Objective. The growing number of recording sites of silicon-based probes means that an increasing amount of neural cell activities can be recorded simultaneously, facilitating the investigation of underlying complex n...
详细信息
Objective. The growing number of recording sites of silicon-based probes means that an increasing amount of neural cell activities can be recorded simultaneously, facilitating the investigation of underlying complex neural dynamics. In order to overcome the challenges generated by the increasing number of channels, highly automated signal processing tools are needed. Our goal was to build a spike sorting model that can perform as well as offline solutions while maintaining high efficiency, enabling high-performance online sorting. Approach. In this paper we present ELVISort, a deep learning method that combines the detection and clustering of different action potentials in an end-to-end fashion. Main results. The performance of ELVISort is comparable with other spike sorting methods that use manual or semi-manual techniques, while exceeding the methods which use an automatic approach: ELVISort has been tested on three independent datasets and yielded average F-1 scores of 0.96, 0.82 and 0.81, which comparable with the results of state-of-the-art algorithms on the same data. We show that despite the good performance, ELVISort is capable to process data in real-time: the time it needs to execute the necessary computations for a sample of given length is only 1/15.71 of its actual duration (i.e. the sampling time multiplied by the number of the sampling points). Significance. ELVISort, because of its end-to-end nature, can exploit the massively parallel processing capabilities of GPUs via deep learning frameworks by processing multiple batches in parallel, with the potential to be used on other cutting-edge AI-specific hardware such as TPUs, enabling the development of integrated, portable and real-time spike sorting systems with similar performance to offline sorters.
Three-dimensional (3D) microstructures are useful for studying the spatial structures and physical properties of porous media. A number of stochastic reconstructions are essential to discover the geometry and topology...
详细信息
Three-dimensional (3D) microstructures are useful for studying the spatial structures and physical properties of porous media. A number of stochastic reconstructions are essential to discover the geometry and topology of the porous media and the its flow behavior. While several deep-learning based generative models have been proposed to deal with the issue, the obstacles about stable training and difficulty in convergence limit the application of these models. To address these problems, a hybrid deep generative model for 3D porous media reconstruction is proposed. The hybrid model is composed of a variant autoencoder (VAE) and a generative adversarial network (GAN). It receives a two-dimensional image as input and generates 3D porous media. The encoder from VAE characterizes the statistical and morphological information of input image and generates a low-dimensional feature vector for generator. Benefiting from the hybrid model, the training becomes more stable and the generative capability is enhanced as well. Furthermore, a simple but useful loss function is used to help improve accuracy. The proposed model is tested on both isotropic and anisotropic porous media. The results show the synthetic realizations have good agreement to the targets on visual inspection, statistical functions and two-phase flow simulation.
Purpose: To investigate whether processing visual field (VF) measurements using a variational autoencoder (VAE) improves the structure-function relationship in glaucoma. Design: Cross-sectional study. Participants: Th...
详细信息
Purpose: To investigate whether processing visual field (VF) measurements using a variational autoencoder (VAE) improves the structure-function relationship in glaucoma. Design: Cross-sectional study. Participants: The training data consisted of 82 433 VF measurements from 16 836 eyes. The testing dataset consisted of 117 eyes of 75 patients with open-angle glaucoma. Methods: A VAE model to reconstruct the threshold of VF was developed using the training dataset. OCT and VF (Humphrey Field Analyzer 24-2, Swedish interactive threshold algorithm standard) measurements were carried out for all eyes in the testing dataset. Visual fields in the testing dataset then were reconstructed using the trained VAE. The structure-function relationship between the circumpapillary retinal nerve fiber layer (cpRNFL) thickness and VF sensitivity was investigated in each of twelve 30 degrees segments of the optic disc (3 nasal sectors were merged). Similarly, the structure-function relationship was investigated using the VAE-reconstructed VF. Main Outcome Measures: Structure-function relationship. Results: The corrected Akaike information criterion values with threshold were found to be smaller than the threshold reconstructed with the VAE (threshold(VAE)) in 9 of 10 sectors. A significant relationship was found between threshold and cpRNFL thickness in 6 of 10 sectors, whereas it was significant in 9 of 10 sectors with threshold(VAE). Conclusions: Applying VAE to VF data results in an improved structure-function relationship. (C) 2020 by the American Academy of Ophthalmology
Spatio-temporal modeling of wireless access latency is of great importance for connected-vehicular systems. The quality of the molded results rely heavily on the number and quality of samples which can vary significan...
详细信息
ISBN:
(纸本)9781728150895
Spatio-temporal modeling of wireless access latency is of great importance for connected-vehicular systems. The quality of the molded results rely heavily on the number and quality of samples which can vary significantly due to the sensor deployment density as well as traffic volume and density. This paper proposes LaMI (Latency Model Inpainting), a novel framework to generate a comprehensive spatio-temporal of wireless access latency of a connected vehicles across a wide geographical area. LaMI adopts the idea from image inpainting and synthesizing and can reconstruct the missing latency samples by a two-step procedure. In particular, it first discovers the spatial correlation between samples collected in various regions using a patching-based approach and then feeds the original and highly correlated samples into a variational autoencoder (VAE), a deep generative model, to create latency samples with similar probability distribution with the original samples. Finally, LaMI establishes the empirical PDF of latency performance and maps the PDFs into the confidence levels of different vehicular service requirements. Extensive performance evaluation has been conducted using the real traces collected in a commercial LTE network in a university campus. Simulation results show that our proposed model can significantly improve the accuracy of latency modeling especially compared to existing popular solutions such as interpolation and nearest neighbor-based methods.
Facial expression and identity are two independent yet intertwined components for representing a face. For facial expression recognition, identity can contaminate the training procedure by providing tangled but irrele...
详细信息
ISBN:
(纸本)9781509066315
Facial expression and identity are two independent yet intertwined components for representing a face. For facial expression recognition, identity can contaminate the training procedure by providing tangled but irrelevant information. In this paper, we propose to learn clearly disentangled and discriminative features that are invariant of identities for expression recognition. However, such disentanglement normally requires annotations of both expression and identity on one large dataset, which is often unavailable. Our solution is to extend conditional VAE to a crossed version named Cross-VAE, which is able to use partially labeled data to disentangle expression from identity. We emphasis the following novel characteristics of our Cross-VAE: (1) It is based on an independent assumption that the two latent representations' distributions are orthogonal. This ensures both encoded representations to be disentangled and expressive. (2) It utilizes a symmetric training procedure where the output of each encoder is fed as the condition of the other. Thus two partially labeled sets can be jointly used. Extensive experiments show that our proposed method is capable of encoding expressive and disentangled features for facial expression. Compared with the baseline methods, our model shows an improvement of 3.56% on average in terms of accuracy.
In this paper, we propose a practical privacy-preserving generative model for data sanitization and sharing, called Sanitizer-variational autoencoder (SVAE). We assume that the data consists of identification-relevant...
详细信息
ISBN:
(纸本)9783030620080;9783030620073
In this paper, we propose a practical privacy-preserving generative model for data sanitization and sharing, called Sanitizer-variational autoencoder (SVAE). We assume that the data consists of identification-relevant and irrelevant components. A variational autoencoder (VAE) based sanitization model is proposed to strip the identification-relevant features and only retain identification-irrelevant components in a privacy-preserving manner. The sanitization allows for task-relevant discrimination (utility) but minimizes the personal identification information leakage (privacy). We conduct extensive empirical evaluations on the real-world face, biometric signal and speech datasets, and validate the effectiveness of our proposed SVAE, as well as the robustness against the membership inference attack.
Artificial neural networks (ANNs) suffer from catastrophic forgetting, a sharp decrease in performance on previously learned tasks, when trained on a new task without constant rehearsal. In this paper, we propose a ne...
详细信息
ISBN:
(纸本)9781509066315
Artificial neural networks (ANNs) suffer from catastrophic forgetting, a sharp decrease in performance on previously learned tasks, when trained on a new task without constant rehearsal. In this paper, we propose a new method for overcoming this phenomenon based on one-class classification. It is not only able to incrementally learn new but also detect unknown classes. This is a desirable property, since it enables the detection of new and unknown classes in a stream of data and adaption to a changing environment. Experiments on commonly used continual learning setups show competitive results and verify the concept.
The conformation spaces (CS) of macromolecules and their associated dynamics are of vital importance in the understanding of many biochemical functions as well as diseases and in the developments of drugs for curing o...
详细信息
ISBN:
(纸本)9781728125473
The conformation spaces (CS) of macromolecules and their associated dynamics are of vital importance in the understanding of many biochemical functions as well as diseases and in the developments of drugs for curing or managing disease conditions. While the exploration of the CS is generally easier for molecules with fewer atoms (such as ligands and short peptides), achieving the same for larger molecules (such as nucleic acids and proteins) beyond a narrow local equilibrium is non-trivial and sometimes computationally prohibitive. In this work, we present Deep Enhanced Sampling of Nucleic Acids' Structures Using Deep-Learning-Derived Biasing Forces (DESNA, pronounced DES-na), that combines variational autoencoder, a special deep neural network (DNN), and molecular dynamics (MD) simulations to create a robust technique for enhanced sampling, in which DNN-learned latent space is used for inferring appropriate biasing potentials for guiding the MD simulations. The results obtained show that DESNA performs better than conventional MD simulations and efficiently samples wider CS than conventional MD simulations even when DESNA is allowed to run for as short as 10% of the length of conventional MD simulations. This suggests that DESNA is at least 10 times more efficient that conventional MD simulations in its sampling of CS of molecules.
Most existing image dehazing methods based learning are less able to perform well to real hazy images. An important reason is that they are trained on synthetic hazy images whose distribution is different from real h...
详细信息
Most existing image dehazing methods based learning are less able to perform well to real hazy images. An important reason is that they are trained on synthetic hazy images whose distribution is different from real hazy images. To relieve this issue, this paper proposes a new hazy scene generation model based on domain adaptation, which uses a variational autoencoder to encode the synthetic hazy image pairs and the real hazy images into the latent space to adapt. The synthetic hazy image pairs guide the model to learn the mapping of clear images to hazy images, the real hazy images are used to adapt the synthetic hazy images’ latent space to real hazy images through generative adversarial loss, so as to make the generative hazy images’ distribution as close to the real hazy images’ distribution as possible. By comparing the results of the model with traditional physical scattering models and Adobe Lightroom CC software, the hazy images generated in this paper is more realistic. Our end-to-end domain adaptation model is also very convenient to synthesize hazy images without depth map. Using traditional method to dehaze the synthetic hazy images generated by this paper, both SSIM and PSNR have been improved, proved that the effectiveness of our method. The non-reference haze density evaluation algorithm and other quantitative evaluation also illustrate the advantages of our method in synthetic hazy images.
Understanding in-network information diffusion is a fundamental problem in many application domains and one of the primary challenges is to predict the size of the information cascade. Most of the existing models rely...
详细信息
ISBN:
(纸本)9781728164120
Understanding in-network information diffusion is a fundamental problem in many application domains and one of the primary challenges is to predict the size of the information cascade. Most of the existing models rely either on hypothesized point process (e.g., Poisson and Hawkes process), or simply predict the information propagation via deep neural networks. However, they fail to simultaneously capture the underlying structure of a cascade graph and the propagation of uncertainty in the diffusion, which may result in unsatisfactory prediction performance. To address these, in this work we propose a novel probabilistic cascade prediction framework: variational Cascade (VaCas) graph learning networks. VaCas allows a non-linear information diffusion inference and models the information diffusion process by learning the latent representation of both the structural and temporal information. It is a pattern-agnostic model leveraging variational inference to learn the node-level and cascade-level latent factors in an unsupervised manner. In addition, VaCas is capable of capturing both the cascade representation uncertainty and node infection uncertainty, while enabling hierarchical pattern learning of information diffusion. Extensive experiments conducted on real-world datasets demonstrate that VaCas significantly improves the prediction accuracy, compared to state-of-the-art approaches, while also enabling interpretability.
暂无评论