The rapid advancement of Artificial Intelligence (AI) has accelerated the development of personalised healthcare. However, the clinical adoption of deep learning remains constrained by a persistent âtrust gapâ surrounding model transparency, security, and data privacy. This talk presents a research vision for building trustworthy human-centred AI in computer audition and biosignal processing by integrating signal processing, machine learning, and healthcare. This vision is structured around three interconnected pillars: Sensing, Understanding, and Empowerment. The Sensing pillar focuses on enabling machines to perceive clinically relevant information from multimodal biosignals, transforming body sounds and physiological signals into non-invasive, accessible, and cost-effective windows into human health and well-being. The Understanding pillar addresses the black-box nature of modern AI systems, with the goal of ensuring transparency, efficiency, robustness, and security in clinical deployment. This is achieved through the development of explainable AI methods, knowledge distillation techniques, and defences against adversarial attacks, fostering AI systems that clinicians can trust and interpret. The Empowerment pillar seeks to restore natural communication for individuals with speech impairments, such as laryngectomy patients. By developing Silent Speech Interfaces (SSIs) that translate facial muscle activity (EMG) directly into audible speech, this research advances speech intelligibility, naturalness, real-time causal architectures, and multi-speaker communication scenarios. Collectively, these research directions aim to establish a foundation of trustworthy, human-centred AI that is not only accurate and secure, but also empathetic, accessible, and impactful, ultimately enhancing healthcare, communication, and quality of life.
论文检索
输入标题、作者或关键词，从 1,034 篇学术成果中精准定位
Mixture-of-Experts (MoE) presents a naturally compatible and scalable framework for multimodal learning, demonstrating strong adaptability across diverse modalities and tasks. Despite its growing success, a comprehensive and systematic evaluation of multimodal MoE remains lacking. Existing surveys tend to address either multimodal learning or MoE independently, overlooking the unique interplay between them. This survey fills that gap by addressing a central question: How does MoE effectively resolve multimodal challenges? We approach this from three key perspectives: (1) MoE as an Efficient Multimodal Framework: enabling scalable multimodal modeling by decoupling computational cost from parameter growth and mitigating modality redundancy through selective expert activation; (2) MoE as a Multimodal Representation Learner: integrating complementary multi-opinion expert knowledge to enrich alignment and interaction representations; and (3)MoE as a Multimodal Adapter: providing a modular and flexible mechanism to model imperfect modality data such as modality imbalance and missing modality. Through an extensive literature review, we identify critical research gaps, including interpretable routing, expert communication, modality integration, and lifelong multimodal learning. We position this survey as a foundation for future research toward interpretable, adaptive, and sustainable multimodal Mixture-of-Experts systems.
Generative models produce long and high probability sequences, yet they often fail to satisfy explicit constraints set by users. Over the past two decades, Constraint Programming (CP) has provided a complementary paradigm: combining generative models with a constraint solver to guarantee feasibility. This survey reviews the main concepts behind these CP-driven hybrid approaches, from enforcing ubiquitous structural rules (e.g., length and patterns) to preventing plagiarism. It synthesizes how learned models can be treated as constraints, compiled structures, or probabilistic factors. We highlight what has remained stable across applications, then discuss how these principles transfer to the Large Language Model era and outline open challenges for controllable and trustworthy generative systems.
Deep neural networks are vulnerable to adversarial examples, posing serious security risks. While GAN-based attacks such as AdvGAN enable fast adversarial example generation, they typically operate in a single step, limiting attack effectiveness and transferability. In this work, we propose PAR-AdvGAN, a progressive auto-regressive GAN framework that iteratively refines adversarial perturbations. By conditioning each iteration on the previous output, PAR-AdvGAN produces stronger, more transferable attacks while maintaining low visual distortion and high generation speed. Experiments on multiple models and datasets demonstrate that PAR-AdvGAN significantly outperforms baseline methods in attack success rate, achieving up to 335.5 frames per second, highlighting its practical efficiency for black-box attack scenarios.
Effective flood risk management relies on accurate forecasting, yet the ”black box” nature of state-of-the-art Deep Learning models creates a barrier to trust and accountability in high-stakes public safety decisions. While existing Explainable AI (XAI) methods offer local attributions, they fail to provide the verifiable, operationally meaningful causal narratives required by disaster response authorities. To address this societal challenge, we propose Context-Aware Concept Distillation (CACD), a framework developed in collaboration with domain experts to distill opaque LSTMs into interpretable, hydrology-aware surrogate models. We introduce an unsupervised pipeline to discover a ”Hydrological Language” and a Residual Hypernetwork that dynamically modulates these concepts based on static basin characteristics. Evaluated on 5,203 basins globally, our model achieves high fidelity (Median NSE 0.70), significantly outperforming black-box baselines (e.g., Multi Layer Perceptrons) on unseen future data. By demonstrating that human-interpretable concepts are sufficient to reconstruct flood dynamics, this work balances AI accuracy with the transparency required for responsible environmental decision-making.
Modern neural networks can be accurate yet poorly calibrated. We present Random Geometric Calibration (RGC), a post hoc method that augments confidence estimation with a geometric signal derived from distances to training data in an aggregated semantic representation space. Prior geometric calibration methods in feature space often rely on architecture-specific layer choices, which can be unstable across models and datasets. To avoid this dependence, RGC samples a fixed number of intermediate blocks uniformly at random and reuses this set at both calibration and test time. We instantiate RGC with RGCL, which applies spatial pyramid pooling to each sampled activation, projects features to a fixed dimension, and computes a nearest-neighbor separation score that is mapped to probabilities via isotonic regression. We also present RGCC, which replaces pooling and projection with uniform coordinate sampling for improved computational efficiency. We evaluate RGCL and RGCC across multiple datasets, including CIFAR-10, CIFAR-100, and Tiny-ImageNet, and across a diverse set of architectures such as ResNet, DenseNet, and DINOv2. Our results show that RGCL achieves state-of-the-art or near state-of-the-art calibration performance, reducing expected calibration error by approximately 78% on average and by up to 97% for some models. Finally, we provide representation analysis that explains why RGCL and RGCC succeed without layer selection and characterize their computational overheads. We observe that layers exhibiting favorable geometric structure tend to emerge in later network blocks, and that randomized aggregation implicitly emphasizes these layers, leading to robust and reliable calibration performance.
Federated learning (FL) enables collaborative training of deep models while keeping raw data on local devices. A critical factor affecting the global performance is server-side aggregation. The most common strategy weights client updates by the amount of local data, which is often unreliable as the server cannot inspect data. More importantly, it ignores substantial variations in data quality (e.g., label noise) across heterogeneous clients. To address this issue, we propose a data-Quality-aware aggregation framework by introducing an Evolutionary-computation-inspired design into Federated learning (FedEvoQ), with a lightweight dual-branch architecture. Each client trains a main-task branch as usual, while a frozen auxiliary branch (EvoQ) performs evolutionary-style controlled sampling with crossover and mutation at the input and representation level to probe data diversity and reliability. EvoQ outputs a compact quality score that jointly reflects learnability (loss-based signal), geometric consistency (stability under perturbations), representation diversity, and out-of-distribution (OOD) deviation, enabling the server to compute quality-only aggregation weights. Extensive experiments on CIFAR10, CIFAR100, and PathMNIST under noisy and Non-independent and identically distributed (Non-IID) settings demonstrate that FedEvoQ consistently outperforms 10 state-of-the-art (SOTA) federated aggregation methods, improving the average test accuracy of the global model by over 2% with modest overhead.
LLMs' overconfidence, particularly when hallucinating, poses a significant challenge for the deployment of the models in safety-critical settings and makes a reliable estimation of uncertainty necessary. Existing approaches for uncertainty quantification typically prioritize lexical or probabilistic measures; however, these techniques often ignore the semantic variance of different responses with similar meaning. In this paper, we propose Adaptive Conformal Semantic Entropy (ACSE), a method for estimating prompt-level uncertainty by adaptively measuring semantic dispersion in LLMs outputs. Our uncertainty scoring function is based on clustering semantic entropy of multiple diverse responses to the same prompt. The function adaptively adjusts the uncertainty score based on semantic features of each cluster. To ensure statistical reliability of our score, we use conformal calibration to apply a decision rule to accept/abstain the prompts, providing a finite-sample, distribution-free guarantee such that the error rate among the accepted responses remains bounded by a user-specified tolerance. Our extensive experimental evaluations using different LLMs and datasets, demonstrate that our approach consistently outperforms state-of-the-art uncertainty quantification baselines using discriminative performance, acceptance rate, conformal guarantees, and probabilistic calibration indicators. As a highlight, for TriviaQA dataset, AUROC of our approach is 0.88 compared to 0.65 produced by the token entropy approach.
Neural Network k-Nearest Neighbor (NN-kNN) was proposed as an interpretable network model that learns feature weights and similarity to retrieve relevant cases for classification. This paper extends it to regression with the goal of generating accurate predictions based on neighboring cases with similar labels. Specifically, we introduce three modular components: an attention mechanism that weights the contribution of retrieved cases, a locality-aware regularizer that favors label-similar neighbors, and an optional case adaptation module that refines the retrieved estimate. Across synthetic and standard tabular regression benchmarks, NN-kNN achieves competitive predictive error against strong baselines (kNN-R, MLKR, and MLPs) while providing cases with similar labels as explanation (later referred as label-similar case). Moreover, NN-kNN supports manual knowledge injection through tuning weights for human-comprehensible features. The result is a simple, general, and interpretable approach to continuous-valued prediction that unifies retrieval, attention, and optional case adaptation within a single neural framework.
We investigate the computational complexity of neural network verification in quantised settings. We distinguish three classes of Feedforward Neural Networks (FNNs): rational FNNs with exact rational weights, quantised FNNs whose weights come from a finite-width arithmetic, and dynamically quantised FNNs in which rational networks are evaluated with respect to a given finite-width arithmetic. We consider two types of specifications used in the literature. Linear programming (LP) specifications are conjunctions of linear constraints, while bit-vector (BV) specifications allow reasoning at the bit level and can express non-linear constraints. Our results give a complexity landscape of these verification problems. For quantised FNNs with fixed arithmetic precision, we show that verification under both LP and BV specifications remains NP-complete, matching the complexity of the rational case. For dynamically quantised FNNs with BV specifications, we establish upper bounds, complementing a previously known PSPACE-hardness result.
Machine learning (ML)-based malware detectors are widely deployed but remain vulnerable to adversarial attacks. However, under hard-label black-box access, existing adversarial attacks on Windows Portable Executable (PE) malware are often query-inefficient and incur large file-size inflation. A common paradigm is to predefine a set of semantics-preserving atomic perturbations and search for evasive combinations under only binary feedback. Among these atomic perturbations, (i) those with highly combinatorial search spaces are difficult to explore effectively under hard-label feedback, leaving their potential untapped, and (ii) those relying on transplanting benign fragments are often laden with evasion-irrelevant bytes and exhibit highly variable adversarial utility. We propose Gradient-seeded Reinforcement Learning And Stealthy Pruning (GRASP), a three-stage framework that tackles these challenges. First, we decouple perturbations with highly combinatorial search spaces from the query-based search and instead apply a gradient-seeded warm-up that uses Gumbel-Softmax relaxation to enable gradient-based updates over the discrete space. This yields a strong warm start that improves evasion and reduces queries in later stages. Second, Reinforcement Learning (RL)-based refinement is accelerated by a perturbation library that filters, caches, and reuses compact high-utility patterns, reducing wasted queries on low-utility benign fragments. Third, a perturbation minimization stage removes redundant bytes while preserving evasion, reducing size inflation and feeding compact patterns back to the library. Experiments show that GRASP outperforms baselines, achieving higher attack success with fewer queries and smaller file-size inflation. We additionally demonstrate its practical effectiveness against commercial Antivirus engines.
Unifying Perspectives: Plausible Counterfactual Explanations on Global, Group-wise, and Local Levels
The growing complexity of AI systems has intensified the need for transparency through Explainable AI (XAI). Counterfactual explanations (CFs) offer actionable "what-if" scenarios on three levels: Local CFs providing instance-specific insights, Global CFs addressing broader trends, and Group-wise CFs (GWCFs) striking a balance and revealing patterns within cohesive groups. Despite the availability of methods for each granularity level, the field lacks a unified method that integrates these complementary approaches. We address this limitation by proposing a gradient-based optimization method for differentiable models that generates Local, Global, and Group-wise Counterfactual Explanations in a unified manner. We especially enhance GWCF generation by combining instance grouping and counterfactual generation into a single efficient process, replacing traditional two-step methods. Moreover, to ensure trustworthiness, we innovatively introduce the integration of plausibility criteria into the GWCF domain, making explanations both valid and realistic. Our results demonstrate the method's effectiveness in balancing validity, proximity, and plausibility while optimizing group granularity, with practical utility validated through practical use cases.
Evaluating LLM forecasting capabilities is constrained by a fundamental tension: prospective evaluation offers methodological rigor but prohibitive latency, while retrospective forecasting (RF)—evaluating on already-resolved events—faces rapidly shrinking clean evaluation data as SOTA models possess increasingly recent knowledge cutoffs. Simulated Ignorance (SI), prompting models to suppress pre-cutoff knowledge, has emerged as a potential solution. We provide the first systematic test of whether SI can approximate True Ignorance (TI). Across 470 competition-level questions and 9 models, we find that SI fails systematically: (1) cutoff instructions leave a 52% performance gap between SI and TI; (2) chain-of-thought reasoning fails to suppress prior knowledge, even when reasoning traces contain no explicit post-cutoff references; (3) reasoning-optimized models exhibit worse SI fidelity despite superior reasoning trace quality. These findings demonstrate that prompts cannot reliably "rewind" model knowledge. We conclude that RF on pre-cutoff events is methodologically flawed; we recommend against using SI-based retrospective setups to benchmark forecasting capabilities.
Vision-language model (VLM) encoders such as CLIP enable strong retrieval and zero-shot classification in a shared image–text embedding space, yet the semantic organization of this space is rarely inspected. We present a post-hoc framework to explain, verify, and align the semantic hierarchies induced by a VLM over a given set of child classes. First, we extract a binary hierarchy by agglomerative clustering of class centroids and name internal nodes by dictionary-based matching to a concept bank. Second, we quantify plausibility by comparing the extracted tree against human ontologies using efficient tree- and edge-level consistency measures, and we evaluate utility via explainable hierarchical tree-traversal inference with uncertainty-aware early stopping (UAES). Third, we propose an ontology-guided post-hoc alignment method that learns a lightweight embedding-space transformation, using UMAP to generate target neighborhoods from a desired hierarchy. Across 13 pretrained VLMs and 4 image datasets, our method finds systematic modality differences: image encoders are more discriminative, while text encoders induce hierarchies that better match human taxonomies. Overall, the results reveal a persistent trade-off between zero-shot accuracy and ontological plausibility and suggest practical routes to improve semantic alignment in shared embedding spaces.
Adversarial training is effective on balanced datasets, but its robustness degrades under long-tailed class distributions, where tail classes suffer high robust error and unstable decision boundaries. We propose \emph{Manifold-Constrained Adversarial Training (MCAT)}, a unified framework that enforces the semantic validity of adversarial examples by penalizing deviations from class-conditional manifolds in feature space, while promoting balanced geometric separation across classes via an ETF-inspired regularization. We provide theoretical results that link geometric separation to lower bounds on adversarially robust margins, and show that manifold-constrained adversarial risk upper-bounds robust risk on high-density semantic regions. Extensive experiments on standard long-tailed benchmarks demonstrate consistent improvements in overall, balanced, and tail-class adversarial robustness. The codes and appendix are available on https://github.com/yneversky/MCAT.
Open‑vocabulary video instance segmentation (OV‑VIS) couples spatial‑temporal reasoning with language grounding, yet its adversarial robustness has remained unexplored. We present the Dual-Objective Triggers (DOT), the first transferable attack on OV-VIS that simultaneously exploits the vision–language coupling and temporal coherence. DOT deploys a Dual Semantic Perturbation Module that overlays two complementary triggers: a Semantic Suppression Trigger erases the alignment between the true object and the query, while a Plausible Replacement Trigger steers the tracker toward a phantom trajectory that is visually plausible and text‑consistent. To amplify cross‑model transferability without sacrificing perceptual fidelity, we introduce Phase‑Guided Adversarial Training, which injects perturbations primarily in the phase spectrum while blending amplitudes with clean references. Extensive experiments on four state‑of‑the‑art OV‑VIS implementations demonstrate that DOT reduces mAP by up to 69.3% and raises attack success rate by up to 98%, outperforming the strongest baselines by a factor of 1.6× on average, while maintaining a PSNR of 52.49 dB, thus exposing critical security vulnerabilities and laying a foundation for future research on robust and trustworthy vision–language systems.
Gradient boosting remains a strong and widely used method for tabular data learning, but its performance often degrades when training labels are noisy. This behavior is largely related to the way boosting algorithms emphasize samples with large gradients, without explicitly accounting for whether such errors originate from informative hard cases or from unreliable labels. We address this issue by reconsidering how sample reliability is evaluated during boosting. Instead of relying on instantaneous error, we examine the evolution of each sample’s residuals across iterations. Based on this insight, we propose Information-Theoretic Trust Boosting (ITBoost), which uses the Minimum Description Length principle to measure the complexity of residual trajectories. Samples whose residual patterns fluctuate in an irregular manner are treated as less trustworthy and are down-weighted during learning. Theoretically, we derive a tighter generalization bound for ITBoost under label noise. Empirical results on various tabular benchmarks indicate that ITBoost provides improved robustness in noisy environments over leading boosting and deep tabular models, while retaining best average performance on clean data.
zkQML: Verifiable and Privacy-Preserving Inference for Quantum Machine Learning (Student Abstract)
PDF ↗Quantum machine learning (QML) has attracted growing interest for their ability to achieve superior performance with significantly fewer parameters. However, the high cost and scarcity of current hardware push inference to cloud-hosted quantum devices, creating a tension between verifiability and confidentiality. This work proposes a novel framework that converts quantum neural network operations into classical arithmetic circuits that faithfully approximate genuine quantum computations. By encrypting these circuits with zero-knowledge proofs, it ensures computational validity while concealing internal parameters. Experimental results show that our classical circuits achieve fidelity above 0.9996 and total variation distance below 1% compared to actual quantum computations, verifying the practicality of trustworthy and privacy-preserving quantum inference.
Machine learning (ML)-powered systems are capable of reproducing and often amplifying undesired biases embedded in society, emphasizing the importance of operating under practices that enable the study and understanding of the intrinsic characteristics of ML pipelines. This supports the emergence of documentation frameworks with the idea that “any remedy for bias starts with awareness of its existence.” However, a resource that can formally describe ML pipelines in terms of detected biases is still missing. To address this gap, we present the Doc-BiasO ontology, a resource that sets out to create an integrated vocabulary of biases defined in the Trustworthy AI literature and their measures, as well as to incorporate relevant domain terminology and relationships between them. Overseeing ontology engineering best practices, we reuse existing vocabularies on machine learning and AI to foster knowledge sharing and interoperability between the actors concerned with its research, development, regulation, and others. In addition, we demonstrate the potential of Doc-BiasO with an experiment on an existing benchmark and as part of a neuro-symbolic system. Overall, our main objective is to contribute towards clarifying existing terminology on bias research as it rapidly expands to all areas of AI and to improve the interpretation of bias in data and downstream impact through its documentation.
Safe reinforcement learning (RL) has emerged as a key paradigm for deploying AI in high-stakes domains such as autonomous driving, robotics, healthcare, and recommender systems. By embedding constraints into the learning process, safe RL enables agents to optimize performance while satisfying critical requirements, including collision avoidance, resource limits, and system reliability. Such guarantees are indispensable for real-world AI, where failures can cause physical harm, economic loss, or loss of trust. At the same time, demand for trustworthy AI continues to grow as machine learning is increasingly deployed in human-centered applications. This makes it essential to design RL algorithms that are not only efficient but also reliable, robust, and aligned with societal needs.