In 3D inverse synthetic aperture radar (ISAR) imaging of space targets, radar neural radiance fields (i.e., ISARNeRF) is an important research direction. However, a significant yet unexplored problem is its deployment in continual learning scenarios, where gradual 3D imaging is expected since target ISAR images often emerge sequentially. To address this issue, this letter proposes a new continual 3D ISAR imaging method, named CL-ISAR-NeRF. Specifically, CL-ISAR-NeRF leverages a multilevel distillation mechanism to simultaneously replay pixel, field, and feature-levels information, to alleviate the forgetting of previously learned knowledge. In addition, an efficient memory selection strategy is designed to enrich the diversity of lineof-sight (LOS) when selecting replayed data, which improves imaging performance and further enhances the stability and plasticity of the method. In order to evaluate the proposed method in continual learning settings, we design a realistic simulation scenario in which the trajectories of space targets are calculated by the Simplified General Perturbations-4 (SGP4) model. The comparative experiments with classic continual learning methods demonstrate the superior performance and robustness of CLISAR-NeRF

Mixture models serve as one fundamental tool with versatile applications. However, their training techniques, like the popular Expectation Maximization (EM) algorithm, are notoriously sensitive to parameter initialization and often suffer from bad local optima that could be arbitrarily worse than the optimal. To address the long-lasting bad-local-optima challenge, we draw inspiration from the recent ground-breaking foundation models and propose to leverage their underlying big learning principle to upgrade the EM. Specifically, we present the Big Learning EM (BigLearn-EM), an EM upgrade that simultaneously performs joint, marginal, and orthogonally transformed marginal matchings between data and model distributions. Through simulated experiments, we empirically show that the BigLearn-EM is capable of delivering the optimal with high probability; comparisons on benchmark clustering datasets further demonstrate its effectiveness and advantages over existing techniques. The code is available at https://github.com/YulaiCong/Big-Learning-Expectation-Maximization.

Text generation is a key component of many natural language tasks. Motivated by the success of generative adversarial networks (GANs) for image generation, many text-specific GANs have been proposed. However, due to the discrete nature of text, these text GANs often use reinforcement learning (RL) or continuous relaxations to calculate gradients during learning, leading to high-variance or biased estimation. Furthermore, the existing text GANs often suffer from mode collapse (i.e., they have limited generative diversity). To tackle these problems, we propose a new text GAN model named text feature GAN (TFGAN), where adversarial learning is performed in a continuous text feature space. In the adversarial game, GPT2 provides the "true" features, while the generator of TFGAN learns from them. TFGAN is trained by maximum likelihood estimation on text space and adversarial learning on text feature space, effectively combining them into a single objective, while alleviating mode collapse. TFGAN achieves appealing performance in text generation tasks, and it can also be used as a flexible framework for learning text representations.

Aircraft detection is a valuable but challenging task in synthetic aperture radar (SAR) automatic target recognition (ATR). Because of the complicated electromagnetic imaging mechanism of SAR, the SAR image of aircraft appears as a distributed collection of discrete scattering points that varies significantly with imaging conditions, like different incident angles, bringing great challenges to existing CNN-based detection methods for accurate aircraft detection. To address these challenges, we analyze and leverage the scattering characteristics of multi-scale SAR aircraft to propose a novel SAR aircraft detector named deformable scattering feature correlation network (DSFCN). First, to deal with the discreteness of SAR aircraft, we propose a new Transformer-based backbone named scalable Swin Transformer backbone (SSTB) to replace a conventional CNN-based one, to effectively extract hierarchical scattering features from multi-scale aircraft. Second, to cope with the varying image appearance of SAR aircraft, we design a deformable region correlation module (DRCM) to flexibly correlate strong scattering regions that carry aircraft salient features. Various interpretable experiments conducted on a real-measured Gaofen-3 SAR aircraft dataset demonstrate the superiority and reliability of our DSFCN over other representative CNN-based methods.

In recent years, deep neural networks are increasingly popular in the field of radar high-resolution range profiles (HRRPs) target recognition. Unfortunately, recent researches have revealed that a deep-learning classifier can be easily fooled by adding small perturbations to the input, named adversarial attack. This provides us an inspiration for radar deceptive jamming signal generation in electronic countermeasures (ECMs). However, the perturbations generated by these adversarial attacks are usually of complex envelopes and quite low power, making it challenging for jammers to generate such actual jamming signals. To solve that issue, we propose a practical deceptive jamming generation method that learns the vulnerable range cells in an HRRP sample and injects several jamming pulses with specific amplitudes into these range cells. Such jamming signals are easy to generate and can deceive the radar automatic target recognition (RATR) model to output the wrong target category prediction with high confidence. To avoid the requirement of the recognition network structure information, we leverage the differential evolution optimization algorithm (non-gradientbased). Further, to provide the potential of real-time jamming signal generation during the test, an encoder is constructed not only to learn the separable features but also to find the vulnerable range cells and the specific amplitudes of the jamming pulses. In the experiments, we apply the proposed attack algorithms to fool the one-dimensional convolutional neural network-based HRRPRATR models. The extensive experimental results on measured aircraft HRRP dataset prove that the proposed algorithms achieve a promising attack performance and serve as a practical and fast deceptive jamming generation method.

To build a flexible and interpretable model for document analysis, we develop deep autoencoding topic model (DATM) that uses a hierarchy of gamma distributions to construct its multi-stochastic-layer generative network. In order to provide scalable posterior inference for the parameters of the generative network, we develop topic-layer-adaptive stochastic gradient Riemannian MCMC that jointly learns simplex-constrained global parameters across all layers and topics, with topic and layer specific learning rates. Given a posterior sample of the global parameters, in order to efficiently infer the local latent representations of a document under DATM across all stochastic layers, we propose a Weibull upward-downward variational encoder that deterministically propagates information upward via a deep neural network, followed by a Weibull distribution based stochastic downward generative model. To jointly model documents and their associated labels, we further propose supervised DATM that enhances the discriminative power of its latent representations. The efficacy and scalability of our models are demonstrated on both unsupervised and supervised learning tasks on big corpora.

Maximum likelihood (ML) and adversarial learning are two popular approaches for training generative models, and from many perspectives these techniques are complementary. ML learning encourages the capture of all data modes, and it is typically characterized by stable training. However, ML learning tends to distribute probability mass diffusely over the data space, e.g., yielding blurry synthetic images. Adversarial learning is well known to synthesize highly realistic natural images, despite practical challenges like mode dropping and delicate training. We propose an α-Bridge to unify the advantages of ML and adversarial learning, enabling the smooth transfer from one to the other via the α-divergence. We reveal that generalizations of the α-Bridge are closely related to approaches developed recently to regularize adversarial learning, providing insights into that prior work, and further understanding of why the α-Bridge performs well in practice.

Recent work has shown generative adversarial networks (GANs) can generate highly realistic images, that are often indistinguishable (by humans) from real images. Most images so generated are not contained in the training dataset, suggesting potential for augmenting training sets with GAN-generated data. While this scenario is of particular relevance when there are limited data available, there is still the issue of training the GAN itself based on that limited data. To facilitate this, we leverage existing GAN models pretrained on large-scale datasets (like ImageNet) to introduce additional knowledge (which may not exist within the limited data), following the concept of transfer learning. Demonstrated by natural-image generation, we reveal that low-level filters (those close to observations) of both the generator and discriminator of pretrained GANs can be transferred to facilitate generation in a perceptually-distinct target domain with limited training data. To further adapt the transferred filters to the target domain, we propose adaptive filter modulation (AdaFM). An extensive set of experiments is presented to demonstrate the effectiveness of the proposed techniques on generation with limited data.

As a fundamental issue in lifelong learning, catastrophic forgetting is directly caused by inaccessible historical data; accordingly, if the data (information) were memorized perfectly, no forgetting should be expected. Motivated by that, we propose a GAN memory for lifelong learning, which is capable of remembering a stream of datasets via generative processes, with \emph{no} forgetting. Our GAN memory is based on recognizing that one can modulate the ``style'' of a GAN model to form perceptually-distant targeted generation. Accordingly, we propose to do sequential style modulations atop a well-behaved base GAN model, to form sequential targeted generative models, while simultaneously benefiting from the transferred base knowledge. The GAN memory -- that is motivated by lifelong learning -- is therefore itself manifested by a form of lifelong learning, via forward transfer and modulation of information from prior tasks. Experiments demonstrate the superiority of our method over existing approaches and its effectiveness in alleviating catastrophic forgetting for lifelong classification problems. Code is available at \url{https://github.com/MiaoyunZhao/GANmemory_LifelongLearning}.

It is challenging to develop stochastic gradient based scalable inference for deep discrete latent variable models (LVMs), due to the difficulties in not only computing the gradients, but also adapting the step sizes to different latent factors and hidden layers. For the Poisson gamma belief network (PGBN), a recently proposed deep discrete LVM, we derive an alternative representation that is referred to as deep latent Dirichlet allocation (DLDA). Exploiting data augmentation and marginalization techniques, we derive a block-diagonal Fisher information matrix and its inverse for the simplex-constrained global model parameters of DLDA. Exploiting that Fisher information matrix with stochastic gradient MCMC, we present topic-layer-adaptive stochastic gradient Riemannian (TLASGR) MCMC that jointly learns simplex-constrained global parameters across all layers and topics, with topic and layer specific learning rates. State-of-the-art results are demonstrated on big data sets.

To infer multilayer deep representations of high-dimensional discrete and nonnegative real vectors, we propose an augmentable gamma belief network (GBN) that factorizes each of its hidden layers into the product of a sparse connection weight matrix and the nonnegative real hidden units of the next layer. The GBN's hidden layers are jointly trained with an upward-downward Gibbs sampler that solves each layer with the same subroutine. The gamma-negative binomial process combined with a layer-wise training strategy allows inferring the width of each layer given a fixed budget on the width of the first layer. Example results illustrate interesting relationships between the width of the first layer and the inferred network structure, and demonstrate that the GBN can add more layers to improve its performance in both unsupervisedly extracting features and predicting heldout data. For exploratory data analysis, we extract trees and subnetworks from the learned deep network to visualize how the very specific factors discovered at the first hidden layer and the increasingly more general factors discovered at deeper hidden layers are related to each other, and we generate synthetic data by propagating random variables through the deep network from the top hidden layer back to the bottom data layer.

To infer a multilayer representation of high-dimensional count vectors, we propose the Poisson gamma belief network (PGBN) that factorizes each of its layers into the product of a connection weight matrix and the nonnegative real hidden units of the next layer. The PGBN's hidden layers are jointly trained with an upward-downward Gibbs sampler, each iteration of which upward samples Dirichlet distributed connection weight vectors starting from the first layer (bottom data layer), and then downward samples gamma distributed hidden units starting from the top hidden layer. The gamma-negative binomial process combined with a layer-wise training strategy allows the PGBN to infer the width of each layer given a fixed budget on the width of the first layer. The PGBN with a single hidden layer reduces to Poisson factor analysis. Example results on text analysis illustrate interesting relationships between the width of the first layer and the inferred network structure, and demonstrate that the PGBN, whose hidden units are imposed with correlated gamma priors, can add more layers to increase its performance gains over Poisson factor analysis, given the same limit on the width of the first layer.