arXiv subjects

Yingchao Yu

Publications and source records attributed to Yingchao Yu.

15 recordsLinked to original sources

arXiv2025

Self-motion as a structural prior for coherent and robust formation of cognitive maps

Most computational accounts of cognitive maps assume that stability is achieved primarily through sensory anchoring, with self-motion contributing to incremental positional updates only. However, biological spatial representations often remain coherent even when sensory cues degrade or conflict, suggesting that self-motion may play a deeper organizational role. Here, we show that self-motion can act as a structural prior that actively organizes the geometry of learned cognitive maps. We embed a path-integration-based motion prior in a predictive-coding framework, implemented using a capacity-efficient, brain-inspired recurrent mechanism combining spiking dynamics, analog modulation and adaptive thresholds. Across highly aliased, dynamically changing and naturalistic environments, this structural prior consistently stabilizes map formation, improving local topological fidelity, global positional accuracy and next-step prediction under sensory ambiguity. Mechanistic analyses reveal that the motion prior itself encodes geometrically precise trajectories under tight constraints of internal states and generalizes zero-shot to unseen environments, outperforming simpler motion-based constraints. Finally, deployment on a quadrupedal robot demonstrates that motion-derived structural priors enhance online landmark-based navigation under real-world sensory variability. Together, these results reframe self-motion as an organizing scaffold for coherent spatial representations, showing how brain-inspired principles can systematically strengthen spatial intelligence in embodied artificial agents.

q-bio.NC↗

arXiv2025

OutSafe-Bench: A Benchmark for Multimodal Offensive Content Detection in Large Language Models

Since Multimodal Large Language Models (MLLMs) are increasingly being integrated into everyday tools and intelligent agents, growing concerns have arisen regarding their possible output of unsafe contents, ranging from toxic language and biased imagery to privacy violations and harmful misinformation. Current safety benchmarks remain highly limited in both modality coverage and performance evaluations, often neglecting the extensive landscape of content safety. In this work, we introduce OutSafe-Bench, the first most comprehensive content safety evaluation test suite designed for the multimodal era. OutSafe-Bench includes a large-scale dataset that spans four modalities, featuring over 18,000 bilingual (Chinese and English) text prompts, 4,500 images, 450 audio clips and 450 videos, all systematically annotated across nine critical content risk categories. In addition to the dataset, we introduce a Multidimensional Cross Risk Score (MCRS), a novel metric designed to model and assess overlapping and correlated content risks across different categories. To ensure fair and robust evaluation, we propose FairScore, an explainable automated multi-reviewer weighted aggregation framework. FairScore selects top-performing models as adaptive juries, thereby mitigating biases from single-model judgments and enhancing overall evaluation reliability. Our evaluation of nine state-of-the-art MLLMs reveals persistent and substantial safety vulnerabilities, underscoring the pressing need for robust safeguards in MLLMs.

cs.LG↗

arXiv2025

STF: Shallow-Level Temporal Feedback to Enhance Spiking Transformers

Transformer-based Spiking Neural Networks (SNNs) suffer from a great performance gap compared to floating-point \mbox{Artificial} Neural Networks (ANNs) due to the binary nature of spike trains. Recent efforts have introduced deep-level feedback loops to transmit high-level semantic information to narrow this gap. However, these designs often span \mbox{multiple} deep layers, resulting in costly feature transformations, higher parameter overhead, increased energy consumption, and longer inference latency. To address this issue, we propose Shallow-level Temporal Feedback (STF), a lightweight plug-and-play module for the encoding layer, which consists of Temporal-Spatial Position Embedding (TSPE) and Temporal Feedback (TF). Extensive experiments show that STF consistently improves performance across various Transformer-based SNN backbones on static datasets, including CIFAR-10, CIFAR-100, and ImageNet-1K, under different spike timestep settings. Further analysis reveals that STF enhances the diversity of spike patterns, which is key to performance gain. Moreover, evaluations on adversarial robustness and temporal sensitivity confirm that STF outperforms direct coding and its variants, highlighting its potential as a new spike encoding scheme for static scenarios. Our code will be released upon acceptance.

cs.NE↗

arXiv2025

TDFormer: A Top-Down Attention-Controlled Spiking Transformer

Traditional spiking neural networks (SNNs) can be viewed as a combination of multiple subnetworks with each running for one time step, where the parameters are shared, and the membrane potential serves as the only information link between them. However, the implicit nature of the membrane potential limits its ability to effectively represent temporal information. As a result, each time step cannot fully leverage information from previous time steps, seriously limiting the model's performance. Inspired by the top-down mechanism in the brain, we introduce TDFormer, a novel model with a top-down feedback structure that functions hierarchically and leverages high-order representations from earlier time steps to modulate the processing of low-order information at later stages. The feedback structure plays a role from two perspectives: 1) During forward propagation, our model increases the mutual information across time steps, indicating that richer temporal information is being transmitted and integrated in different time steps. 2) During backward propagation, we theoretically prove that the feedback structure alleviates the problem of vanishing gradients along the time dimension. We find that these mechanisms together significantly and consistently improve the model performance on multiple datasets. In particular, our model achieves state-of-the-art performance on ImageNet with an accuracy of 86.83%.

cs.NE↗

arXiv2025

SpiLiFormer: Enhancing Spiking Transformers with Lateral Inhibition

Spiking Neural Networks (SNNs) based on Transformers have garnered significant attention due to their superior performance and high energy efficiency. However, the spiking attention modules of most existing Transformer-based SNNs are adapted from those of analog Transformers, failing to fully address the issue of over-allocating attention to irrelevant contexts. To fix this fundamental yet overlooked issue, we propose a Lateral Inhibition-inspired Spiking Transformer (SpiLiFormer). It emulates the brain's lateral inhibition mechanism, guiding the model to enhance attention to relevant tokens while suppressing attention to irrelevant ones. Our model achieves state-of-the-art (SOTA) performance across multiple datasets, including CIFAR-10 (+0.45%), CIFAR-100 (+0.48%), CIFAR10-DVS (+2.70%), N-Caltech101 (+1.94%), and ImageNet-1K (+1.6%). Notably, on the ImageNet-1K dataset, SpiLiFormer (69.9M parameters, 4 time steps, 384 resolution) outperforms E-SpikeFormer (173.0M parameters, 8 time steps, 384 resolution), a SOTA spiking Transformer, by 0.46% using only 39% of the parameters and half the time steps. The code and model checkpoints are publicly available at https://github.com/KirinZheng/SpiLiFormer.

cs.NE↗

arXiv2025

IP$^{2}$-RSNN: Bi-level Intrinsic Plasticity Enables Learning-to-learn in Recurrent Spiking Neural Networks

Learning-to-learn (L2L), defined as progressively faster learning across similar tasks, is fundamental to both neuroscience and artificial intelligence. However, its neural basis remains elusive, as most studies emphasize neural population dynamics induced by synaptic plasticity while overlooking adaptations driven by intrinsic neuronal plasticity, which point-neuron models cannot capture. To address the above issue, we develop a recurrent spiking neural network with bi-level intrinsic plasticity (IP$^{2}$-RSNN). First, based on task demands, a slow meta-intrinsic plasticity determines which intrinsic neuronal properties are learnable, which is preserved throughout subsequent task learning once configured. Second, a fast intrinsic plasticity fine-tunes those learnable properties within each task. Our results indicate that the proposed bi-level intrinsic plasticity plays a critical role in enabling L2L in RSNNs and show that IP$^{2}$-RSNNs outperform point-neuron recurrent neural networks and self-attention models. Furthermore, our analysis of multi-scale neural dynamics reveals that the bi-level intrinsic plasticity is essential to task-type-specific adaptations at both the neuronal and network levels during L2L, while such adaptations cannot be captured by point-neuron models. Our results suggest that intrinsic plasticity provides significant computational advantages in L2L, shedding light on the design of brain-inspired deep learning models and algorithms.

cs.NE↗

arXiv2024

Heterogeneous Federated Learning with Convolutional and Spiking Neural Networks

Federated learning (FL) has emerged as a promising paradigm for training models on decentralized data while safeguarding data privacy. Most existing FL systems, however, assume that all machine learning models are of the same type, although it becomes more likely that different edge devices adopt different types of AI models, including both conventional analogue artificial neural networks (ANNs) and biologically more plausible spiking neural networks (SNNs). This diversity empowers the efficient handling of specific tasks and requirements, showcasing the adaptability and versatility of edge computing platforms. One main challenge of such heterogeneous FL system lies in effectively aggregating models from the local devices in a privacy-preserving manner. To address the above issue, this work benchmarks FL systems containing both convoluntional neural networks (CNNs) and SNNs by comparing various aggregation approaches, including federated CNNs, federated SNNs, federated CNNs for SNNs, federated SNNs for CNNs, and federated CNNs with SNN fusion. Experimental results demonstrate that the CNN-SNN fusion framework exhibits the best performance among the above settings on the MNIST dataset. Additionally, intriguing phenomena of competitive suppression are noted during the convergence process of multi-model FL.

cs.LG↗

arXiv2024

Privacy-preserving Quantification of Non-IID Degree in Federated Learning

Federated learning (FL) offers a privacy-preserving approach to machine learning for multiple collaborators without sharing raw data. However, the existence of non-independent and non-identically distributed (non-IID) datasets across different clients presents a significant challenge to FL, leading to a sharp drop in accuracy, reduced efficiency, and hindered implementation. To address the non-IID problem, various methods have been proposed, including clustering and personalized FL frameworks. Nevertheless, to date, a formal quantitative definition of the non-IID degree between different clients' datasets is still missing, hindering the clients from comparing and obtaining an overview of their data distributions with other clients. For the first time, this paper proposes a quantitative definition of the non-IID degree in the federated environment by employing the cumulative distribution function (CDF), called Fully Homomorphic Encryption-based Federated Cumulative Distribution Function (FHE-FCDF). This method utilizes cryptographic primitive fully homomorphic encryption to enable clients to estimate the non-IID degree while ensuring privacy preservation. The experiments conducted on the CIFAR-100 non-IID dataset validate the effectiveness of our proposed method.

cs.CR↗

arXiv2021

Improving Fake News Detection by Using an Entity-enhanced Framework to Fuse Diverse Multimodal Clues

Recently, fake news with text and images have achieved more effective diffusion than text-only fake news, raising a severe issue of multimodal fake news detection. Current studies on this issue have made significant contributions to developing multimodal models, but they are defective in modeling the multimodal content sufficiently. Most of them only preliminarily model the basic semantics of the images as a supplement to the text, which limits their performance on detection. In this paper, we find three valuable text-image correlations in multimodal fake news: entity inconsistency, mutual enhancement, and text complementation. To effectively capture these multimodal clues, we innovatively extract visual entities (such as celebrities and landmarks) to understand the news-related high-level semantics of images, and then model the multimodal entity inconsistency and mutual enhancement with the help of visual entities. Moreover, we extract the embedded text in images as the complementation of the original text. All things considered, we propose a novel entity-enhanced multimodal fusion framework, which simultaneously models three cross-modal correlations to detect diverse multimodal fake news. Extensive experiments demonstrate the superiority of our model compared to the state of the art.

cs.MM↗

arXiv2016

Nanomaterial datasets to advance tomography in scanning transmission electron microscopy

Electron tomography in materials science has flourished with the demand to characterize nanoscale materials in three dimensions (3D). Access to experimental data is vital for developing and validating reconstruction methods that improve resolution and reduce radiation dose requirements. This work presents five high-quality scanning transmission electron microscope (STEM) tomography datasets in order to address the critical need for open access data in this field. The datasets represent the current limits of experimental technique, are of high quality, and contain materials with structural complexity. Included are tomographic series of a hyperbranched Co2P nanocrystal, platinum nanoparticles on a carbon nanofibre imaged over the complete 180{\deg} tilt range, a platinum nanoparticle and a tungsten needle both imaged at atomic resolution by equal slope tomography, and a through-focal tilt series of PtCu nanoparticles. A volumetric reconstruction from every dataset is provided for comparison and development of post-processing and visualization techniques. Researchers interested in creating novel data processing and reconstruction algorithms will now have access to state of the art experimental test data.

cond-mat.mes-hall↗

arXiv2014

Breaking the Crowther Limit: Combining Depth-Sectioning and Tilt Tomography for High-Resolution, Wide-Field 3D Reconstructions

To date, high-resolution (< 1 nm) imaging of extended objects in three-dimensions (3D) has not been possible. A restriction known as the Crowther criterion forces a tradeoff between object size and resolution for 3D reconstructions by tomography. Further, the sub-Angstrom resolution of aberration-corrected electron microscopes is accompanied by a greatly diminished depth of field, causing regions of larger specimens (> 6 nm) to appear blurred or missing. Here we demonstrate a three-dimensional imaging method that overcomes both these limits by combining through-focal depth sectioning and traditional tilt-series tomography to reconstruct extended objects, with high-resolution, in all three dimensions. The large convergence angle in aberration corrected instruments now becomes a benefit and not a hindrance to higher quality reconstructions. A through-focal reconstruction over a 390 nm 3D carbon support containing over one hundred dealloyed and nanoporous PtCu catalyst particles revealed with sub-nanometer detail the extensive and connected interior pore structure that is created by the dealloying instability.

cond-mat.mtrl-sci↗

arXiv2013

Nanoscale Imaging of Lithium Ion Distribution During In Situ Operation of Battery Electrode and Electrolyte

A major challenge in the development of new battery materials is understanding their fundamental mechanisms of operation and degradation. Their microscopically inhomogeneous nature calls for characterization tools that provide operando and localized information from individual grains and particles. Here we describe an approach that images the nanoscale distribution of ions during electrochemical charging of a battery in a transmission electron microscope liquid flow cell. We use valence energy-loss spectroscopy to track both solvated and intercalated ions, with electronic structure fingerprints of the solvated ions identified using an ab initio non-linear response theory. Equipped with the new electrochemical cell holder, nanoscale spectroscopy and theory, we have been able to determine the lithiation state of a LiFePO4 electrode and surrounding aqueous electrolyte in real time with nanoscale resolution during electrochemical charge and discharge. We follow lithium transfer between electrode and electrolyte and observe charging dynamics in the cathode that differ among individual particles. This technique represents a general approach for the operando nanoscale imaging of electrochemically active ions in a wide range of electrical energy storage systems.

physics.chem-ph↗

arXiv2012

In Situ Electron Energy-Loss Spectroscopy in Liquids

In situ scanning transmission electron microscopy (STEM) through liquids is a promising approach for exploring biological and materials processes. However, options for in situ chemical identification are limited: X-ray analysis is precluded because the liquid cell holder shadows the detector, and electron energy-loss spectroscopy (EELS) is degraded by multiple scattering events in thick layers. Here, we explore the limits of EELS for studying chemical reactions in their native environments in real time and on the nanometer scale. The determination of the local electron density, optical gap and thickness of the liquid layer by valence EELS is demonstrated. By comparing theoretical and experimental plasmon energies, we find that liquids appear to follow the free-electron model that has been previously established for solids. Signals at energies below the optical gap and plasmon energy of the liquid provide a high signal-to-background ratio regime as demonstrated for LiFePO4 in aqueous solution. The potential for using valence EELS to understand in situ STEM reactions is demonstrated for beam-induced deposition of metallic copper: as copper clusters grow, EELS develops low-loss peaks corresponding to metallic copper. From these techniques, in situ imaging and valence EELS offer insight into the local electronic structure of nanoparticles and chemical reactions.

physics.chem-ph↗

arXiv2011

3-D Tracking and Visualization of Hundreds of Pt-Co Fuel Cell Nanocatalysts During Electrochemical Aging

We present an electron tomography method that allows for the identification of hundreds of electrocatalyst nanoparticles with one-to-one correspondence before and after electrochemical aging. This method allows us to track, in three-dimensions (3-D), the trajectories and morphologies of each Pt-Co nanocatalyst on a fuel cell carbon support. The use of atomic-scale electron energy loss spectroscopic imaging enables the correlation of performance degradation of the catalyst with changes in particle/inter-particle morphologies, particle-support interactions and the near-surface chemical composition. We found that, aging of the catalysts under normal fuel cell operating conditions (potential scans from +0.6 V to +1.0 V for 30,000 cycles) gives rise to coarsening of the nanoparticles, mainly through coalescence, which in turn leads to the loss of performance. The observed coalescence events were found to be the result of nanoparticle migration on the carbon support during potential cycling. This method provides detailed insights into how nanocatalyst degradation occurs in proton exchange membrane fuel cells (PEMFCs), and suggests that minimization of particle movement can potentially slow down the coarsening of the particles, and the corresponding performance degradation.

cond-mat.mtrl-sci↗

arXiv2010

Pt-Decorated PdCo@Pd/C Core-Shell Nanoparticles with Enhanced Stability and Electrocatalytic Activity for Oxygen Reduction Reaction

A simple method for the preparation of PdCo@Pd core-shell nanoparticles supported on carbon has been developed using an adsorbate-induced surface segregation effect. The stability and electrocatalytic activity for the oxygen reduction of PdCo@Pd nanoparticles was enhanced by a small amount of Pt, deposited via a spontaneous displacement reaction. The facile method described herein is suitable for large-scale lower cost production and significantly lowers the Pt loading and thus cost. The as-prepared PdCo@Pd and Pd-decorated PdCo@Pd nanocatalysts have higher methanol-tolerance for the ORR when compared to Pt/C, and are promising cathode catalysts for fuel cell applications.

cond-mat.mtrl-sci↗