Collapse Modes Research Articles

Wind-induced collapse mode and mechanism of super-large hyperbolic steel reticulated shell cooling tower with stiffening rings

With advancement of dry-air cooling technology, it has become feasible to use steel reticulated shell structures in design and construction of industrial cooling towers. Steel reticulated shell cooling towers (SRSCTs) with light weight, high strength, and rapid construction have development potential. Wind-induced collapse accidents in cooling towers frequently occur, characterized by pronounced nonlinear behavior involving significant deformation and material failure. Nevertheless, current stability and collapse analyses of cooling towers based on linear theory and elastic deformation are inadequate and unsafe. Thus, the wind-resistant stability of SRSCTs must be evaluated considering their inherent high flexibility and low damping levels. Considering a 220 m high hyperbolic SRSCT as a case study, a wind-induced collapse FEM simulation and analysis were performed. The average and fluctuating wind loads were measured in a reduced-scale model during wind tunnel tests and used as the time-variant wind pressure in a structural finite-element model. The dynamic performance and buckling collapse analyses of the cooling tower were conducted considering geometric and material nonlinearities. The entire wind-induced failure process of the SRSCT in extreme winds was systematically investigated. The weaknesses in the structural stiffness and the failure process were quantified, with five failure stages and four failure modes. The failure stages included linear elasticity, local member failure, local regional failure, overall structural failure, and collapse. The failure modes included circumferential plastic buckling zone failure, windward buckling failure of the shell, crosswind failure of the stiffening ring (hinged curved-beam mechanism), and windward failure of the stiffening ring (triangular arch-like mechanism). This study confirms that development of plastic buckling zones and formation of plastic hinges, considering geometric and material nonlinearity, are the dominant factor in wind-induced failure and collapse of SRSCTs.

Research on Spatiotemporal Continuous Information Perception of Overburden Compression-Tensile Strain Transition Zone during Mining and Integrated Safety Guarantee System.

The mining of deep underground coal seams induces the movement, failure, and collapse of the overlying rock-soil body, and the development of this damaging effect on the surface causes ground fissures and ground subsidence on the surface. To ensure safety throughout the life cycle of the mine, fully distributed, real-time, and continuous sensing and early warning is essential. However, due to mining being a dynamic process with time and space, the overburden movement and collapse induced by mining activities often have a time lag effect. Therefore, how to find a new way to resolve the issue of the existing discontinuous monitoring technology of overburden deformation, obtain the spatiotemporal continuous information of the overlying strata above the coal seam in real time and accurately, and clarify the whole process of deformation in the compression-tensile strain transition zone of overburden has become a key breakthrough in the investigation of overburden deformation mechanism and mining subsidence. On this basis, firstly, the advantages and disadvantages of in situ observation technology of mine rock-soil body were compared and analyzed from the five levels of survey, remote sensing, testing, exploration, and monitoring, and a deformation and failure perception technology based on spatiotemporal continuity was proposed. Secondly, the evolution characteristics and deformation failure mechanism of the compression-tensile strain transition zone of overburden were summarized from three aspects: the typical mode of deformation and collapse of overlying rock-soil body, the key controlling factors of deformation and failure in the overburden compression-tensile strain transition zone, and the stability evaluation of overburden based on reliability theory. Finally, the spatiotemporal continuous perception technology of overburden deformation based on DFOS is introduced in detail, and an integrated coal seam mining overburden safety guarantee system is proposed. The results of the research can provide an important evaluation basis for the design of mining intensity, emergency decisions, and disposal of risks, and they can also give important guidance for the assessment of ground geological and ecological restoration and management caused by underground coal mining.

Side-Scan Sonar Image Augmentation Method Based on CC-WGAN

The utilization of deep learning algorithms for side-scan sonar target detection is impeded by the restricted quantity and representativeness of side-scan sonar (SSS) samples. To address this issue, this paper proposes a method for image augmentation using a CC-WGAN network. First, the generator incorporates the Convolutional Block Attention Module (CBAM) to enhance the assimilation of global information and local features in the input images. This integration also improves stability and avoids mode collapse problems associated with the original Generative Adversarial Network. Subsequently, the CBAM is incorporated into the discriminator to facilitate a better understanding of the relevance and significance of input data, thereby enhancing the model’s generalization ability. Finally, based on this model, existing few-sample SSS images are augmented, and we utilize the augmented images for discrimination and detection with YOLOv5. The experimental results show that following training with the SSS dataset that is augmented by this network, the accuracy of target detection increased by 7.6%, validating the feasibility of our proposed method. This method presents a novel solution to the problem of low model accuracy in underwater target detection with side-scan sonar due to limited samples.

Local reinforcement of adhesively bonded square shape aluminum energy absorber with CFRP under quasi-static axial loading: An experimental and numerical study

This study presents experimental and numerical investigations on the effects of composite ply angle, thickness and number of layers on energy absorption behavior of adhesively bonded carbon fiber-reinforced plastic/aluminum hybrid structure under quasi-static axial loading. This paper aims to use the advantages of adhesive bonding for the local strengthening of a square shaped aluminum energy absorber by composite and as a result achieving desirable energy absorption behavior prior to failure in the connection and without heterogeneous deformation of the structure while reducing the weight of the structure and usage of material and at the same time obtaining favorable results compared to the traditional ways like reinforcing the aluminum structure with composite globally. For this purpose, two models were made and analyzed; one reinforced with four L-shaped composites adhesively bonded using Araldite 2015 to four outside corners of the aluminum square and the other reinforced locally to four corners from within the square. Finite element model was developed for analysis of these hybrid structures. Five different ply angles for composite and multiple number of composite layers varied from 2 to 8 layers under the same and different thicknesses were investigated. Moreover an alpha factor has been determined to measure the importance of local reinforement of the square shape aluminum energy absorber by L-shaped composites. The results showed that compared to aluminum and aluminum with global composite reinforcement energy absorbers, due to the adhesive connection between aluminum and composite in the locally reinforced CFRP/aluminum specimen, the composite had better energy absorption by following the collapse pattern of aluminum and creating a continuous collapse and failure modes. The results indicated good coordination and agreement between the simulated models and the experimental tests. The findings from this experiment demonstrates significant potential for local reinforcement of structures employing adhesives. Utilizing such connections and the ability to strengthen specific areas based on arbitrary geometry and strengthening location could offer substantial opportunities in constructing lightweight structures with high energy absorption across diverse applications.

Path signatures for diversity in probabilistic trajectory optimisation

Motion planning can be cast as a trajectory optimisation problem where a cost is minimised as a function of the trajectory being generated. In complex environments with several obstacles and complicated geometry, this optimisation problem is usually difficult to solve and prone to local minima. However, recent advancements in computing hardware allow for parallel trajectory optimisation where multiple solutions are obtained simultaneously, each initialised from a different starting point. Unfortunately, without a strategy preventing two solutions to collapse on each other, naive parallel optimisation can suffer from mode collapse diminishing the efficiency of the approach and the likelihood of finding a global solution. In this paper, we leverage on recent advances in the theory of rough paths to devise an algorithm for parallel trajectory optimisation that promotes diversity over the range of solutions, therefore avoiding mode collapses and achieving better global properties. Our approach builds on path signatures and Hilbert space representations of trajectories and connects parallel variational inference for trajectory estimation with diversity-promoting kernels. We empirically demonstrate that this strategy achieves lower average costs than competing alternatives on a range of problems, from 2D navigation to robotic manipulators operating in cluttered environments.

Fabrication and in-plane compressive collapse of CFRP honeycomb metamaterials

A pin-slot positioning method was proposed to fabricate carbon fiber reinforced polymer (CFRP) metamaterials of hexagonal and zero Poisson’s ratio semi-re-entrant honeycombs with the same mass of monolayer continuous twill-woven carbon fiber/epoxy prepregs. The in-plane compressive collapse of CFRP honeycombs was explored through experiments and finite element (FE) simulations. Furthermore, analytical models were developed to predict the modulus and initial collapse stress of the hexagonal and semi-re-entrant honeycombs. Good agreement is achieved between analytical predictions, FE simulations and experimental results. It is shown that the initial collapse of CFRP honeycombs is by the bending of cell walls. Once initial collapse has been attained, both hexagonal and semi-re-entrant honeycombs have a stress softening. The collapse and failure modes of CFRP honeycombs strongly depend on cellular configurations and loading directions.

Experimental investigation on damage development and failure mechanism of shield tunnel lining under internal blast considering stratum-structure interaction

With the expansion of international terrorism and the potential threat of attacks against civil infrastructure, the dynamic response and failure modes of underground tunnels under explosive loads have become a prominent research topic. The high cost and inherent danger associated with explosion experiments have limited current research on tunnel internal explosions, particularly in the context of scaled model tests of shield tunnels. This study presents a series of scaled model tests under 1 g-condition simulating internal blast events within a shield tunnel based on the prototype of the Shantou Bay Tunnel, considering the influences of surrounding stratum and equivalent explosive yield. Three different TNT explosive yields are considered in the model tests, namely 0.2, 0.4, and 1.0 kg. The model tests focus on the damage behavior and collapse modes of the shield tunnel lining under internal explosive loads. The model tests reveal that the shield tunnel is prone to damage at the joints of the tunnel crown and shoulder when subjected to internal explosive loads, with the upper half of the tunnel lining experiencing segment collapse, while the lower half remains largely undamaged. As the TNT equivalent increases, the damage area at the tunnel joints expands, and the number of segment failures in the upper half of the tunnel rises, transitioning from a damaged state to a collapsed state. The influence of “stratum-structure” interaction is investigated by comparing two models, one with overburden soil and the other positioned at the ground surface. The model tests reveal that the presence of soil pressure and confinement can significantly enhance the tunnel resistance to internal blast loads. Based on the observation of the model tests, five different damage modes of segment joints under internal explosion are proposed in this study.

Generative spatiotemporal image exploitation for datacenter traffic prediction

The tremendous growth rate of global internet traffic in past years increases the importance of traffic prediction for network operators to ensure seamless Quality of Service (QoS) with proactive traffic engineering. Even minor anomalies in traffic management can lead to service disruptions that affect a vast user base, necessitating highly accurate traffic predictions. While recent studies have exploited Deep Learning for accurate traffic predictions, most of them have targeted mobile network traffic and they often fall short in delivering precise long-range predictions and effective spatiotemporal feature extraction from single-stream time-series data. This research addresses these limitations by proposing a Convolutional Recurrent Generative Adversarial Network (CoRe-GAN) consisting of generator and discriminator neural networks for high-accuracy traffic prediction. The generator with Convolutional Long Short-Term Memory (ConvLSTM) model effectively captures intricate features, whereas the discriminator utilizes a Convolutional Neural Network (CNN) to train the generator through feedback. Moreover, advanced training techniques like fact forcing and feature matching increase the learning convergence rate, avoid mode collapse, and amplify prediction accuracy of CoRe-GAN. The evaluation with Pangyo Network Dataset (PND) and synthetic Intrusion Detection Dataset (IDD) confirms CoRe-GAN superiority. The results show that it outperforms ConvLSTM models with an average 20% and 16% lower Mean Square Error (MSE) with PND and IDD traffic data, respectively.

Destination intention estimation-based convolutional encoder-decoder for pedestrian trajectory multimodality forecast

Forecasting pedestrian trajectory is a vital area of research in smart urban mobility, which can be applied to intelligent transportation and intelligent surveillance. Current approaches employ conditional variational autoencoders to model future trajectory multimodality. However, these methods generate multi-modal trajectories for one single destination, ignoring the trajectory multimodality caused by the uncertainty of the pedestrians’ destination intention. Besides, they can lead to mode collapse and training instability. To address this issue, we propose a novel destination intention estimation-based convolutional encoder-decoder framework for multimodal trajectory forecast. Specially, we design a destination intention estimator to forecast pedestrian future destination intentions at the last time step. Then, we devise a trajectory decoder module to forecast pedestrian trajectories at each time step with the assistance of the destination intentions. To evaluate our method, we perform experiments on publicly available benchmark datasets and demonstrate that our proposed method achieves the superior results compared with state-of-the-art approaches.

DynGAN: Solving Mode Collapse in GANs With Dynamic Clustering.

Generative Adversarial Networks (GANs) are widely-used generative models for synthesizing complex and realistic data. However, mode collapse, where the diversity of generated samples is significantly lower than that of real samples, poses a major challenge for further applications. Our theoretical analysis demonstrates that the generator loss function is non-convex with respect to its parameters when there are multiple modes in real data. In particular, parameters that result in generated distributions with perfect partial mode coverage of the real distribution are the local minima of the generator loss function. To address mode collapse, we propose a unified framework called Dynamic GAN. This method detects collapsed samples in the generator by thresholding on observable discriminator outputs, divides the training set based on these collapsed samples, and trains a dynamic conditional model on the partitions. The theoretical outcome ensures progressive mode coverage and experiments on synthetic and real-world data sets demonstrate that our method surpasses several GAN variants. In conclusion, we examine the root cause of mode collapse and offer a novel approach to quantitatively detect and resolve it in GANs.

Reducing Mode Collapse With Monge-Kantorovich Optimal Transport for Generative Adversarial Networks.

Mode collapse has been a persisting challenge in generative adversarial networks (GANs), and it directly affects the applications of GAN in many domains. Existing works that attempt to solve this problem have some serious limitations: models using optimal transport (OT) strategies (e.g., Wasserstein distance) lead to vanishing or exploding gradients; increasing the number of generators can cause several generators focusing on the same mode; and approaches that modify the loss also do not satisfactorily resolve mode collapse. In this article, we reduce mode collapse by formulating it as a Monge problem of OT map. We show that the Monge problem can be transformed to the distribution transformation problem in GAN, and a rectified affine neural network can be considered as a measurable function. In this way, we propose Monge GAN that uses this measurable function to transform the generated data distribution into the original data distribution. We utilize the Kantorovich formulation to obtain the OT cost, which is regarded as the OT distance between the two distributions. Finally, we conduct extensive experiments on both image and numerical datasets to validate our Monge GAN in reducing model collapse.

Leveraging GANs data augmentation for imbalanced medical image classification

Machine learning algorithms have been widely applied and researched in the field of medical image classification. Most of the current algorithms are designed based on the assumption of data balance. However, in practical applications, medical data often face the problem of imbalanced data. The traditional data augmentation approaches can balance the original data, but they are unable to generate more effective features. Generative Adversarial Networks (GANs) can effectively perform sample augmentation. However, GAN often encounter issues such as intra-class mode collapse and generating noisy samples when faced with imbalanced data. This study proposes a GAN model called IBGAN that focuses on generating intra-class sparse samples and class boundary samples. Specifically, IBGAN addresses the issue of imbalanced data through two stages. In the first stage, the Isolation Forest (iForest) algorithm and boundary sample detection algorithm are designed to identify sparse region samples and boundary samples within the intra-class data. This enables the GAN model to effectively focus on generating such samples during training. The second stage involves the refinement process of the samples. We propose a sample evaluation method based on Support Vector Data Description (SVDD) to filter out noisy data in the generated samples, ensuring the quality of the generated data. We conducted extensive and in-depth experiments on five real-world datasets. The experimental results demonstrate that IBGAN can generate high-quality and diverse augmented samples, which contribute to the improvement of classifier performance. We compared our proposed method with seven baseline methods, including traditional approaches and classic GAN models. The experimental results, including the visualization of the generated data and the evaluation of classification performance, consistently show that our proposed method achieves more competitive results.

DSM design of cold-formed steel fixed-ended lipped channel columns undergoing distortional-global interaction

This work reports a numerical investigation dealing with the post-buckling behaviour, strength, and Direct Strength Method (DSM) design of cold-formed steel fixed-ended lipped channel columns experiencing coupling between distortional and global (flexural-torsional) buckling − distortional-global (D–G) interaction. Various cross-section dimensions and lengths are considered, to ensure that the columns selected undergo different levels and types (“true”, “secondary-bifurcation distortional” or “secondary-bifurcation global”) of D–G interaction. The results presented and discussed, determined by means of Abaqus shell finite element geometrically and materially non-linear analyses, consist of elastic and elastic-plastic post-buckling equilibrium paths, failure loads and collapse modes. The steel material behaviour is deemed elastic-perfectly plastic and several yield stresses are considered, thus making it possible to cover a wide column D–G slenderness range. Particular attention is devoted to identifying the most detrimental initial geometrical imperfection shapes, in the sense that they lead to the lowest column failure loads. The numerical failure load data gathered are subsequently used to assess the merits of the available DSM-based design approaches developed to handle cold-formed steel columns undergoing D–G interaction. Since these design approaches are shown to be either inefficient and/or improvable, the above failure load data are also used to propose modifications/improvements aimed at achieving an efficient failure load prediction in the specific context of fixed-ended lipped channel columns. The success of this endeavour provides encouragement to the authors in their search for a safe, accurate and reliable DSM-based design approach capable of handling cold-formed steel columns with arbitrary cross-section shapes and/or end support conditions that fail in D–G interactive modes.

Melanoma classification using generative adversarial network and proximal policy optimization.

In oncology, melanoma is a serious concern, often arising from DNA changes caused mainly by ultraviolet radiation. This cancer is known for its aggressive growth, highlighting the necessity of early detection. Our research introduces a novel deep learning framework for melanoma classification, trained and validated using the extensive SIIM-ISIC Melanoma Classification Challenge-ISIC-2020 dataset. The framework features three dilated convolution layers that extract critical feature vectors for classification. A key aspect of our model is incorporating the Off-policy Proximal Policy Optimization (Off-policy PPO) algorithm, which effectively handles data imbalance in the training set by rewarding the accurate classification of underrepresented samples. In this framework, the model is visualized as an agent making a series of decisions, where each sample represents a distinct state. Additionally, a Generative Adversarial Network (GAN) augments training data to improve generalizability, paired with a new regularization technique to stabilize GAN training and prevent mode collapse. The model achieved an F-measure of 91.836% and a geometric mean of 91.920%, surpassing existing models and demonstrating the model's practical utility in clinical environments. These results demonstrate its potential in enhancing early melanoma detection and informing more accurate treatment approaches, significantly advancing in combating this aggressive cancer.

Leveraging Conditional Generative Adversarial Networks (cGANs) for enhanced artistic creation: Exploring quality improvement and content control through conditional inputs

Conditional Generative Adversarial Networks (cGANs) have revolutionized digital art by enabling the creation of high-quality, artistically coherent images guided by conditional inputs. This paper examines key factors influencing the performance of cGANs, including the quality of training data, network architecture improvements, and loss function optimization. We introduce a mathematical model to quantify training data quality, emphasizing dataset diversity, data augmentation, and cleaning. Network architectural enhancements such as residual connections, attention mechanisms, and progressive growing are explored for their impact on image quality. Additionally, we discuss the integration of conditional inputs, such as labels and textual descriptions, for precise content control. Challenges in balancing realism with artistic expression, managing mode collapse, and interpreting conditional inputs are also addressed. This study provides a comprehensive framework for enhancing cGAN-generated artworks, offering insights into applications in personalized art generation, art restoration, and collaborative art projects.

Circular synthetic aperture radar sub‐aperture angle information complementation based on azimuth‐controllable generative adversarial network

AbstractA conditional generative adversarial network (CGAN) framework is proposed to address the issue of incomplete circular synthetic aperture radar (CSAR) azimuthal information due to motion errors. Specifically, the authors propose a novel CGAN architecture that can control the azimuth angle for arbitrary angle generation, capable of complementing missing CSAR sub‐aperture information. The network incorporates angular labels for various scenarios and integrates a dynamic region‐aware convolution (DRconv) module. Additionally, to counteract the common challenge of mode collapse in GAN training, a mode seeking regularisation technique is innovativrly introduced into the authors’ loss function. The efficacy of the proposed network is rigorously tested using both the MSTAR dataset and an X‐band SAR dataset. The results demonstrate that the authors’ network can generate high‐fidelity SAR images with controllable azimuths, closely resembling authentic images. Furthermore, the proposed method excels in complementing missing CSAR sub‐aperture information, effectively supplying the lost angular information due to motion errors. A new technical approach for SAR image generation is not only offered but it also has the potential to significantly expand SAR datasets. This advancement is expected to enhance the quality and utility of SAR imagery in applications such as surveillance, reconnaissance, and environmental monitoring.

Novel hybrid integrated Pix2Pix and WGAN model with Gradient Penalty for binary images denoising

This paper introduces a novel approach to image denoising that leverages the advantages of Generative Adversarial Networks (GANs). Specifically, we propose a model that combines elements of the Pix2Pix model and the Wasserstein GAN (WGAN) with Gradient Penalty (WGAN-GP). This hybrid framework seeks to capitalize on the denoising capabilities of conditional GANs, as demonstrated in the Pix2Pix model, while mitigating the need for an exhaustive search for optimal hyperparameters that could potentially ruin the stability of the learning process. In the proposed method, the GAN’s generator is employed to produce denoised images, harnessing the power of a conditional GAN for noise reduction. Simultaneously, the implementation of the Lipschitz continuity constraint during updates, as featured in WGAN-GP, aids in reducing susceptibility to mode collapse. This innovative design allows the proposed model to benefit from the strong points of both Pix2Pix and WGAN-GP, generating superior denoising results while ensuring training stability. Drawing on previous work on image-to-image translation and GAN stabilization techniques, the proposed research highlights the potential of GANs as a general-purpose solution for denoising. The paper details the development and testing of this model, showcasing its effectiveness through numerical experiments. The dataset was created by adding synthetic noise to clean images. Numerical results based on real-world dataset validation underscore the efficacy of this approach in image-denoising tasks, exhibiting significant enhancements over traditional techniques. Notably, the proposed model demonstrates strong generalization capabilities, performing effectively even when trained with synthetic noise.

MEvo-GAN: A Multi-Scale Evolutionary Generative Adversarial Network for Underwater Image Enhancement

Inunderwater imaging, achieving high-quality imagery is essential but challenging due to factors such as wavelength-dependent absorption and complex lighting dynamics. This paper introduces MEvo-GAN, a novel methodology designed to address these challenges by combining generative adversarial networks with genetic algorithms. The key innovation lies in the integration of genetic algorithm principles with multi-scale generator and discriminator structures in Generative Adversarial Networks (GANs). This approach enhances image details and structural integrity while significantly improving training stability. This combination enables more effective exploration and optimization of the solution space, leading to reduced oscillation, mitigated mode collapse, and smoother convergence to high-quality generative outcomes. By analyzing various public datasets in a quantitative and qualitative manner, the results confirm the effectiveness of MEvo-GAN in improving the clarity, color fidelity, and detail accuracy of underwater images. The results of the experiments on the UIEB dataset are remarkable, with MEvo-GAN attaining a Peak Signal-to-Noise Ratio (PSNR) of 21.2758, Structural Similarity Index (SSIM) of 0.8662, and Underwater Color Image Quality Evaluation (UCIQE) of 0.6597.

Physics constrained pedestrian trajectory prediction with probability quantification

Accurately understanding and predicting the future trajectories of pedestrians around autonomous vehicles is a major challenge. Due to the uncertainty and diversity of pedestrian trajectories, current research focuses on the multimodality of trajectory prediction. Currently, multimodal pedestrian prediction methods mainly use deep generative networks to generate mutually independent trajectories through independent sampling strategy, thus focusing on modes with a large number of samples. Moreover, deep generative networks are black-box models that are entirely data-driven and cannot effectively explain trajectory generation. In addition, current research lacks an effective measure for each trajectory. Therefore, in our paper, we propose a physics constrained pedestrian trajectory prediction with probability quantification. We divide multimodal trajectory prediction into two phases: trajectory generation phase and trajectory selection phase. During the trajectory generation phase, rather than directly predicting trajectories, we incorporate a differential constraint module to ensure that the generated trajectories adhere to pedestrian motion laws. This approach enhances the interpretability of the model prediction process. Subsequently, we generate a set of candidate trajectory proposals with specific correlations through the correlated sampling module. During the trajectory selection phase, a probability selection module is proposed to establish explicit probability distributions for the candidate trajectory proposals and to select the proposal with the highest probability as the final output. Extensive experiments on a public real-world pedestrian trajectory dataset show that our proposed model exhibits significant advantages over existing models in multimodal trajectory prediction, which not only effectively reduces the error but also mitigates mode collapse. Moreover, we provide probabilities for each trajectory to better capture the model uncertainty and provide more comprehensive information for downstream decision-making systems.

Hamiltonian quantum generative adversarial networks

We propose Hamiltonian quantum generative adversarial networks (HQuGANs) to learn to generate unknown input quantum states using two competing quantum optimal controls. The game-theoretic framework of the algorithm is inspired by the success of classical generative adversarial networks in learning high-dimensional distributions. The quantum optimal control approach not only makes the algorithm naturally adaptable to the experimental constraints of near-term hardware, but also offers a more natural characterization of overparameterization compared to the circuit model. We numerically demonstrate the capabilities of the proposed framework to learn various highly entangled many-body quantum states, using simple two-body Hamiltonians and under experimentally relevant constraints such as low-bandwidth controls. We analyze the computational cost of implementing HQuGANs on quantum computers and show how the framework can be extended to learn quantum dynamics. Furthermore, we introduce a cost function that circumvents the problem of mode collapse that prevents convergence of HQuGANs and demonstrate how to accelerate the convergence of them when generating a pure state. Published by the American Physical Society 2024