General Learning Research Articles

Backdoor Attacks and Defenses Targeting Multi-Domain AI Models: A Comprehensive Review

Since the emergence of security concerns in artificial intelligence (AI), there has been significant attention devoted to the examination of backdoor attacks. Attackers can utilize backdoor attacks to manipulate model predictions, leading to significant potential harm. However, current research on backdoor attacks and defenses in both theoretical and practical fields still has many shortcomings. To systematically analyze these shortcomings and address the lack of comprehensive reviews, this paper presents a comprehensive and systematic summary of both backdoor attacks and defenses targeting multi-domain AI models. Simultaneously, based on the design principles and shared characteristics of triggers in different domains and the implementation stages of backdoor defense, this paper proposes a new classification method for backdoor attacks and defenses. We use this method to extensively review backdoor attacks in the fields of computer vision and natural language processing, and also examine the current applications of backdoor attacks in audio recognition, video action recognition, multimodal tasks, time series tasks, generative learning, and reinforcement learning, while critically analyzing the open problems of various backdoor attack techniques and defense strategies. Finally, this paper builds upon the analysis of the current state of AI security to further explore potential future research directions for backdoor attacks and defenses.

Fault diagnosis method for elevator carriages based on temporal generative federated distillation

Addressing the ‘data silo’ issue among different elevator operating units and the temporal correlations in elevator vibration signals, a novel small-sample fault diagnosis method for elevator carriages based on temporal generative federated distillation is proposed. This method incorporates a temporal generative adversarial network into Federated Distillation via Generative Learning (FedGen). FedGen combines federated learning, knowledge distillation, and generative models to enhance model aggregation efficiency while mitigating data heterogeneity. However, the original generative model struggles to maintain dynamic correlations between signals when extracting temporal features. Therefore, an improved Time Series Generative Adversarial Networks (TimeGAN) model is introduced, substituting the initial logarithmic loss function with a least squares error function, thereby enhancing training stability and data quality. This approach eliminates the need for proxy datasets in knowledge distillation, avoiding the loss of temporal information during central server feature extraction. Simulation results demonstrate that this method enables data sharing while protecting data privacy, and enhances model generalization capabilities.

A Machine-Learning Approach to Mitigate Ground Clutter Effects in the GPM Combined Radar-Radiometer Algorithm (CORRA) Precipitation Estimates

Abstract In this study, a machine-learning based methodology is developed to mitigate the effects of ground clutter on precipitation estimates from the Global Precipitation Mission Combined Radar-Radiometer Algorithm. Ground clutter can corrupt and obscure precipitation echo in radar observations, leading to inaccuracies in precipitation estimates. To improve upon previous work, this study introduces a general machine learning (ML) approach that enables a systematic investigation and a better understanding of uncertainties in clutter mitigation. To allow for a less restrictive exploration of conditional relations between precipitation above the lowest clutter-free bin and surface precipitation, reflectivity observations above the clutter are included in a fixed-size set of predictors along with the precipitation type, surface type, and freezing level to estimate surface precipitation rates, and several ML-based estimation methods are investigated. A Neural Network Model (NN) is ultimately identified as the best candidate for systematic evaluations, as it is computationally fast to apply while effective in applications. The NN provides unbiased estimates; however, it does not significantly outperform a simple bias correction approach in reducing random errors in the estimates. The similar performance of other ML approaches suggests that the NN’s limited improvement beyond bias removal is due to indeterminacies in the data rather than limitations in the ML approach itself.

Lembar Kerja Peserta Didik Berdasarkan Model Pembelajaran Generatif untuk Materi Gelombang Cahaya Kelas XI SMA

This research aims to develop LKPD based on a generative learning model on light wave material for class XI SMA that is valid and practical. The generative learning model was chosen because it is able to encourage students to actively build their understanding through interaction, reflection and discussion. This model has proven effective in improving critical thinking and problem-solving skills in physics learning. The developed worksheet is designed following the syntax of the generative learning model: orientation, cognitive conflict, disclosure, construct, application, and reflection, to support in-depth learning and link theory with real application. This research uses the Research and Development (R&D) method with the 4D development model (define, design, develop, disseminate). The define stage includes needs analysis and problem identification. At the design stage, the LKPD was designed by taking into account the content, structure, and aesthetic components. The development stage involves validity testing by experts, well as practicality testing by teachers and students. The dissemination stage is planned for wider application. The results showed that this LKPD has a very high level of validity with a validation index value of 0.94 in the content aspect and 0.86 in the structure aspect, which means that this LKPD meets the eligibility standards of the Merdeka curriculum. The practicality test shows that the LKPD is very practical with a teacher assessment score of 95% and a student assessment of 87%, so it is feasible to use it in the real field because it has been declared valid and practical.

Neurophysiological and Autonomic Dynamics of Threat Processing during Sustained Social Fear Generalization.

Survival in dynamic environments requires that organisms learn to predict danger from situational cues. One key facet of threat prediction is generalization from a predictive cue to similar cues, ensuring that a cue-outcome contingency is applied beyond the original learning environment. Generalization has been observed in laboratory studies of aversive conditioning: Behavioral and physiological processes generalize responses from a stimulus paired with threat (the conditioned stimulus [CS+]) to unpaired stimuli, with response magnitudes varying with CS+ similarity. In contrast, work focusing on sensory responses in visual cortex has found a sharpening pattern, in which responses to stimuli closely resembling the CS+ are maximally suppressed, potentially reflecting lateral inhibitory interactions with the CS+ representation. Originally demonstrated with simple visual cues, changes in visuocortical tuning have also been observed in threat generalization learning across facial identities. It is unclear to what extent these visuocortical changes represent transient or sustained effects and if generalization learning requires prior conditioning to the CS+. The present study addressed these questions using EEG and pupillometry in an aversive generalization paradigm involving hundreds of trials using a gradient of facial identities. Visuocortical steady-state visual evoked potential sharpening occurred after dozens of trials of generalization learning without prior differential conditioning, but diminished as learning continued. By contrast, generalization of alpha power suppression, pupil dilation, and self-reported valence and arousal was seen throughout the experiment. Findings are consistent with threat processing models emphasizing the role of changing visucocortical and attentional dynamics when forming, curating, and shaping fear memories as observers continue learning about stimulus-outcome contingencies.

Family income and polygenic scores are independently but not interactively associated with cognitive performance among youth genetically similar to European reference populations.

Cognitive abilities are heritable and influenced by socioeconomic status (SES). It is critical to understand the association between SES and cognition beyond genetic propensity to inform potential benefits of SES-based interventions and to determine if such associations vary across (i) cognitive domains, (ii) facets of SES, and/or (iii) genetic propensity for different aspects of cognition. We examined the contributions of neighborhood socioeconomic advantage, family income, and polygenic scores (PGS) for domains of cognition (i.e., general cognitive ability, executive function, learning and memory, fluid reasoning) in a sample of children (ages 9-10; n = 5549) most genetically similar to reference populations from Europe. With some variability across cognitive outcomes, family income and PGS were independently significantly associated with cognitive performance. Within-sibling analyses revealed that cognitive PGS associations were predominantly driven by between-family effects suggestive of non-direct genetic mechanisms. These findings provide evidence that SES and genetic propensity to cognition have unique associations with cognitive performance in middle childhood. These results underscore the importance of environmental factors and genetic influences in the development of cognitive abilities and caution against overinterpreting associations with PGS of cognitive and educational outcomes as predominantly direct genetic effects.

FNContra: Frequency-domain Negative Sample Mining in Contrastive Learning for limited-data image generation

Substantial training data is necessary to train an effective generative adversarial network(GANs), without which the discriminator is easily overfitting, causing the sub-optimal models. To solve these problems, this work explores the Frequency-domain Negative Sample Mining in Contrastive learning (FNContra) to improve data efficiency, which requires the discriminator to differentiate the definite relationships between the negative samples and real images. Concretely, this work first constructs multiple-level negative samples in the frequency domain and then proposes Discriminated Wavelet-instance Contrastive Learning (DWCL) and Generated Wavelet-prototype Contrastive Learning (GWCL). The former helps the discriminator learn the fine-grained texture features, and the latter impels the generated feature distribution to be close to real. Considering the learning difficulty of multi-level negative samples, this work proposes a dynamic weight driven by self-information, which ensures the resultant force is positive from the multi-level negative samples during the training. Finally, this work performs experiments on eleven datasets with different domains and resolutions. The quantitative and qualitative results demonstrate the superiority and effectiveness of the FNContra trained on limited data, and it indicates that FNContra can synthesize high-quality images. Notably, FNContra achieves the best FID scores on 10 out of 11 datasets, with improvements of 17.90 and 29.24 on Moongate and Shells, respectively, compared to the baseline. The code can be found at https://github.com/YQX1996/FNContra.

Workshop training to facilitate parental involvement in their children’s mathematics education: parents’ perceptions

This study investigated the effectiveness of workshops on parental involvement in their children’s mathematical education. The study employed a quasi-experimental with a pre-test-post-test design. This study involved 76 parents from one senior secondary school in the Kweneng Region, Botswana. The study included three steps for data analysis: thematic content analysis, hypothesis testing, and descriptive statistical analysis. The results showed that the parental involvement level before and after the intervention marked a difference with a significant increase. Parental involvement variables such as parenting style, parental expectations, home rules, parental supervision, communication between parents and children, children’s home mathematics activities, and parental attitude toward school all improved significantly. The findings further indicated that parental involvement in their child's general and mathematical learning can be improved. As a result, the study recommended that all basic education schools reach out to parents through parent education programs that can assist parents in understanding the value of being active in their children’s mathematics education, which will probably affect their education performance positively.

Optimizing data transmission in 6G software defined networks using deep reinforcement learning for next generation of virtual environments

Data transmission of Virtual Reality (VR) plays an important role in delivering a powerful VR experience. This increasing demand on both high bandwidth and low latency. 6G emerging technologies like Software Defined Network (SDN) and resource slicing are acting as promising technologies for addressing the transmission requirements of VR users. Efficient resource management becomes dominant to ensure a satisfactory user experience. The integration of Deep Reinforcement Learning (DRL) allows for dynamic network resource balancing, minimizing communication latency and maximizing data transmission rates wirelessly. Employing slicing techniques further aids in managing distributed resources across the network for different services as enhanced Mobile Broadband (eMBB) and Ultra-Reliable and Low Latency Communications (URLLC). The proposed VR-based SDN system model for 6G cellular networks facilitates centralized administration of resources, enhancing communication between VR users. This innovative solution seeks to contribute to the effective and streamlined resource management essential for VR video transmission in 6G cellular networks. The utilization of Deep Reinforcement Learning (DRL) approaches, is presented as an alternative solution, showcasing significant performance and feature distinctions through comparative results. Our results show that implementing strategies based on DRL leads to a considerable improvement in the resource management process as well as in the achievable data rate and a reduction in the necessary latency in dynamic and large scale networks.

A survey on person and vehicle re‐identification

AbstractPerson/vehicle re‐identification aims to use technologies such as cross‐camera retrieval to associate the same person (same vehicle) in the surveillance videos at different locations, different times, and images captured by different cameras so as to achieve cross‐surveillance image matching, person retrieval and trajectory tracking. It plays an extremely important role in the fields of intelligent security, criminal investigation etc. In recent years, the rapid development of deep learning technology has significantly propelled the advancement of re‐identification (Re‐ID) technology. An increasing number of technical methods have emerged, aiming to enhance Re‐ID performance. This paper summarises four popular research areas in the current field of re‐identification, focusing on the current research hotspots. These areas include the multi‐task learning domain, the generalisation learning domain, the cross‐modality domain, and the optimisation learning domain. Specifically, the paper analyses various challenges faced within these domains and elaborates on different deep learning frameworks and networks that address these challenges. A comparative analysis of re‐identification tasks from various classification perspectives is provided, introducing mainstream research directions and current achievements. Finally, insights into future development trends are presented.

Ablation Study of Diffusion Model with Transformer Backbone for Low-count PET Denoising.

Diffusion models (DM) built from a hierarchy of denoising autoencoders have achieved remarkable progress in image generation, and are increasingly influential in the field of image restoration (IR) tasks. In the meantime, its backbone of autoencoders also evolved from UNet to vision transformer, e.g. Restormer. Therefore, it is important to disentangle the contribution of backbone networks and the additional generative learning scheme. Notably, DM shows varied performance across IR tasks, and the performance of recent advanced transformer-based DM on PET denoising is under-explored. In this study, we further raise an intuitive question, "{if we have a sufficiently powerful backbone, whether DM can be a general add-on generative learning scheme to further boost PET denoising}". Specifically, we investigate one of the best-in-class IR models, i.e., DiffIR, which is a latent DM based on the Restormer backbone. We provide a qualitative and quantitative comparison with UNet, SR3 (UNet+pixel DM), and Restormer, on the 25% low dose 18F-FDG whole-body PET denoising task, aiming to identify the best practices. We trained and tested on 93 and 12 subjects, and each subject has 644 slices. It appears that Restormer outperforms UNet in terms of PSNR and MSE. However, additional latent DM over Restormer does not contribute to better MSE, SSIM, or PSNR in our task, which is even inferior to the conventional UNet. In addition, SR3 with pixel space DM is not stable to synthesize satisfactory results. The results are consistent with the natural image super-resolution tasks, which also suffer from limited spatial information. A possible reason would be the denoising iteration at latent feature space cannot well support detailed structure and texture restoration. This issue is more crucial in the IR tasks taking inputs with limited details, e.g., SR and PET denoising.

Airborne LiDAR Point Cloud Classification Using Ensemble Learning for DEM Generation.

Airborne laser scanning (ALS) point clouds have emerged as a predominant data source for the generation of digital elevation models (DEM) in recent years. Traditionally, the generation of DEM using ALS point clouds involves the steps of point cloud classification or ground point filtering to extract ground points and labor-intensive post-processing to correct the misclassified ground points. The current deep learning techniques leverage the ability of geometric recognition for ground point classification. However, the deep learning classifiers are generally trained using 3D point clouds with simple geometric terrains, which decrease the performance of model inferencing. In this study, a point-based deep learning model with boosting ensemble learning and a set of geometric features as the model inputs is proposed. With the ensemble learning strategy, this study integrates specialized ground point classifiers designed for different terrains to boost classification robustness and accuracy. In experiments, ALS point clouds containing various terrains were used to evaluate the feasibility of the proposed method. The results demonstrated that the proposed method can improve the point cloud classification and the quality of generated DEMs. The classification accuracy and F1 score are improved from 80.9% to 92.2%, and 82.2% to 94.2%, respectively, by using the proposed methods. In addition, the DEM generation error, in terms of mean squared error (RMSE), is reduced from 0.318-1.362 m to 0.273-1.032 m by using the proposed ensemble learning.

GDMol: Generative Double-Masking Self-Supervised Learning for Molecular Property Prediction.

Effective molecular feature representation is crucial for drug property prediction. Recent years have seen increased attention on graph neural networks (GNNs) that are pre-trained using self-supervised learning techniques, aiming to overcome the scarcity of labeled data in molecular property prediction. Traditional GNNs in self-supervised molecular property prediction typically perform a single masking operation on the nodes and edges of the input molecular graph, masking only local information and insufficient for thorough self-supervised training. Hence, we propose a model for molecular property prediction based on generative double-masking self-supervised learning, termed as GDMol. This integrates generative learning into the self-supervised learning framework for latent representation, and applies a second round of masking to these latent representations, enabling the model to better capture global information and semantic knowledge of the molecules for a richer, more informative representation, thereby achieving more accurate and robust molecular property prediction. Our experiments on 5 datasets demonstrated superior performance of GDMol in predicting molecular properties across different domains. Moreover, we used the masking operation to traverse through the gradient changes of each node, the magnitude and sign of which reflect the positive and negative contribution respectively of the local structure in the molecule to the prediction outcome. This in-depth interpretative analysis not only enhances the model's interpretability, but also provides more targeted insights and direction for optimizing drug molecules. In summary, this research offers novel insights on improving molecular property prediction tasks, and paves the way for further research on the application of generative learning and self-supervised learning in the field of chemistry.

Decomposed Prototype Learning for Few-Shot Scene Graph Generation

Today's scene graph generation (SGG) models typically require abundant manual annotations to learn new predicate types. Therefore, it is difficult to apply them to real-world applications with massive uncommon predicate categories whose annotations are hard to collect. In this paper, we focus on Few-Shot SGG (FSSGG) , which encourages SGG models to be able to quickly transfer previous knowledge and recognize unseen predicates well with only a few examples. However, current methods for FSSGG are hindered by the high intra-class variance of predicate categories in SGG: On one hand, each predicate category commonly has multiple semantic meanings under different contexts. On the other hand, the visual appearance of relation triplets with the same predicate differs greatly under different subject-object compositions. Such great variance of inputs makes it hard to learn generalizable representation for each predicate category with current few-shot learning (FSL) methods. However, we found that this intra-class variance of predicates is highly related to the composed subjects and objects. To model the intra-class variance of predicates with subject-object context, we propose a novel Decomposed Prototype Learning (DPL) model for FSSGG. Specifically, we first construct a decomposable prototype space to capture diverse semantics and visual patterns of subjects and objects for predicates by decomposing them into multiple prototypes. Afterwards, we integrate these prototypes with different weights to generate query-adaptive predicate representation with more reliable semantics for each query sample. We conduct extensive experiments and compare with various baseline methods to show the effectiveness of our method.

Quantifying population-level neural tuning functions using Ricker wavelets and the Bayesian bootstrap

BackgroundExperience changes visuo-cortical tuning. In humans, re-tuning has been studied during aversive generalization learning, in which the similarity of generalization stimuli (GSs) with a conditioned threat cue (CS+) is used to quantify tuning functions. Previous work utilized pre-defined tuning shapes (generalization and sharpening patterns). This approach may constrain the ways in which re-tuning can be characterized since the tuning patterns may not match the prototypical functions. New methodThe present study proposes a flexible and data-driven method for precisely quantifying changes in tuning based on the Ricker wavelet function and the Bayesian bootstrap. This method was applied to EEG and psychophysics data from an aversive generalization learning paradigm. ResultsThe Ricker wavelet model fitted the steady-state visual event potentials (ssVEP), alpha-band power, and detection accuracy data well. A Morlet wavelet function was used for comparison and fit the data better in some situations, but was more challenging to interpret. The pattern of re-tuning in the EEG data, predicted by the Ricker model, resembled the shapes of the best fitting a-priori patterns. Comparison with existing methodsAlthough the re-tuning shape modeled by the Ricker function resembled the pre-defined shapes, the Ricker approach led to greater Bayes factors and more interpretable results compared to a-priori models. The Ricker approach was more easily fit and led to more interpretable results than a Morlet wavelet model. ConclusionThis work highlights the promise of the current method for capturing the precise nature of visuo-cortical tuning, unconstrained by the implementation of a-priori models.

SVQ-MAE: an efficient speech pre-training framework with constrained computational resources

Self-supervised learning for speech pre-training models has achieved remarkable success in acquiring superior speech contextual representations by learning from unlabeled audio, excelling in numerous downstream speech tasks. However, the pre-training of these models necessitates significant computational resources and training duration, presenting a high barrier to entry into the realm of pre-training learning. In our efforts, by amalgamating the resource-efficient benefits of the generative learning model, Masked Auto Encoder, with the efficacy of the vector quantization method in discriminative learning, we introduce a novel pre-training framework: Speech Vector Quantization Masked Auto Encoder (SVQ-MAE). Distinct from the majority of SSL frameworks, which require simultaneous construction of speech contextual representations and mask reconstruction within an encoder-only module, we have exclusively designed a decoupled decoder for pre-training SVQ-MAE. This allows the additional decoupled decoder to undertake the mask reconstruction task solely, reducing the learning complexity of pretext tasks and enhancing the encoder’s efficiency in extracting speech contextual representations. Owing to this innovation, by using only 4 GPUs, SVQ-NAE can achieve high performance comparable to wav2vec 2.0, which requires 64 GPUs for training. In the Speech Processing Universal Performance Benchmark, SVQ-MAE surpasses wav2vec 2.0 in both keyword spotting and emotion recognition tasks. Furthermore, in cross-lingual ASR for Mandarin, upon fine-tuning on AISHELL-1, SVQ-MAE achieves a Character Error Rate of 4.09%, outperforming all supervised ASR models.

Feature Selection as Deep Sequential Generative Learning

Feature selection aims to identify the most pattern-discriminative feature subset. In prior literature, filter (e.g., backward elimination) and embedded (e.g., LASSO) methods have hyperparameters (e.g., top- k , score thresholding) and tie to specific models, thus, hard to generalize; wrapper methods search a feature subset in a huge discrete space and is computationally costly. To transform the way of feature selection, we regard a selected feature subset as a selection decision token sequence and reformulate feature selection as a deep sequential generative learning task that distills feature knowledge and generates decision sequences. Our method includes three steps: (1) We develop a deep variational transformer model over a joint of sequential reconstruction, variational, and performance evaluator losses. Our model can distill feature selection knowledge and learn a continuous embedding space to map feature selection decision sequences into embedding vectors associated with utility scores. (2) We leverage the trained feature subset utility evaluator as a gradient provider to guide the identification of the optimal feature subset embedding; (3) We decode the optimal feature subset embedding to autoregressively generate the best feature selection decision sequence with autostop. Extensive experimental results show this generative perspective is effective and generic, without large discrete search space and expert-specific hyperparameters. The code is available at http://tinyurl.com/FSDSGL .

Generative learning for forecasting the dynamics of high-dimensional complex systems

We introduce generative models for accelerating simulations of high-dimensional systems through learning and evolving their effective dynamics. In the proposed Generative Learning of Effective Dynamics (G-LED), instances of high dimensional data are down sampled to a lower dimensional manifold that is evolved through an auto-regressive attention mechanism. In turn, Bayesian diffusion models, that map this low-dimensional manifold onto its corresponding high-dimensional space, operate on batches of physics correlated, time sequences of data and capture the statistics of the system dynamics. We demonstrate the capabilities and drawbacks of G-LED in simulations of several benchmark systems, including the Kuramoto-Sivashinsky (KS) equation, two-dimensional high Reynolds number flow over a backward-facing step, and simulations of three-dimensional turbulent channel flow. The results demonstrate that generative learning offers new frontiers for the accurate forecasting of the statistical properties of high-dimensional systems at a reduced computational cost.

General Machine Learning Approaches for Lithium-Ion Battery Capacity Fade Compared to Empirical Models

Today’s growing demand for lithium-ion batteries across various industrial sectors has introduced a new concern: battery aging. This issue necessitates the development of tools and models that can accurately predict battery aging. This study proposes a general framework for constructing battery aging models using machine learning techniques and compares these models with two existing empirical models, including a commercial one. To build the models, the databases produced by EVERLASTING and Bills et al. were utilized. The aim is to create universally applicable models that can address any battery-aging scenario. In this study, three types of models were developed: a vanilla neural network, a neural network inspired by extreme learning machines, and an encoder coupled with a neural network. The inputs for these models are derived from established knowledge in battery science, allowing the models to capture aging effects across different use cases. The models were trained on cells subjected to specific aging conditions and they were tested on other cells from the same database that experienced different aging conditions. The results obtained during the test for the vanilla neural network showed an RMSE of 1.3% on the Bills et al. test data and an RMSE of 2.7% on the EVERLASTING data, demonstrating similar or superior performance compared to the empirical models and proving the ability of the models to capture battery aging.

Implementing Vocabulary Journals through Digital Narrative Text for EFL Secondary Students in Learning General Vocabulary

Four fundamental language skills such as listening, speaking, reading, and writing are important skills for EFL secondary school students to develop it. Learning English requires fundamental knowledge of vocabulary since it influences these abilities greatly. However, vocabulary instruction is often overlooked in traditional language learning strategies. On the other hand, the use of digital narrative texts, such as videos, provides a modern approach to facilitating students in general vocabulary learning. This study explores the integration of vocabulary journals with digital narrative texts to enhance vocabulary acquisition among EFL secondary students specifically in junior high school students. This study used action research and involved 25 students from a junior high school in Karawang. Data collection methods included vocabulary journal logs, observations, interviews, and reflective journals. The findings revealed that vocabulary journals significantly assist students in understanding and retaining new vocabulary, enhancing overall language comprehension. Students expressed positive attitudes towards this method, highlighting its engaging and interactive learning strategy. This study concludes that combining vocabulary journals with digital narrative texts effectively supports vocabulary acquisition, making learning more enjoyable and impactful. This approach fosters deeper understanding and long-term retention of vocabulary, improving students' language skills and academic performance. The findings suggest that this innovative strategy could be beneficial in educational settings.