Event-based object detection aims to directly recognize and localize objects from asynchronous event streams, facilitating reliable perception in scenarios with fast motion and complex lighting conditions. Its performance largely depends on how temporal information is organized during event sampling. In such dynamic scenarios, fixed sampling strategies struggle to adapt to temporal variations in event streams, whereas recent learnable approaches often exhibit limitations in either their coupling with downstream detection objectives or their ability to fully exploit temporal cues. In this paper, we propose MAS-SNN, a memory-augmented Spiking Neural Network (SNN) framework for event-based object detection, which formulates event sampling as a task-driven, learnable process tightly integrated with representation learning. MAS-SNN introduces a Spiking Memory Attention Embedding (SMAE), composed of a Spiking Memory Buffer (SMB) and a Spiking Memory Attention (SMA) mechanism, to preserve and selectively reuse historical spiking states, thereby fully leveraging the intrinsic temporal dynamics of spiking neurons while maintaining the inherent sparsity and asynchronicity of event-driven vision. MAS-SNN achieves superior performance on N-Caltech 101 and Gen1, demonstrating the effectiveness of memory augmented spiking sampling for event-based object detection.

As the third generation of neural networks, Spiking Neural Networks (SNNs) have demonstrated remarkable potential across diverse applications owing to their unique temporal dynamics. In recent years, analyzing the robustness of SNNs from a temporal perspective has become an emerging research focus. However, most existing works examine only the overall temporal behavior of SNNs, typically applying adversarial attacks that rely on time-averaged gradients.In this study, we revisit SNN robustness through the lens of temporal granularity, emphasizing the distinct behaviors that occur at individual time steps. We first introduce a Temporal Granularity Attack (TG-Attack), which selectively perturbs gradients at specific time steps. This approach enables a finer-grained evaluation of SNN robustness across time and demonstrates higher attack success rates than traditional gradient-averaging methods.Furthermore, we theoretically show that the robustness of SNNs at a given time step is determined by the Hessian of the input–output gradient at that step, which we define as Temporal Sensitivity (TS). By calculating the Temporal Sensitivity Value (TSV) for each time step, robustness can be effectively estimated without generating adversarial examples. Finally, we propose a Temporal Granularity Regularization (TG-Reg) term that constrains the TSV across all time steps, thereby improving the model’s overall robustness. Experimental evaluations confirm that our framework consistently outperforms existing state-of-the-art methods.

Abstract Whole-body intelligent locomotion systems face persistent challenges of redundant actuation and poor energy efficiency, limiting real-world deployment. Bio-inspired central pattern generators offer a promising framework for rhythmic control, yet hardware implementations struggle to match the efficiency and adaptability of biological systems. Here, we introduce an in-situ spike-malleable artificial plateau neuron integrating a bistable plateau gate with a transient threshold-switch. The neuron generates amplitude-programmable rhythmic spike bursts, achieving energy-efficient, antagonistic activation of extensors and flexors via a scalable circuit comprising two paired units (plateau gate and threshold-switch). The design leverages distributed encoding for coordinated muscle control, operating at ultra-low energy dissipation (141.37 pJ/spike). An expanded four-unit circuit enhances dynamic spike malleability, enabling parallel processing for multi-joint coordination. On a quadruped robot (Unitree Go2), these distributed circuits directly drive joint-level proportional derivative controllers using the Gaussian-filtered rhythmic spikes, enabling energy-efficient trotting without centralized computation. Critically, the system achieves stable on-ground locomotion and demonstrates adaptive gait transitions in real-world environments. Our approach merges ultra-compact hardware with bio-inspired architecture, advancing neuromorphic systems for energy-efficient autonomous robotics.

In recent years, Recurrent Spiking Neural Networks (RSNNs) have shown promising potential in long-term temporal modeling. Many studies focus on improving neuron models and also integrate recurrent structures, leveraging their synergistic effects to improve the long-term temporal modeling capabilities of Spiking Neural Networks (SNNs). However, these studies often place an excessive emphasis on the role of neurons, overlooking the importance of analyzing neurons and recurrent structures as an integrated framework. In this work, we consider neurons and recurrent structures as an integrated system and conduct a systematic analysis of gradient propagation along the temporal dimension, revealing a challenging gradient vanishing problem. To address this issue, we propose the Skip Recurrent Connection (SRC) as a replacement for the vanilla recurrent structure, effectively mitigating the gradient vanishing problem and enhancing long-term temporal modeling performance. Additionally, we propose the Adaptive Skip Recurrent Connection (ASRC), a method that can learn the skip span of skip recurrent connection in each layer of the network. Experiments show that replacing the vanilla recurrent structure in RSNN with SRC significantly improves the model's performance on temporal benchmark datasets. Moreover, ASRC-SNN outperforms SRC-SNN in terms of temporal modeling capabilities and robustness.

Object-goal visual navigation requires robots to reason over semantic structure and act effectively under partial observability. Recent approaches based on object-level topological maps enable long-horizon navigation without dense geometric reconstruction, but their execution remains limited by the gap between global topological guidance and local perception-driven control. In particular, local decisions are made solely from the current egocentric observation, without access to information beyond the robot's field of view. As a result, the robot may persist along its current heading even when initially oriented away from the goal, moving toward directions that do not decrease the global topological distance. In this work, we propose IntentReact, an intent-conditioned object-centric navigation framework that introduces a compact interface between global topological planning and reactive object-centric control. Our approach encodes global topological guidance as a low-dimensional directional signal, termed intent, which conditions a learned waypoint prediction policy to bias navigation toward topologically consistent progression. This design enables the robot to promptly reorient when local observations are misleading, guiding motion toward directions that decrease global topological distance while preserving the reactivity and robustness of object-centric control. We evaluate the proposed framework through extensive experiments, demonstrating improved navigation success and execution quality compared to prior object-centric navigation methods.

Spiking neural networks (SNNs) have garnered significant attention in reinforcement learning tasks for their low power consumption. However, traditional spiking reinforcement learning (SRL) methods, which rely on local-connected encoding and fixed-threshold learning, struggle to capture the inter-dimensional correlations of input information within short timesteps, limiting the network’s expressive capacity. While increasing timesteps can significantly enhance performance, excessive timesteps result in substantial delays. In this paper, we propose an adaptive global-connected encoding method that directly captures global dependencies (such as long-range feature interactions), allowing for efficient encoding of input state information even within a single timestep. Additionally, threshold-learnable spiking neurons are introduced to effectively capture and distinguish subtle changes in input signals at each timestep, thereby enhancing the spatial-temporal state representation during spike information transmission. Extensive experimental results demonstrate that the proposed method achieves performance comparable to state-of-the-art algorithms using only a single timestep, while significantly reducing inference latency and energy consumption. When extended to multiple timesteps, our approach consistently outperforms existing methods, showing substantial improvements across eight continuous control tasks from OpenAI Gym.

Bottleneck states, which connect distinct regions of the state space, provide a principled and interpretable basis for constructing temporal abstractions in Hierarchical Reinforcement Learning (HRL). However, existing bottleneck identification methods primarily rely on topological analysis of the state-transition graph, limiting their scalability to high-dimensional or continuous domains. To address this challenge, we introduce Value Power Strength (VPS), a value function-based metric inspired by the analogy between the Bellman equation and Kirchhoff’s current law, to quantify bottleneck property via the diffusion of reward in Markov Decision Processes (MDPs). VPS is estimated efficiently using value functions learned from random reward signals and captures reward diffusion bottlenecks in both discrete and continuous state spaces. Leveraging VPS, we design options that guide agents toward or away from bottleneck regions. Experimental results on classic tabular domains, MiniGrid, and Atari 2600 games demonstrate that the VPS-based framework discovers semantically meaningful subgoals and substantially improves exploration efficiency.

Dynamic Vision Sensor (DVS) asynchronously records sparse events triggered by changes in pixel intensity, offering high temporal resolution and low latency. Existing frame-based methods process event data densely, violating its inherent sparsity and introducing computational redundancy. While asynchronous models preserve the event stream's native format, they often neglect spatial information, compromising their adaptability and efficiency. To address these limitations, we propose a Spatiotemporally Separated Sparse Network (S3Net) for efficient event stream encoding and learning. Specifically, we employ a learnable sparse encoding scheme to construct a voxel-structured representation that effectively extracts spatiotemporal relationships among event data. After that, we propose a dual-branch architecture to capture localized spatial dependencies and dynamic temporal patterns of event data. By explicitly decoupling spatial and temporal modeling, S3Net enables end-to-end asynchronous processing of variable-length event sequences, achieving both strong representational capacity and high computational efficiency. Experimental results on six event-based datasets demonstrate that S3Net achieves state-of-the-art performance. Compared to frame-based methods, it significantly reduces computational overhead and model complexity, while also outperforming existing asynchronous approaches in inference speed without compromising accuracy. Extensive experiments across six event-based datasets show that S3Net establishes new state-of-the-art performance. Our method reduces computational costs by 35% and model parameters by 27% compared to frame-based approaches, while delivering 1.58× faster inference than existing point-based methods at comparable accuracy levels.

Neuromorphic computing, inspired by the funda mental structures and principles of biological neural systems, has emerged as a promising paradigm for achieving energy efficient and robust artificial intelligence. However, effective temporal processing in dynamic scenarios remains a fundamental challenge, as neuromorphic systems must handle temporally varying signals and discriminate their underlying temporal structures over time, a capability that remains limited in existing spiking neural networks (SNNs). To address this challenge, we propose Temporal-structure Encoding spiking neural networks (TE-SNNs), a framework equipped with a Temporal-structure Encoding (TE) module that explicitly encodes salient rhythmic and non-rhythmic temporal structures. By leveraging TE to modulate neuronal dynamics, TE-SNNs substantially enhance the capability to discriminate various temporal structures in cognitive timing tasks. Extensive experiments on sequential benchmarks further reveal that this temporal discrimination supports accurate and robust temporal processing. Finally, we demonstrate the practical significance of TE-SNNs in a closed loop autonomous driving task, where they are implemented on neuromorphic hardware to generate optimized driving strategies with orders-of-magnitude improvements in energy efficiency.

The spiking federated learning (FL) is an emerging distributed learning paradigm that allows resource-constrained devices to train collaboratively at low power consumption without exchanging local data. It takes advantage of both the privacy computation property in FL and the energy efficiency in spiking neural networks (SNNs). However, existing spiking FL methods employ a random selection approach for client aggregation, assuming unbiased client participation. This neglect of statistical heterogeneity significantly affects the convergence and precision of the global model. In this work, we propose a credit assignment-based active client selection strategy for spiking federated learning, the SFedCA, to aggregate clients contributing to the global sample distribution balance judiciously. Specifically, the client credits are assigned by the firing intensity state before and after local model training, which reflects the difference in local data distribution from the global model. The comprehensive experiments are conducted on various non-identical and independent distribution (non-IID) scenarios. The experimental results demonstrate that the SFedCA outperforms the existing state-of-the-art spiking FL methods and requires fewer communication rounds.

The surrogate gradient (SG) method has shown significant promise in enhancing the performance of deep spiking neural networks (SNNs), but it also introduces vulnerabilities to adversarial attacks. Although spike coding strategies and neural dynamics parameters have been extensively studied for their impact on robustness, the critical role of gradient magnitude, which reflects the model's sensitivity to input perturbations, remains underexplored. In SNNs, the gradient magnitude is primarily determined by the interaction between the membrane potential distribution (MPD) and the SG function. In this study, we investigate the relationship between the MPD and SG and their implications for improving the robustness of SNNs. Our theoretical analysis reveals that reducing the proportion of membrane potentials lying within the gradient-available range of the SG function effectively mitigates the sensitivity of SNNs to input perturbations. Building upon this insight, we propose a novel MPD-driven surrogate gradient regularization (MPD-SGR) method, which enhances robustness by explicitly regularizing the MPD based on its interaction with the SG function. Extensive experiments across multiple image classification benchmarks and diverse network architectures confirm that the MPD-SGR method significantly enhances the resilience of SNNs to adversarial perturbations and exhibits strong generalizability across diverse network configurations, SG functions, and spike encoding schemes.

Parameter-Efficient Fine-Tuning (PEFT), which adapts pre-trained models to downstream tasks using only a limited number of trainable parameters, has attracted significant attention. Existing PEFT methods are predominantly based on Artificial Neural Networks (ANNs), yet they often incur substantial computational costs, hindering their deployment in resource-constrained scenarios. In contrast, Spiking Neural Networks (SNNs), a representative paradigm in neurocomputing, leverage sparse activations and event-driven operation to offer a promising low-power alternative. To leverage these advantages, we propose SpikeLoRA, a novel fine-tuning method that incorporates spike-driven processing to improve energy efficiency. To address the parameter update discrepancy between SNNs and ANNs, we further introduce a dedicated initialization strategy for SpikeLoRA that uses knowledge distillation to align its low-rank matrices with the pre-trained representations, effectively bridging SNN fine-tuning with the prior knowledge of ANNs. Extensive experiments show that SpikeLoRA achieves performance comparable to or better than standard LoRA and its variants, while significantly reducing power consumption. Compared to LoRA, SpikeLoRA reduces power consumption by 35% with only a 0.08% average performance drop on GLUE. On the Wikipedia dataset, SpikeLoRA achieves a 2.07% average improvement with the larger Llama-3.1-8B model. SpikeLoRA also exhibits stronger robustness, outperforming LoRA by an average of 6.27% on AdvGLUE. To the best of our knowledge, this is the first SNN-based variant of LoRA that successfully incorporates the energy efficiency and robustness of SNNs into fine-tuning ANN language models.

Spiking neural networks (SNNs) have garnered significant attention in reinforcement learning tasks for their low power consumption. However, traditional spiking reinforcement learning (SRL) methods, which rely on local-connected encoding and fixed-threshold learning, struggle to capture the inter-dimensional correlations of input information within short timesteps, limiting the network’s expressive capacity at low timesteps. While increasing timesteps can significantly enhance performance, excessive timesteps result in substantial delays. To address this contradiction and enhance the expressive and decision-making capabilities of SNNs within short timesteps, we propose Mask-Adaptive Global Connection (MAGC), a novel encoding method that efficiently captures long-range dependencies via sparse, adaptively masked connections—enabling global feature interaction in a single timestep. Additionally, dynamic-threshold spiking neurons are introduced to effectively capture and distinguish subtle changes in input signals at each timestep, thereby enhancing the spatial-temporal state representation during spike information transmission. Extensive experimental results demonstrate that the proposed method achieves performance comparable to state-of-the-art algorithms using only a single timestep, while significantly reducing inference latency and energy consumption. When extended to multiple timesteps, our approach consistently outperforms existing methods, showing substantial improvements across eight continuous control tasks from OpenAI Gym.

Recent brain decoding studies have primarily emphasized the development of brain decoders, while largely neglecting the segmentation step. Existing methods typically adopt fixed-length segmentation, which might overlook subject- or task-level variability and disrupt temporal patterns within brain signals. To address this gap, we propose S3, which leverages spiking neurons as an isolating segmenter for brain signal decoding. S3 segments brain signals adaptively, considering subject- and task-level variability while preserving intrinsic temporal patterns of brain signals. It exploits the unique reset mechanism of spiking neurons to isolate previous irrelevant temporal patterns during the generation of each segmentation point. To optimize S3 for enhancing task performance in the absence of segmentation labels, we develop an optimization method where segmentation pseudo-labels are created with a stochastic-greedy algorithm to optimize them, while circumventing gradient blockade between S3 and task performance. Experiments on 10 downstream tasks across 13 public datasets demonstrate that S3 consistently outperforms existing methods, validating its effectiveness, generalizability and interpretability.

Spiking Neural Networks (SNNs) have attracted increasing attention for their biologically inspired temporal dynamics. As their applications expand, understanding their robustness has become an important research focus. However, little is known about how the intrinsic temporal properties of SNNs affect robustness. In this work, we revisit SNN robustness from an information-theoretic perspective and reveal the pivotal role of temporal dynamics. We establish a theoretical link between robustness error and the mutual information (MI) between inputs and latent representations along the temporal dimension, grounded in the information bottleneck principle. Through an analysis of spike-based information transmission, we show that temporal dynamics inherently compress MI, thereby tightening the robustness error bound. Building on this insight, we propose a Temporal Mutual Information (TMI) regularizer that explicitly exploits temporal characteristics to enhance robustness. Extensive experiments on CIFAR-10, CIFAR-100, DVS-CIFAR10, and Tiny-ImageNet demonstrate that our method consistently improves SNN robustness across various architectures and attack settings.

Event cameras, designed to mimic biological perception principles, are a new type of visual sensor offering advantages such as low latency, low power consumption, and high dynamic range. However, the non-uniform, discontinuous nature of their spatially sparse yet temporally dense data presents significant challenges for extracting meaningful spatio-temporal features. Representing such novel data paradigms to provide high-quality inputs for neural network models remains a formidable task. Consequently, a key consideration is how to extract spatio-temporal features from event streams effectively. Currently, most work on event data typically performs feature extraction at either a single temporal or spatial scale, which often struggles to fully utilize the complex spatio-temporal relationships, leading to inaccurate capture of dynamic changes. Inspired by the brain's multi-scale processing mechanisms, we propose the Multi-scale Event Representation (MSER) model to enable a comprehensive understanding of spatio-temporal information. Among these, Temporal Multi-scale Encoding (TMSE) leverages a logarithmic perception algorithm with an adaptive decay factor to capture temporal features across multiple scales, effectively establishing temporal correlations. Then Spatial Multi-scale Fusion strategy (SMSF) is designed to address spatial variations by employing a feature pyramid structure with a bottom-up strategy to extract and integrate multi-scale spatial features. We validate the proposed method on a series of publicly available neuromorphic datasets, including N-CARS, N-MNIST, N-Caltech101, CIFAR10-DVS, and DvsGesture. Experimental results demonstrate that the features we extract exhibit excellent expressiveness and generalization, achieving state-of-the-art performance.

Why have modern artificial neural networks not adopted the nonlinear dendritic structures found in biological brain cells, and what is the core advantage of such active dendritic units? While early studies suggested that dendritic nonlinearities can enhance learning capabilities by boosting capacity, we provide empirical evidence reassessing this. Using extensive machine learning experiments, we show that dendritic nonlinearities in neural networks offer comparable learning capacity to standard point-neuron models when controlled for parametric complexity. Instead, we believe that their key advantage lies in enabling network scaling while substantially reducing communication costs via localized feature aggregation. Our experiments and analysis suggest that incorporating nonlinear dendritic architectures can significantly lower memory access or data transfer overhead during neural network inference—the primary sources of energy consumption in modern AI systems—and potentially during training as well. We argue that these insights motivate further theoretical and architectural exploration of dendritic-like structures in artificial neural networks.

Against the backdrop of the rapid advancement of Large Language Models (LLMs), their application in the field of psychology has garnered significant academic attention. A central issue is whether LLMs possess the capability to accurately perceive and measure complex, latent human psychological constructs, such as personality, emotions, and cognitive states. This paper provides a systematic review focused on the use of LLMs as instruments for human psychological measurement. To organize this domain, we propose a comprehensive analytical framework structured around three critical dimensions: Theoretical Plausibility (why measurement might be possible), Measurement Methodology (how to measure), and Application Effectiveness (what has been measured). We first explore the theoretical foundations supporting LLM-based measurement, examining the debate on their emergent cognitive properties from a psychometric perspective. Next, we systematically analyze existing measurement paradigms, categorizing them into active conversational assessment, passive natural language analysis, and multimodal fusion. Subsequently, we review the practical effectiveness and limitations of LLMs in core application areas, including personality trait assessment and mental health evaluation. Distinct from prior reviews focusing on general applications or the “psychology” of LLMs themselves, this paper centers on the psychometric properties of LLMs as measurement tools.

Perception under low illumination remains a major challenge for computer vision systems, as RGB sensors often fail to capture sufficient structural and color information in extremely dark environments. Event cameras, with their high dynamic range and temporal resolution, provide complementary cues that are well suited for such conditions. In this work, we present eRetinexGS, a novel framework that jointly leverages event streams and low-light frames through 3D Gaussian Splatting for scene-level enhancement and reconstruction. Unlike previous approaches that operate on individual frames, eRetinexGS enforces geometric and photometric consistency across multiple views, bridging the gap between degraded images and noisy event signals. By introducing an event-assisted Retinex decomposition and a reflectance–illumination representation within the 3DGS pipeline, our method reconstructs normal-light radiance fields with fine-grained details and accurate color. Extensive experiments on both synthetic and real datasets demonstrate that eRetinexGS achieves state-of-the-art performance in low-light scene enhancement while maintaining real-time rendering capability. The code and dataset will be released upon publication.