CLAHE with U-NET for MRI brain tumour segmentation

Semantic brain tumour segmentation is a significant contribution to medical image processing. This research is helpful for doctors making diagnoses and determining the severity of lesions. In recent years, convolutional neural networks have sometimes been known as self-remarkable representations in computer vision tasks. In three-dimensional medical image processing, CNN with architecture and skip connections has become increasingly popular. However, they are inadequate at both acquiring nearby and far-off semantic knowledge. However, the transformer's use of a vision. This accomplishment was made possible by the global information modelling capabilities of the transformer. When tryingto perform challenging prediction tasks, such as segmenting 3D medical pictures, it is vital to consider local and global features. In this study, we propose a hypothesis that predicts pre-processing steps that enhance the image quality and significantly impact accuracy with any statistical image classification method. We used a better picture enhancement methodology to back up our idea. This method has three unique stages: a median filter to eliminate noise, a histogram equalisation technique to boost contrast, and a grayscale-to-RGB conversion. The clipping limit of contrast-limited Adaptive Histogram Equalisation (CLAHE)is calculated by the Sailfish Optimization Algorithm (SOA) to achieve optimal picture enhancement. This study classifies semantically segmenting 3D brain tumours as a sequence-to-sequence prediction problem. U-net is proposed in the article as a novel method for segmenting 3D medical images; it incorporates a structure. A new technique for segmenting 3D medical images, U-net, is emerging. We trained and assessed our method on the BRATS 2019 dataset, which was found to outperform state-of-the-art algorithms and be competitive under similar settings.

Multi-step Transfer Learning in Natural Language Processing for the Health Domain

The restricted access to data in healthcare facilities due to patient privacy and confidentiality policies has led to the application of general natural language processing (NLP) techniques advancing relatively slowly in the health domain. Additionally, because clinical data is unique to various institutions and laboratories, there are not enough standards and conventions for data annotation. In places without robust death registration systems, the cause of death (COD) is determined through a verbal autopsy (VA) report. A non-clinician field agent completes a VA report using a set of standardized questions as guide to identify the symptoms of a COD. The narrative text of the VA report is used as a case study to examine the difficulties of applying NLP techniques to the healthcare domain. This paper presents a framework that leverages knowledge across multiple domains via two domain adaptation techniques: feature extraction and fine-tuning. These techniques aim to improve VA text representations for COD classification tasks in the health domain. The framework is motivated by multi-step learning, where a final learning task is realized via a sequence of intermediate learning tasks. The framework builds upon the strengths of the Bidirectional Encoder Representations from Transformers (BERT) and Embeddings from Language Models (ELMo) models pretrained on the general English and biomedical domains. These models are employed to extract features from the VA narratives. Our results demonstrate improved performance when initializing the learning of BERT embeddings with ELMo embeddings. The benefit of incorporating character-level information for learning word embeddings in the English domain, coupled with word-level information for learning word embeddings in the biomedical domain, is also evident.

Two-stage fine-tuning with ChatGPT data augmentation for learning class-imbalanced data

Classification of long-tailed distributed data is a challenging problem, which suffers from serious class imbalance and hence poor performance on tail classes, which have only a few samples. Owing to this paucity of samples, learning on the tail classes is especially challenging for fine-tuning when transferring a pretrained model to a downstream task. In this work, we present a simple modification of standard fine-tuning to cope with these challenges. Specifically, we propose a two-stage fine-tuning. In Stage 1, we fine-tune the final layer of the pretrained model with class-balanced augmented data, generated using ChatGPT. As a large generative language model, ChatGPT is capable of generating novel and contextually similar responses to a given prompt, which makes it an excellent candidate for data augmentation. In Stage 2, we perform the standard fine-tuning. Our modification has several benefits: (1) it leverages pretrained representations by only fine-tuning a small portion of the model parameters while keeping the rest untouched; (2) it allows the model to learn an initial representation of the specific task; and importantly (3) it protects the learning of tail classes from being at a disadvantage during the model updating. We conduct extensive experiments on synthetic datasets of both two-class and multi-class tasks of text classification as well as a real-world application to ADME (i.e., absorption, distribution, metabolism, and excretion) semantic drug labeling. The experimental results show that the proposed two-stage fine-tuning outperforms vanilla fine-tuning and state-of-the-art methods on the above datasets.