Due to the fixed weights and simple topology of traditional unidirectional ring neural networks, it is challenging to generate complex firing dynamics. This paper introduces two threshold memristor models to mimic neuronal autapses, constructing a novel dual memristor-coupled unidirectional ring neural network (DMCURNN). Benefiting from the dual memristor-based autapse emulation, the DMCURNN enriches diverse hidden firing dynamics that are absent in conventional unidirectional ring neural networks, and the richness and complexity of these hidden firing dynamics are highly sensitive to the memristive coupling strengths. Numerical simulations indicate that the DMCURNN is able to generate various hidden firing activities, pattern transitions, homogeneous firing multistability of infinite coexisting homogeneous firing patterns and heterogeneous firing multistability of seven heterogeneous firing patterns. Furthermore, through adjusting the coupling parameters along with the external current intensity, the regulatory mechanisms of hidden firing patterns and their amplitude and frequency regulation are revealed. Subsequently, the hidden firing dynamics of this network are verified on an FPGA hardware platform. Finally, Using the hidden firing multistability characteristic of the DMCURNN, a novel bitstream-based hardware image encryption scheme with a dynamic key update mechanism is designed. Performance evaluations demonstrate that the proposed scheme exhibits exceptional robustness and high efficiency for IoT security applications.

Memristors possess unique characteristics, including nano-scale dimensions, non-volatility, programmability and synaptic plasticity, enabling them to simulate biological neuronal synapses. Despite existing memristor-synaptic coupled neuron models demonstrating diverse firing patterns under varying initial conditions, the diversity of their multistable firing behaviors remains limited. To address this issue, a new multistable ReLU-type memristor with a piecewise linear function is proposed to couple FitzHugh–Nagumo (FHN) and Hindmarsh–Rose (HR) neurons, establishing a novel multistable ReLU-type memristive heterogeneous neuron model (MRMHNM). Numerical calculations indicate that this MRMHNM displays rich firing dynamics, including periodic spiking firing, chaotic bursting firing, and multi-scroll firing with adjustable scroll numbers. Significantly, it demonstrates distinct firing multistability, encompassing the homogeneous multistability of initial offset-boosted coexisting single-scroll firings, as well as peculiar behaviors in both homogeneous and heterogeneous coexisting scenarios, where the amount of the heterogeneous and homogeneous firings can theoretically increase without bound. In addition, pattern transitions, phase synchronization and geometrically controllable firing behaviors are also comprehensively revealed and investigated. Moreover, a digital hardware platform based on the STM32 microcontroller is employed to implement and verify these intriguing discoveries. Lastly, using the distinctive multistability and multi-scroll firing dynamics of the MRMHNM, a new image secure communication system is built, and experimental results confirm its excellent reliability and security performance.

Large vision Transformers (ViTs) have achieved competitive performance in various computer vision tasks based on large-scale pre-training. However, large ViTs still remain vulnerable to adversarial examples, emphasizing the necessity of enhancing their adversarial robustness. While adversarial training is an effective defense for deep convolutional models, it often faces scalability issues with large ViTs due to high computational costs. Recent approaches propose robust fine-tuning methods, such as adversarial tuning of low-rank adaptation (LoRA) in ViT, however, they still struggle to match the accuracy of full parameter adversarial fine-tuning. An effective synergy of various defense mechanisms offers a promising approach to enhancing the robustness of ViT, yet this paradigm remains largely underexplored. To address this, we propose hyper adversarial tuning (HyperAT), a meta learning approach, which captures shared defensive knowledge among different methods to improve model robustness efficiently and effectively simultaneously. Specifically, adversarial tuning of each defense method is formulated as a learning task, and a HyperNetwork generates LoRA specific to this defense. Then, a random sampling and tuning strategy is proposed to extract and facilitate the defensive knowledge transfer between different defenses. Finally, diverse LoRAs are merged adaptively to further enhance the adversarial robustness. Experiments on various datasets and model architectures demonstrate that HyperAT significantly enhances the adversarial robustness of pretrained large vision models without excessive computational overhead, establishing a new state-of-the-art benchmark.

Continuous Conditional Generative Modeling(CCGM) estimates high-dimensional data distributions, such as images, conditioned on scalar continuous variables (aka regression labels). WhileContinuous Conditional Generative Adversarial Networks(CcGANs) were designed for this task, their instability during adversarial learning often leads to suboptimal results.Conditional Diffusion Models(CDMs) offer a promising alternative, generating more realistic images, but their diffusion processes, label conditioning, and model fitting procedures are either not optimized for or incompatible with CCGM, making it difficult to integrate CcGANs' vicinal approach. To address these issues, we introduceContinuous Conditional Diffusion Models(CCDMs), the first CDM specifically tailored for CCGM. CCDMs address existing limitations with specially designed conditional diffusion processes, a novel hard vicinal image denoising loss, a customized label embedding method, and efficient conditional sampling procedures. Through comprehensive experiments on four datasets with resolutions ranging from$64\times 64$to$192\times 192$, we demonstrate that CCDMs outperform state-of-the-art CCGM models, establishing a new benchmark. Ablation studies further validate the model design and implementation, highlighting that some widely used CDM implementations are ineffective for the CCGM task. Our code is publicly available athttps://github.com/UBCDingXin/CCDM.

The Hopfield neural network with unidirectional fixed resistance weights has been shown to exhibit limited complex dynamical behaviors due to its relatively simple architecture. To address this limitation, this paper proposes a new tri-memristor hyperchaotic ring neural network (THRNN). The THRNN facilitates the generation of hidden chaos and demonstrates homogeneous/heterogeneous multistability. Homogeneous coexisting attractors, when tightly connected across barriers, exhibit significant self-growth behavior over time. The number of growth directions can be freely regulated, and the multidirectional initial offset boosting characteristics of these growing attractors can also be readily observed. Furthermore, abundant hidden firing patterns are well-tuned by the coupling parameters of the memristors, resulting in chaotic bursting firing, periodic bursting firing, chaotic spiking firing, and periodic spiking firing. Particularly, a more complicated hidden hyperchaotic firing pattern is also discovered and captured. Moreover, an STM32H7 digital circuit is built to verify the findings presented in this paper. Finally, a hardware image blocking encryption system based on FPGA and the THRNN is proposed. This encryption system constructs a framework based on the hyperchaotic firing attractors and homogeneous multistability attractors. It realizes dynamic key update through block encryption strategy, and completes key scrambling by combining Cat mapping and sequence sorting, which significantly enhances encryption security. Relying on FPGA hardware implementation, its parallel processing capability greatly improves encryption efficiency, and the hardware deployment feature enhances the system’s stability and practicality, providing an efficient solution for high-security image encryption.

Continuous Conditional Generative Adversarial Networks (CcGANs) enable generative modeling conditional on continuous scalar variables (termed regression labels). However, they can produce subpar fake images due to limited training data. Although Negative Data Augmentation (NDA) effectively enhances unconditional and class-conditional GANs by introducing anomalies into real training images, guiding the GANs away from low-quality outputs, its impact on CcGANs is limited, as it fails to replicate negative samples that may occur during the CcGAN sampling. We present a novel NDA approach called Dual-NDA specifically tailored for CcGANs to address this problem. Dual-NDA employs two types of negative samples: visually unrealistic images generated from a pre-trained CcGAN and label-inconsistent images created by manipulating real images' labels. Leveraging these negative samples, we introduce a novel discriminator objective alongside a modified CcGAN training algorithm. Empirical analysis on UTKFace and Steering Angle reveals that Dual-NDA consistently enhances the visual fidelity and label consistency of fake images generated by CcGANs, exhibiting a substantial performance gain over the vanilla NDA. Moreover, by applying Dual-NDA, CcGANs demonstrate a remarkable advancement beyond the capabilities of state-of-the-art conditional GANs and diffusion models, establishing a new pinnacle of performance. Our codes can be found at https://github.com/UBCDingXin/Dual-NDA.

This article focuses on conditional generative modeling (CGM) for image data with continuous, scalar conditions (termed regression labels). We propose the first model for this task which is called continuous conditional generative adversarial network (CcGAN). Existing conditional GANs (cGANs) are mainly designed for categorical conditions (e.g., class labels). Conditioning on regression labels is mathematically distinct and raises two fundamental problems: (P1) since there may be very few (even zero) real images for some regression labels, minimizing existing empirical versions of cGAN losses (a.k.a. empirical cGAN losses) often fails in practice; and (P2) since regression labels are scalar and infinitely many, conventional label input mechanisms (e.g., combining a hidden map of the generator/discriminator with a one-hot encoded label) are not applicable. We solve these problems by: (S1) reformulating existing empirical cGAN losses to be appropriate for the continuous scenario; and (S2) proposing a naive label input (NLI) mechanism and an improved label input (ILI) mechanism to incorporate regression labels into the generator and the discriminator. The reformulation in (S1) leads to two novel empirical discriminator losses, termed the hard vicinal discriminator loss (HVDL) and the soft vicinal discriminator loss (SVDL) respectively, and a novel empirical generator loss. Hence, we propose four versions of CcGAN employing different proposed losses and label input mechanisms. The error bounds of the discriminator trained with HVDL and SVDL, respectively, are derived under mild assumptions. To evaluate the performance of CcGANs, two new benchmark datasets (RC-49 and Cell-200) are created. A novel evaluation metric (Sliding Fréchet Inception Distance) is also proposed to replace Intra-FID when Intra-FID is not applicable. Our extensive experiments on several benchmark datasets (i.e., RC-49, UTKFace, Cell-200, and Steering Angle with both low and high resolutions) support the following findings: the proposed CcGAN is able to generate diverse, high-quality samples from the image distribution conditional on a given regression label; and CcGAN substantially outperforms cGAN both visually and quantitatively.

Deep learning based image classification models are shown vulnerable to adversarial attacks by injecting deliberately crafted noises to clean images. To defend against adversarial attacks in a training-free and attack-agnostic manner, this work proposes a novel and effective reconstruction-based defense framework by delving into deep image prior (DIP). Fundamentally different from existing reconstruction-based defenses, the proposed method analyzes and explicitly incorporates the model decision process into our defense. Given an adversarial image, firstly we map its reconstructed images during DIP optimization to the model decision space, where cross-boundary images can be detected and on-boundary images can be further localized. Then, adversarial noise is purified by perturbing on-boundary images along the reverse direction to the adversarial image. Finally, on-manifold images are stitched to construct an image that can be correctly predicted by the victim classifier. Extensive experiments demonstrate that the proposed method outperforms existing state-of-the-art reconstruction-based methods both in defending white-box attacks and defense-aware attacks. Moreover, the proposed method can maintain a high visual quality during adversarial image reconstruction.