Baselines In Terms Of Accuracy Research Articles

A dual-ways feature fusion mechanism enhancing active learning based on TextCNN

Active Learning (AL) is a technique being widely employed to minimize the time and labor costs in the task of annotating data. By querying and extracting the specific instances to train the model, the relevant task’s performance is improved maximally within limited iterations. However, rare work was conducted to fully fuse features from different hierarchies to enhance the effectiveness of active learning. Inspired by the thought of information compensation in many famous deep learning models (such as ResNet, etc.), this work proposes a novel TextCNN-based Two ways Active Learning model (TCTWAL) to extract task-relevant texts. TextCNN takes the advantage of little hyper-parameter tuning and static vectors and achieves excellent results on various natural language processing (NLP) tasks, which are also beneficial to human-computer interaction (HCI) and the AL relevant tasks. In the process of the proposed AL model, the candidate texts are measured from both global and local features by the proposed AL framework TCTWAL depending on the modified TextCNN. Besides, the query strategy is strongly enhanced by maximum normalized log-probability (MNLP), which is sensitive to detecting the longer sentences. Additionally, the selected instances are characterized by general global information and abundant local features simultaneously. To validate the effectiveness of the proposed model, extensive experiments are conducted on three widely used text corpus, and the results are compared with with eight manual designed instance query strategies. The results show that our method outperforms the planned baselines in terms of accuracy, macro precision, macro recall, and macro F1 score. Especially, to the classification results on AG’s News corpus, the improvements of the four indicators after 39 iterations are 40.50%, 45.25%, 48.91%, and 45.25%, respectively.

Distance-Aware Learning for Inductive Link Prediction on Temporal Networks.

Inductive link prediction on temporal networks aims to predict the future links associated with node(s) unseen in the historical timestamps. Existing methods generate the predictions mainly by learning node representation from the node/edge attributes as well as the network dynamics or by measuring the distance between nodes on the temporal network structure. However, the attribute information is unavailable in many realistic applications and the structure-aware methods highly rely on nodes' common neighbors, which are difficult to accurately detect, especially in sparse temporal networks. Thus, we propose a distance-aware learning (DEAL) approach for inductive link prediction on temporal networks. Specifically, we first design an adaptive sampling method to extract temporal adaptive walks for nodes, increasing the probability of including the common neighbors between nodes. Then, we design a dual-channel distance measuring component, which simultaneously measures the distance between nodes in the embedding space and on the dynamic graph structure for predicting future inductive edges. Extensive experiments are conducted on three public temporal network datasets, i.e., MathOverflow, AskUbuntu, and StackOverflow. The experimental results validate the superiority of DEAL over the state-of-the-art baselines in terms of accuracy, area under the ROC curve (AUC), and average precision (AP), where the improvements are especially obvious in scenarios with only limited data.

ECOD: Unsupervised Outlier Detection Using Empirical Cumulative Distribution Functions

Outlier detection refers to the identification of data points that deviate from a general data distribution. Existing unsupervised approaches often suffer from high computational cost, complex hyperparameter tuning, and limited interpretability, especially when working with large, high-dimensional datasets. To address these issues, we present a simple yet effective algorithm called ECOD (Empirical-Cumulative-distribution-based Outlier Detection), which is inspired by the fact that outliers are often the "rare events" that appear in the tails of a distribution. In a nutshell, ECOD first estimates the underlying distribution of the input data in a nonparametric fashion by computing the empirical cumulative distribution per dimension of the data. ECOD then uses these empirical distributions to estimate tail probabilities per dimension for each data point. Finally, ECOD computes an outlier score of each data point by aggregating estimated tail probabilities across dimensions. Our contributions are as follows: (1) we propose a novel outlier detection method called ECOD, which is both parameter-free and easy to interpret; (2) we perform extensive experiments on 30 benchmark datasets, where we find that ECOD outperforms 11 state-of-the-art baselines in terms of accuracy, efficiency, and scalability; and (3) we release an easy-to-use and scalable (with distributed support) Python implementation for accessibility and reproducibility.

Machine fault diagnosis based on data reassembly for a deep neural network trained by imbalanced acoustic recordings

For high-tech equipment, online monitoring and diagnostics play a critical role in ensuring production under normal conditions. While the machines are generally in normal condition, fault conditions are often difficult, if not impossible, to simulate. In this paper, we propose a data augmentation approach to address the challenge of machine fault diagnosis for the above-mentioned scenarios where the normal and abnormal acoustic data sets are imbalanced. To overcome this limitation, we propose a data reassembly approach to generate new signal samples by randomly combining data segments from different machine states. The resulting sample is annotated as a vector according to the proportion of the corresponding machine state. The training with data reassembly is performed for a deep neural network (DNN)-based classifier using the log Mel spectrum as the acoustic features. To validate the proposed approach, we conducted experiments using a rotor kit. Acoustic data associated with the normal and seven abnormal conditions were recorded using a four-microphone uniform circular array (UCA). The results showed that the classifier with data reassembly had superior diagnostic performance compared to the classifier without data reassembly. The proposed method is also compared to several data generation baselines in terms of accuracy and F1 score.

Beyond the Meta: Leveraging Game Design Parameters for Patch-Agnostic Esport Analitics

Esport games comprise a sizeable fraction of the global games market, and is the fastest growing segment in games. This has given rise to the domain of esports analytics, which uses telemetry data from games to inform players, coaches, broadcasters and other stakeholders. Compared to traditional sports, esport titles change rapidly, in terms of mechanics as well as rules. Due to these frequent changes to the parameters of the game, esport analytics models can have a short life-spam, a problem which is largely ignored within the literature. This paper extracts information from game design (i.e. patch notes) and utilises clustering techniques to propose a new form of character representation. As a case study, a neural network model is trained to predict the number of kills in a Dota 2 match utilising this novel character representation technique. The performance of this model is then evaluated against two distinct baselines, including conventional techniques. Not only did the model significantly outperform the baselines in terms of accuracy (85% AUC), but the model also maintains the accuracy in two newer iterations of the game that introduced one new character and a brand new character type. These changes introduced to the design of the game would typically break conventional techniques that are commonly used within the literature. Therefore, the proposed methodology for representing characters can increase the life-spam of machine learning models as well as contribute to a higher performance when compared to traditional techniques typically employed within the literature.

Intent Disentanglement and Feature Self-Supervision for Novel Recommendation

One key property in recommender systems is the long-tail distribution in user-item interactions where most items only have few user feedback. Improving the recommendation of tail items can promote novelty and bring positive effects to both users and providers, and thus is a desirable property of recommender systems. Current novel recommendation studies over-emphasize the importance of tail items without differentiating the degree of users' intent on popularity and often incur a sharp decline of accuracy. Moreover, none of existing methods has ever taken the extreme case of tail items, i.e., cold-start items without any interaction, into consideration. In this work, we first disclose the mechanism that drives a user's interaction towards popular or niche items by disentangling her intent into conformity influence (popularity) and personal interests (preference). We then present a unified end-to-end framework to simultaneously optimize accuracy and novelty targets based on the disentangled intent of popularity and that of preference. We further develop a new paradigm for novel recommendation of cold-start items which exploits the self-supervised learning technique to model the correlation between collaborative features and content features. We conduct extensive experimental results on three real-world datasets. The results demonstrate that our proposed model yields significant improvements over the state-of-the-art baselines in terms of accuracy, novelty, coverage, and trade-off.

A homologous and heterogeneous multi-view inter-patient adaptive network for arrhythmia detection

Background and ObjectiveElectrocardiogram (ECG) is a widely used diagnostic tool for arrhythmia assessment in clinical practice. However, current arrhythmia detection algorithms rely heavily on signal-based data, while cardiologists often use image-based data. This discrepancy, combined with individual differences in physiological signals, poses challenges for accurate arrhythmia detection. To address these challenges and improve arrhythmia detection performance, we propose a homologous and heterogeneous multi-view inter-patient adaptive network. MethodsWe designed a multi-view representation learning module to capture dynamic and morphological characteristics from ECG signals and electrocardiographic images. Expert knowledge was also elicited to gain internally-invariant characteristics of each category. Finally, we designed a new loss function that aligns the embedding of the source and target domains in the feature space to minimize the negative effects of individual differences. ResultsExperiments on the MIT-BIH arrhythmia database demonstrate that our proposed method outperforms state-of-the-art baselines in terms of accuracy, positive predictive value, sensitivity and F1-score. These results indicate the effectiveness of our method in accurately detecting arrhythmias. ConclusionsOur homologous and heterogeneous multi-view inter-patient adaptive network successfully addresses the challenges of utilizing both ECG signal and electrocardiographic image data for arrhythmia detection and overcoming individual differences in physiological signals. Our proposed method has the potential to improve early diagnosis and treatment outcomes of arrhythmias in clinical practice.

Meta-GCN: A Dynamically Weighted Loss Minimization Method for Dealing with the Data Imbalance in Graph Neural Networks

Although many real-world applications, such as disease prediction, and fault detection suffer from class imbalance, most existing graph-based classification methods ignore the skewness of the distribution of classes; therefore, tend to be biased towards the majority class(es). Conventional methods typically tackle this problem through the assignment of weights to each one of the class samples based on a function of their loss, which can lead to over-fitting on outliers. In this paper, we propose a meta-learning algorithm, named Meta-GCN, for adaptively learning the example weights by simultaneously minimizing the unbiased meta-data set loss and optimizing the model weights through the use of a small unbiased meta-data set. Through experiments, we have shown that Meta-GCN outperforms state-of-the-art frameworks and other baselines in terms of accuracy, the area under the receiver operating characteristic (AUC-ROC) curve, and macro F1 Score for classification tasks on two different datasets.

Multi-View Clustering for Integration of Gene Expression and Methylation Data With Tensor Decomposition and Self-Representation Learning.

The accumulated DNA methylation and gene expression provide a great opportunity to exploit the epigenetic patterns of genes, which is the foundation for revealing the underlying mechanisms of biological systems. Current integrative algorithms are criticized for undesirable performance because they fail to address the heterogeneity of expression and methylation data, and the intrinsic relations among them. To solve this issue, a novel multi-view clustering with self-representation learning and low-rank tensor constraint (MCSL-LTC) is proposed for the integration of gene expression and DNA methylation data, which are treated as complementary views. Specifically, MCSL-LTC first learns the low-dimensional features for each view with the linear projection, and then these features are fused in a unified tensor space with low-rank constraints. In this case, the complementary information of various views is precisely captured, where the heterogeneity of omic data is avoided, thereby enhancing the consistency of different views. Finally, MCSL-LTC obtains a consensus cluster of genes reflecting the structure and features of various views. Experimental results demonstrate that the proposed approach outperforms state-of-the-art baselines in terms of accuracy on both the social and cancer data, which provides an effective and efficient method for the integration of heterogeneous genomic data.

Multi-Modal Interaction-Aware Motion Prediction at Unsignalized Intersections

Autonomous vehicle technologies have evolved quickly over the last few years, with safety being one of the key requirements for their full deployment. However, ensuring their safety while navigating through highly interactive and complex scenarios remains a critical challenge. To tackle this problem, intention estimation and motion prediction are fundamental. In this work, a method to infer the intentions, based on a Dynamic Bayesian Network (DBN), and predict the motion, using Markov Chains, of the nearby vehicles at unsignalized intersections is proposed. This approach considers all possible corridors of the surrounding traffic participants and takes into account their interactions to infer the probabilities of stopping or crossing the intersection, as well as the probability of being in each of the possible navigable corridors. To achieve this, the DBN is used to model the relationships between the observed states and the unobserved intentions of the nearby agents. The Markov Chain model, obtained from a kinematic model, is used to predict the future motions of the vehicles, taking into account their current state, their inferred intentions, and the uncertainty associated with the prediction. The resulting multi-modal motion predictions are sent to the ego vehicle to navigate through the scene. The proposed method is evaluated in 6 real situations extracted from publicly available datasets and is compared with a model-based and a learn-based baseline models. The results showed that the proposed method outperformed both baselines in terms of accuracy considering the metrics ADE and FDE.

OCPHN: Outfit Compatibility Prediction with Hypergraph Networks

With the rapid development of the online shopping, the pursuit of outfit compatibility has become a basic requirement for an increasing number of customers. The existing work on outfit compatibility prediction largely focuses on modeling pairwise item compatibility without considering modeling the whole outfit directly. To address the problem, in this paper, we propose a novel hypergraph-based compatibility modeling scheme named OCPHN, which is able to better model complex relationships among outfits. In OCPHN, we represent the outfit as a hypergraph, where each hypernode represents a category and each hyperedge represents the interactions between multiple categories (i.e., they appear in the same outfit). To better predict outfit compatibility, the hypergraph is transformed into a simple graph, and the message propagation mechanism in the graph convolution network is used to aggregate the neighbours’ information on the node and update the node representations. Furthermore, with learned node representations, an attention mechanism is introduced to compute the outfit compatibility score. Using a benchmark dataset, the experimental results show that the proposed method is an improvement over the strangest baselines in terms of accuracy by about 3% and 1% in the fill-in-the-blank and compatibility prediction tasks, respectively.

UFKT: Unimportant filters knowledge transfer for CNN pruning

As the deep learning models have been widely used in recent years, there is a high demand for reducing the model size in terms of memory and computation without much compromise in the model performance. Filter pruning is a very widely adopted strategy for model compression. The existing filter pruning methods identify the unimportant filters and prune them without worrying about information loss. They try to recover the same by fine-tuning the remaining filters, limiting their performance. In this paper, we tackle this problem by utilizing the knowledge from unimportant filters before pruning to minimize information loss. First, the proposed method identifies the unimportant and important filters by exploiting the lower and higher importance, respectively, using the L1-norm of filters. Next, the proposed custom UFKT-Reg regularizer (Rufkt) transfers the knowledge from unimportant filters before pruning to remaining filters, notably to a fixed number of important filters. Hence, the proposed method minimizes information loss due to the removal of unimportant filters. The experiments are conducted using the three benchmark datasets, including MNIST, CIFAR-10, and ImageNet. The proposed filter pruning method outperforms many recent state-of-the-art filter pruning methods. An improvement over the baseline in terms of accuracy is observed even after removing 95.15%, 62.28%, and 62.39% of the Floating Point OPerations (FLOPs) from architectures LeNet-5, ResNet-56, and ResNet-110, respectively. After pruning 53.25% of FLOPS from ResNet-50, only 1.02% and 0.47% of drops are observed in top-1 and top-5 accuracies, respectively. The code used in this paper will be publicly available at( https://github.com/sarvanichinthapalli/UFKT).

ASCUE: An Adversarial Network-Based Semantical Conformance Checking Method for Unsupervised Event Extraction in Social Internet of Things

Abstract Event extraction (EE) methods are widely used in the Social Internet of Things (SIoT) to help objects obtain the key information from messages shared by other objects. Existing supervised EE methods can only extract predefined events and can hardly extract events with unseen event types. To address this issue, we propose an unsupervised EE method based on the idea of semantic role labeling. However, after extracting all possible events from a given message text, some of these events face the semantically inconsistent issue that will destroy the information credibility in SIoT. To solve this issue, we present ASCUE, an adversarial network-based semantic conformance checking method for unsupervised EE in SIoT. Briefly, ASCUE first mines all the event candidates from a given text unsupervisedly. Next, ASCUE introduces two independent Bidirectional Encoder Representations from Transformers models to capture the semantics of the event candidate and text, respectively. Moreover, to enable the model to achieve competitive performance as label embedding for semantic conformance checking, an adversarial network is added into ASCUE’s training stage. Finally, ASCUE extracts all the events that are semantically consistent with the original text by classifying them into two categories: semantically consistent and semantically inconsistent. In addition, since the existing datasets do not label whether the events are semantically consistent with the original text, we re-annotate an existing dataset to fit our task. Experimental results on the re-annotated dataset show that our model outperforms the state-of-the-art baselines in terms of accuracy and F1 scores.

Combining Unsupervised and Supervised Learning for Sample Efficient Continuous Language Grounding

Natural and efficient communication with humans requires artificial agents that are able to understand the meaning of natural language. However, understanding natural language is non-trivial and requires proper grounding mechanisms to create links between words and corresponding perceptual information. Since the introduction of the “Symbol Grounding Problem” in 1990, many different grounding approaches have been proposed that either employed supervised or unsupervised learning mechanisms. The latter have the advantage that no other agent is required to learn the correct groundings, while the former are often more sample-efficient and accurate but require the support of another agent, like a human or another artificial agent. Although combining both paradigms seems natural, it has not achieved much attention. Therefore, this paper proposes a hybrid grounding framework which combines both learning paradigms so that it is able to utilize support from a tutor, if available, while it can still learn when no support is provided. Additionally, the framework has been designed to learn in a continuous and open-ended manner so that no explicit training phase is required. The proposed framework is evaluated through two different grounding scenarios and its unsupervised grounding component is compared to a state-of-the-art unsupervised Bayesian grounding framework, while the benefit of combining both paradigms is evaluated through the analysis of different feedback rates. The obtained results show that the employed unsupervised grounding mechanism outperforms the baseline in terms of accuracy, transparency, and deployability and that combining both paradigms increases both the sample-efficiency as well as the accuracy of purely unsupervised grounding, while it ensures that the framework is still able to learn the correct mappings, when no supervision is available.

Distinguishing natural and computer generated images using Multi-Colorspace fused EfficientNet

The problem of distinguishing natural images from photo-realistic computer generated ones either addresses natural images versus computer graphics or natural images versus GAN images at a time. But in a real-world image forensic scenario, it is highly essential to consider all categories of image generation since in most cases image generation is unknown. We for the first time to our best knowledge, approach the problem of distinguishing natural images from photo-realistic computer generated images as a three-class classification task classifying natural, computer graphics and GAN images. For the task, we propose a Multi-Colorspace fused EfficientNet model by parallelly fusing three EfficientNet networks that follow transfer learning methodology where each of the three networks operates in a different colorspace, one in RGB, the other in LCH and the last in HSV that are chosen after analyzing the efficacy of various colorspace transformations in this image forensics problem. Our model outperforms the baselines in terms of accuracy, robustness towards post-processing and generalizability towards other datasets. We conduct psychophysics experiments to understand how accurately humans can distinguish natural, computer graphics and GAN images where we could observe that humans find difficulty in classifying these images, particularly the computer generated images, indicating the necessity of computational algorithms for the task. We also analyze the behavior of our model through visual explanations to understand salient regions that contribute to model’s decision making and compare with manual explanations provided by human participants in the form of region markings where we could observe similarities in both the explanations indicating powerful nature of our model to take the decisions meaningfully.

Geotagging Social Media Posts to Landmarks Using Hierarchical BERT (Student Abstract)

Geographical information provided in social media data is useful for many valuable applications. However, only a small proportion of social media posts are explicitly geotagged with their posting locations, which makes the pursuit of these applications challenging. Motivated by this, we propose a 2-level hierarchical classification method that builds upon a BERT model, coupled with textual information and temporal context, which we denote HierBERT. As far as we are aware, this work is the first to utilize a 2-level hierarchical classification approach alongside BERT and temporal information for geolocation prediction. Experimental results based on two social media datasets show that HierBERT outperforms various state-of-art baselines in terms of accuracy and distance error metrics.

RESCUE: Enabling green healthcare services using integrated IoT‐edge‐fog‐cloud computing environments

AbstractInternet of Things (IoT) has a pivotal role in developing intelligent and computational solutions to facilitate varied real‐life applications. To execute high‐end computations and data analytics, IoT and cloud‐based solutions play the most significant role. However, frequent communication with long distant cloud servers is not a delay‐aware and energy‐efficient solution while providing time‐critical applications such as healthcare. This article explores the possibilities and opportunities of integrating cloud technology with fog and edge‐based computing to provide healthcare services to users in exigency. Here, we propose an end‐to‐end framework namedRESCUE(enabling green healthcare services using integrated iot‐edge‐fog‐cloud computing environments), consisting efficient spatio‐temporal data analytics module for efficient information sharing, spatio‐temporal data analysis to predict the path for users to reach the destination (healthcare center or relief camps) with minimum delay in the time of exigency (say, natural disaster). This module analyzes the collected information through crowd‐sourcing and assists the user by extracting optimal path postdisaster when many regions are nonreachable. Our work is different from the existing literature in varied aspects: it analyses the context and semantics by augmenting real‐time volunteered geographical information (VGI) and refines it. Furthermore, the novel path prediction module incorporates such VGI instances and predicts routes in emergencies avoiding all possible risks. Also, the design of development of a latency‐aware, power‐aware data‐driven analytics system helps to resolve any spatio‐temporal query more efficiently compared to the existing works for any time‐critical application. The experimental and simulation results outperform the baselines in terms of accuracy, delay, and power consumption.

A Contrastive learning-based Task Adaptation model for few-shot intent recognition

Few-shot intent recognition aims to identify user’s intent from the utterance with limited training data. A considerable number of existing methods mainly rely on the generic knowledge acquired on the base classes to identify the novel classes. Such methods typically ignore the characteristics of each meta task itself, resulting in the inability to make full use of limited given samples when classifying unseen classes. To deal with such issues, we propose a Contrastive learning-based Task Adaptation model (CTA) for few-shot intent recognition. In detail, we leverage contrastive learning to help achieve task adaptation and make full use of the limited samples of novel classes. First, a self-attention layer is employed in the task adaptation module, which aims to establish interactions between samples of different categories so that new representations are task-specific rather than relying entirely on the base classes. Then, the contrastive-based loss functions and the semantics of the label name are respectively used for reducing the similarity between sample representations in different categories while increasing it in the same categories. Experimental results on a public dataset OOS verify the effectiveness of our proposal by beating the competitive baselines in terms of accuracy. Besides, we conduct the cross-domain experiments on three datasets, i.e., OOS, SNIPS as well as ATIS. We find that CTA gains obvious improvements in terms of accuracy in all cross-domain experiments, indicating that it has a better generalization ability than other competitive baselines in both cross-domain and single-domain settings.

A knowledge-driven activity recognition framework for learning unknown activities

Human activity recognition has increasingly received attention in recent years to track regular activities of people. Existing activity recognition approaches considerably contributed to the analysis of human behavior. However, they still confront numerous issues related to the variability of activities performed by people within dynamic environments. Generally, this variability renders the used training or ontology models with predefined activities unsuitable. Therefore, creating an activity recognition approach that is able to leverage dynamically new and unknown activities at runtime becomes important. In this paper, we propose a novel knowledge-driven activity recognition framework using smartphone. This framework envisions taking a knowledge-driven approach to reinforce the recognition accuracy and people's quality of life in the context of dynamic environments at runtime. More specifically, we propose an ontology-based context evolution along with a dynamic decision-making, so that new and unknown performed activities can be accurately recognized. Furthermore, we use a public activity recognition dataset to demonstrate the effectiveness of the proposed framework and show its advantage over a data-driven baselines in terms of accuracy. Experimental results reveal that our framework not only reinforces the accuracy, but also enables an effective activity learning when facing unknown activities at runtime.

MBFFNet: Multi-Branch Feature Fusion Network for Colonoscopy.

Colonoscopy is currently one of the main methods for the detection of rectal polyps, rectal cancer, and other diseases. With the rapid development of computer vision, deep learning–based semantic segmentation methods can be applied to the detection of medical lesions. However, it is challenging for current methods to detect polyps with high accuracy and real-time performance. To solve this problem, we propose a multi-branch feature fusion network (MBFFNet), which is an accurate real-time segmentation method for detecting colonoscopy. First, we use UNet as the basis of our model architecture and adopt stepwise sampling with channel multiplication to integrate features, which decreases the number of flops caused by stacking channels in UNet. Second, to improve model accuracy, we extract features from multiple layers and resize feature maps to the same size in different ways, such as up-sampling and pooling, to supplement information lost in multiplication-based up-sampling. Based on mIOU and Dice loss with cross entropy (CE), we conduct experiments in both CPU and GPU environments to verify the effectiveness of our model. The experimental results show that our proposed MBFFNet is superior to the selected baselines in terms of accuracy, model size, and flops. mIOU, F score, and Dice loss with CE reached 0.8952, 0.9450, and 0.1602, respectively, which were better than those of UNet, UNet++, and other networks. Compared with UNet, the flop count decreased by 73.2%, and the number of participants also decreased. The actual segmentation effect of MBFFNet is only lower than that of PraNet, the number of parameters is 78.27% of that of PraNet, and the flop count is 0.23% that of PraNet. In addition, experiments on other types of medical tasks show that MBFFNet has good potential for general application in medical image segmentation.