Zero-Shot Cross-Modal Retrieval (ZS-CMR) has recently drawn increasing attention as it focuses on a practical retrieval scenario, i.e, the multimodal test set consists of unseen classes that are disjoint with seen cla...
详细信息
ISBN:
(纸本)9781450387323
Zero-Shot Cross-Modal Retrieval (ZS-CMR) has recently drawn increasing attention as it focuses on a practical retrieval scenario, i.e, the multimodal test set consists of unseen classes that are disjoint with seen classes in the training set. The recently proposed methods typically adopt the generative model as the main framework to learn a joint latent embedding space to alleviate the modality gap. Generally, these methods largely rely on auxiliary semantic embeddings for knowledge transfer across classes and unconsciously neglect the effect of the data reconstruction manner in the adopted generative model. To address this issue, we propose a novel ZS-CMR model termed Multimodal Disentanglement variational autoencoders (MD-VAE), which consists of two coupled disentanglement variational autoencoders (DVAEs) and a fusion-exchange VAE (FVAE). Specifically, DVAE is developed to disentangle the original representations of each modality into modality-invariant and modality-specific features. FVAE is designed to fuse and exchange information of multimodal data by the reconstruction and alignment process without pre-extracted semantic embeddings. Moreover, an advanced counter-intuitive cross-reconstruction scheme is further proposed to enhance the informativeness and generalizability of the modality-invariant features for more effective knowledge transfer. The comprehensive experiments on four image-text retrieval and two image-sketch retrieval datasets consistently demonstrate that our method establishes the new state-of-the-art performance.
The rapid synthesis of radar waveform modulations is key to enabling a radar to react to the environment in order to optimize performance. This paper proposes the use of generative models for radar waveform generation...
详细信息
ISBN:
(纸本)9781728153681
The rapid synthesis of radar waveform modulations is key to enabling a radar to react to the environment in order to optimize performance. This paper proposes the use of generative models for radar waveform generation. Specifically, variational autoencoders (VAEs) comprising neural networks that are trained with a novel reconstruction loss are proposed. It is shown for simple classes of non-linear FM waveforms that the decoder from the proposed VAE can generate new radar waveform modulations that possess required ambiguity function characteristics, even though they were not represented in the training data.
To enhance flexibility and facilitate resource cooperation, a novel fully-decoupled radio access network (FD-RAN) architecture is proposed for 6G. However, the decoupling of uplink (UL) and downlink (DL) in FD-RAN mak...
详细信息
To enhance flexibility and facilitate resource cooperation, a novel fully-decoupled radio access network (FD-RAN) architecture is proposed for 6G. However, the decoupling of uplink (UL) and downlink (DL) in FD-RAN makes the existing feedback mechanism ineffective. To this end, we propose an end-to-end data-driven MIMO solution without the conventional channel feedback procedure. Data-driven MIMO can alleviate the drawbacks of feedback including overheads and delay, and can provide customized precoding design for different BSs based on their historical channel data. It essentially learns a mapping from geolocation to MIMO transmission parameters. We first present a codebook-based approach, which selects transmission parameters from the statistics of discrete channel state information (CSI) values and utilizes nearest neighbor interpolation for spatial inference. We further present a non-codebook-based approach, which 1) derives the optimal precoder from the singular value decomposition (SVD) of the channel;2) utilizes variational autoencoder (VAE) to select the representative precoder from the latent Gaussian representations;and 3) exploits Gaussian process regression (GPR) to predict unknown precoders in the space domain. Extensive simulations are performed on a link-level 5G simulator using realistic ray-tracing channel data. The results demonstrate the effectiveness of data-driven MIMO, showcasing its potential for application in FD-RAN and 6G.
While vast amounts of personal data are shared daily on public online platforms and used by companies and analysts to gain valuable insights, privacy concerns are also on the rise: Modern authorship attribution techni...
详细信息
ISBN:
(纸本)9781450390965
While vast amounts of personal data are shared daily on public online platforms and used by companies and analysts to gain valuable insights, privacy concerns are also on the rise: Modern authorship attribution techniques have proven effective at identifying individuals from their data, such as their writing style or behavior of picking and judging movies. It is hence crucial to develop data sanitization methods that allow sharing of users' data while protecting their privacy and preserving quality and content of the original data. In this paper, we tackle anonymization of textual data and propose an end-to-end differentially private variational autoencoder architecture. Unlike previous approaches that achieve differential privacy on a per-word level through individual perturbations, our solution works at an abstract level by perturbing the latent vectors that provide a global summary of the input texts. Decoding an obfuscated latent vector thus not only allows our model to produce coherent, high-quality output text that is human-readable, but also results in strong anonymization due to the diversity of the produced data. We evaluate our approach on IMDb movie and Yelp business reviews, confirming its anonymization capabilities and preservation of the semantics and utility of the original sentences.
The de novo design of drug molecules is recognized as a time-consuming and costly process, and computational approaches have been applied in each stage of the drug discovery pipeline. variational autoencoder is one of...
详细信息
ISBN:
(纸本)9783981926361
The de novo design of drug molecules is recognized as a time-consuming and costly process, and computational approaches have been applied in each stage of the drug discovery pipeline. variational autoencoder is one of the computer-aided design methods which explores the chemical space based on an existing molecular dataset. Quantum machine learning has emerged as an atypical learning method that may speed up some classical learning tasks because of its strong expressive power. However, near-term quantum computers suffer from limited number of qubits which hinders the representation learning in high dimensional spaces. We present a scalable quantum generative autoencoder (SQ-VAE) for simultaneously reconstructing and sampling drug molecules, and a corresponding vanilla variant (SQ-AE) for better reconstruction. The architectural strategies in hybrid quantum classical networks such as, adjustable quantum layer depth, heterogeneous learning rates, and patched quantum circuits are proposed to learn high dimensional dataset such as, ligand-targeted drugs. Extensive experimental results are reported for different dimensions including 8x8 and 32x32 after choosing suitable architectural strategies. The performance of quantum generative autoencoder is compared with the corresponding classical counterpart throughout all experiments. The results show that quantum computing advantages can be achieved for normalized low-dimension molecules, and that high-dimension molecules generated from quantum generative autoencoders have better drug properties within the same learning period.
It is proposed to use analogs of the forecast mean to generate an ensemble of perturbations for use in ensemble optimal interpolation (EnOI) or ensemble variational (EnVar) methods. A new method of constructing analog...
详细信息
It is proposed to use analogs of the forecast mean to generate an ensemble of perturbations for use in ensemble optimal interpolation (EnOI) or ensemble variational (EnVar) methods. A new method of constructing analogs using variational autoencoders (VAEs;a machine learning method) is proposed. The resulting analog methods using analogs from a catalog (AnEnOI), and using constructed analogs (cAnEnOI), are tested in the context of a multiscale Lorenz-'96 model, with standard EnOI and an ensemble square root filter for comparison. The use of analogs from a modestly-sized catalog is shown to improve the performance of EnOI, with limited marginal improvements resulting from increases in the catalog size. The method using constructed analogs (cAnEnOI) is found to perform as well as a full ensemble square root filter, and to be robust over a wide range of tuning parameters.
Given the growing number of volatile energy producers and consumers and the limitations of traditional static load prediction models, we have analyzed the ability of neural networks to predict the loads of an electric...
详细信息
ISBN:
(数字)9781665499965
ISBN:
(纸本)9781665499965
Given the growing number of volatile energy producers and consumers and the limitations of traditional static load prediction models, we have analyzed the ability of neural networks to predict the loads of an electrical transformer and to understand the physical relationships in the electric grid. To do this, we use a variational autoencoder to learn the load behavior of a neighborhood with three houses. Since a variational autoencoder learns a latent representation, we analyzed the possibility of learning physical relationships of the electric grid. By adapting the prediction model to learn a physical variable, namely phase shift, we show that a variational autoencoder can learn physical relations. Our results show a significant improvement in terms of the correlation of the latent and physical variables by integrating prior knowledge in the form of the corresponding power values as the training objective.
With the development of Industry 4.0, more and more attention has been paid to system intelligent maintenance by various industries, among which rolling bearing is an indispensable and most important component. Existi...
详细信息
Civil structures may deteriorate during their service life due to degradation or damage imposed by natural hazards such as earthquakes, wind, and impact. Structural performance anomaly detection is essential to provid...
详细信息
Civil structures may deteriorate during their service life due to degradation or damage imposed by natural hazards such as earthquakes, wind, and impact. Structural performance anomaly detection is essential to provide an early warning of structural degradation limit states in order to prevent potential catastrophic failure. Data-driven machine learning approaches have been widely used for this, due to their capability in capturing features sensitive to damage-induced anomalies from structural health monitoring (SHM) data, assuming that such data are available. Although machine learning models have been used, many are challenged by the vast operational and environmental variability that can corrupt SHM data and by (typically) strongly correlated information from different sensors in the SHM data. This paper proposes an unsupervised deep learning approach for the detection of structural anomaly based on a deep convolutional variational autoencoder (DCVAE) for feature extraction coupled with support vector data description (SVDD) for anomaly detection. The proposed DCVAE-SVDD method has several appealing strengths. First, the variational latent encoding is used to capture the features of monitoring data through a probability distribution. The integration of the Kullback-Leibler divergence in the loss function provides accurate estimation of the probability distributions. Second, the DCVAE designed with convolutional and deconvolutional operations utilizes the correlation among multisensor data to avoid loss of correlation features and achieve better performance in feature extraction. Third, the SVDD is utilized to create a minimum-volume hypersphere that contains the anomaly-sensitive statistical features of the state. The hypersphere accurately separates anomaly-sensitive statistical features of reference states of structure from the anomalous ones. A computational frame model and a laboratory grandstand model are used to evaluate the performance of the proposed meth
We propose a novel approach to structure-aware topology optimization (SATO) to generate physically plausible multi-component structures with diverse stylistic variations. Traditional TO methods often operate within a ...
详细信息
We propose a novel approach to structure-aware topology optimization (SATO) to generate physically plausible multi-component structures with diverse stylistic variations. Traditional TO methods often operate within a discrete voxel-defined design space, overlooking the underlying structure-aware, which limits their ability to accommodate stylistic design preferences. Our approach leverages variational autoencoders (VAEs) to encode both geometries and corresponding structures into a unified latent space, capturing part arrangement features. The design target is carefully formulated as a topology optimization problem taking the VAE code as design variables under physical constraints, and solved numerically via analyzing the associated sensitivity with respect to the VAE variables. Our numerical examples demonstrate the ability to generate lightweight structures that balance geometric plausibility and structural performance with much enhanced stiffness that outperforms existing generative techniques. The method also enables the generation of diverse and reliable designs, maintaining structural integrity throughout, via a direct smooth interpolation between the optimized designs. The findings highlight the potential of our approach to bridge the gap between generative design and physics-based optimization by incorporating deep learning techniques.
暂无评论