arXiv:2607.15446v1 Announce Type: new
Abstract: The cost of healthcare remains a concern in the United States and may have been influenced by disruptions associated with the COVID-19 pandemic. This study examines healthcare financial vulnerability before and after the pandemic using Medical Expenditure Panel Survey (MEPS) data from 2019 and 2021. High financial burden was defined as out-of-pocket healthcare expenditures exceeding 10% of family income. Survey-weighted subgroup analyses were performed to obtain nationally representative estimates across demographic and socioeconomic groups.
Descriptive analyses were complemented by interpretable logistic regression and machine learning models. Logistic regression was used to estimate adjusted odds ratios, while random forest and gradient boosting models were used to evaluate predictive performance. Temporal generalization assessed whether models trained on pre-pandemic data remained predictive when applied to post-pandemic observations.
Financial vulnerability was strongly associated with poverty status, insurance coverage, and prescription drug spending. Subgroup analyses indicated persistent disparities across population groups, with some evidence of increased burden among vulnerable populations in 2021. Despite these differences, models trained on pre-pandemic data exhibited only modest reductions in predictive performance when evaluated on post-pandemic data, suggesting that the principal predictors of healthcare financial vulnerability remained relatively stable over time.
These findings provide a population-level assessment of healthcare financial vulnerability during the COVID-19 period and demonstrate the value of combining interpretable statistical modeling with machine learning for population health research. The results may support future population health surveillance, risk stratification, and healthcare policy research aimed at reducing financial barriers to care.
Science Journals
arXiv:2607.15525v1 Announce Type: new
Abstract: Kolmogorov--Arnold Networks (KANs) replace fixed node activations with learned one-dimensional edge functions, offering an explicit interface for interpretation and a possible alternative to transformer feed-forward networks. We test these claims separately. In a six-layer, 10M-parameter B-spline KAN, we reconstruct all 884,736 feed-forward edges: 87.8\% exceed (NLS>0.1) and 0.4\% are inactive. Pruning the lowest-activity 20--25\% causes negligible loss increase, although structured MLP neuron pruning tolerates comparable sparsity. The audit replicates on BabyLM, but grid-size sweeps show that near-total fPCA compression and high closed-form-fit coverage are properties of the low-capacity grid-2 basis, not universal KAN behavior. For replacement, we evaluate MLP, SwiGLU, grouped Chebyshev, and rational GR-KAN networks on BabyLM. The KAN-family and gated variants improve validation loss over the GELU MLP, but this ordering does not transfer to standardized benchmarks: across ten seeds and 59,875 BLiMP pairs, accuracies span 62.4--63.1\%, EWoK remains at chance, and a (+0.7)-point GR-KAN effect on BLiMP reverses on the supplement. Larger tests are also cautionary: parameter-matched MLPEdge underperforms the MLP on Wikitext-103, and 286M-parameter GR-KAN remains below a SwiGLU ClimbMix baseline after stabilization. Thus, small-basis KANs provide a practical, corpus-transferable interface for auditing learned scalar transformations, but the tested replacements show no consistent benchmark, quality, or latency advantage over strong MLP baselines.
arXiv:2607.15529v1 Announce Type: new
Abstract: Outbound enterprise messaging pipelines face a fundamental reliability challenge: delivery status callbacks (webhooks) from messaging providers are subject to network failures, endpoint unavailability, and provider-side retry exhaustion, resulting in stale status records in the CRM system of record. A naive single-path architecture that relies exclusively on webhooks leaves a population of messages permanently in an intermediate state when callbacks fail. This paper presents a two-path status verification architecture, generalized from patterns observed in production CRM-native messaging systems built on multi-tenant platform-as-a-service infrastructure. The primary path uses a real-time webhook received by a REST endpoint, which publishes an internal event for asynchronous record update. The fallback path uses a configurable scheduled polling job that detects records still in transitional status after a configurable interval and queries the provider's status API directly to reconcile state. We describe the event-driven primary path, the scheduler-based fallback, deduplication via idempotent upsert, the sync failure detection mechanism, and the platform resource-limit considerations that shape each design decision.
arXiv:2607.15469v1 Announce Type: new
Abstract: Federated Learning (FL) over 5G cellular networks protects raw data but remains vulnerable to side-channel leakage. Prior fingerprinting attacks assume packet-level network visibility, an assumption that does not hold at the 5G Physical (PHY) layer, where user payloads are encrypted and Radio Network Temporary Identifiers (RNTIs) may change over time. However, we demonstrate that PHY-layer scheduling metadata broadcast over the Physical Downlink Control Channel (PDCCH) preserves architecture-associated temporal patterns. We introduce FLINT, a novel black-box fingerprinting framework that infers FL model architecture families, including CNNs, RNNs, and Transformers, using only coarse PHY-layer observations. FLINT overcomes the lack of network-layer visibility by decoding PDCCH scheduling information, mapping changing RNTIs to physical user devices, and applying multi-view temporal modeling to distinguish architecture-specific training behavior. This leakage is security-critical because knowledge of a client's model architecture can transform passive reconnaissance into targeted downstream exploitation. Extensive experiments on an over-the-air srsRAN-based 5G testbed demonstrate that FLINT achieves a macro F1-score of 0.930 for architecture-family classification. To our knowledge, FLINT is the first work to fingerprint AI/ML model architectures using lower-layer 5G side-channel information obtainable by any protocol-aware adversary.
arXiv:2607.15478v1 Announce Type: new
Abstract: This article investigates several parallelizable alternatives to DTW for estimating the alignment between two long sequences. Whereas most previous work has focused on reducing the total computation and/or memory costs of DTW, our focus is instead on reducing wall clock time by utilizing common hardware like GPUs that are optimized for parallel processing. We propose and study four different parallelizable alignment algorithms: the first three algorithms compute approximations of DTW by breaking the pairwise cost matrix into rectangular regions and processing the regions in parallel, and the fourth algorithm computes an exact DTW alignment by processing the cost matrix along diagonals rather than rows or columns. We characterize the performance of our proposed alignment algorithms on an audio-audio alignment task, and we develop GPU-based implementations for the two best-performing algorithms, which we call weakly-ordered Segmental DTW (WSDTW) and Parallelized Diagonal DTW (ParDTW). Our experiments indicate that ParDTW is the most practical and useful of the four algorithms: it computes an exact DTW alignment and reduces runtime by 1.5 to 2 orders of magnitude on long sequences compared to current alternatives. We present a comprehensive evaluation and study of the alignment accuracy, runtime, and practical limitations of the proposed alignment algorithms.
arXiv:2607.15497v1 Announce Type: new
Abstract: Recent experiments by Sandeep \textit{et al.} [Angew. Chem. Int. Ed. 65, e16917 (2026)] suggest that vibrational strong coupling (VSC) in molecular liquids can generate mesoscopic phenomena beyond single-molecule observables, including resonantly enhanced Rayleigh scattering, abrupt concentration thresholds, and thermal collapse. Motivated by these observations, we construct a mesoscopic Ginzburg--Landau model with two coupled fields: a cavity-controlled collective vibrational polarization $P$ and a secondary structural field $m$ whose long-wavelength susceptibility is renormalized by the collective vibrational polarization intensity $P^2$, assumed to govern long-wavelength density/dielectric fluctuations. With calibrated parameters, the model captures the observed Rayleigh enhancement, collective scaling relations, and threshold-like behavior, while explaining why polaritonic/IR signatures may persist when Rayleigh scattering disappears. The model further predicts enhanced long-wavelength density/dielectric correlations, enlarged mesoscopic correlation lengths, and slowed structural dynamics in the regime with strong Rayleigh enhancement, providing direct experimental tests through small-angle X-ray/neutron scattering and dynamic light-scattering probes.
arXiv:2607.15491v1 Announce Type: new
Abstract: Cross-view geo-localization matches ground-level observations against geo-tagged satellite imagery. Recent methods show that sequential queries such as video clips yield richer spatiotemporal cues than single images, yet they overlook a complementary sequential modality: route descriptions -- which capture the same trajectory at a higher level of abstraction and are often the only input available (e.g., a user directing an autonomous vehicle to a pickup point). To bridge this gap, we introduce SeqGeo-VL, a dataset of $\sim$39K video-text-satellite triplets, and TrajLoc, a unified framework capable of processing both video clips and route descriptions. By leveraging both dense visual and abstract linguistic semantics, TrajLoc enables these modalities to mutually reinforce cross-view matching. We further propose TrajMod, a lightweight module that conditions query embeddings on trajectory geometry, yielding spatially-aware representations. Experiments show that TrajLoc achieves substantial gains over state-of-the-art methods on both video and text geo-localization. The project page is available at https://humblegamer.github.io/trajloc/.
A Tool-Invariant Framework for Teaching and Assessing Computational Methods in the Age of Agentic AI
arXiv:2607.15518v1 Announce Type: new
Abstract: Learning a computational method has always meant learning to operate a tool -- pencil, slide rule, calculator, or programming language. Agentic artificial intelligence, which writes, executes, and revises simulation code from natural-language specifications, is the latest and largest step in a centuries-long migration of mechanical work from human to tool. I argue that what a learner must know has remained remarkably stable: the inputs and outputs of a method, the concept of what it does, the terminology to communicate about it, the judgment to evaluate its results, and the skill of operating the current tool. This paper organizes these requirements into a tool-invariant framework spanning single-digit addition to agent-orchestrated molecular dynamics, argues that verification -- not code authorship -- is now the load-bearing skill, and draws the consequence for assessment: when artifacts can be generated on demand, the artifact no longer certifies the student. I describe a practical response, designed for the small classes where the subject lives -- AI-free in-class coding quizzes paired with oral defenses of comment-stripped, AI-assisted work -- and argue that the real product of a computational physics course is the student's ability to explain and defend computational artifacts in the language of the discipline.
arXiv:2607.15524v1 Announce Type: new
Abstract: Under model--harness co-evolution, harnesses are not merely inference-time scaffolds but data-generating components whose execution traces can shape future foundation models. This motivates harness-in-the-loop learning: optimizing harnesses for both immediate agent performance and the quality of traces used for future model training. However, continually updating provider-built scaffolds is costly and labor-intensive. We therefore investigate whether optimizing user-constructed harnesses in a task-specific manner can improve execution-trace quality while remaining computationally lightweight and requiring only a few update iterations. To this end, we introduce Recursive Harness Self-Improvement (RHI), which represents the harness as a prompt-level specification of the agent loop and iteratively refines it using pairwise feedback over its own revision history. Across 30 synthetic machine-learning research tasks spanning quantitative finance, robotics, and pharmacy, a few RHI iterations suffice to substantially raise the performance ceiling of low-reasoning-effort agents, exceeding the corresponding maximum-reasoning-effort setting while reducing inference cost by up to 60%. We show that these gains arise primarily from improved task-specific context management through more effective inter-agent information flow rather than longer reasoning traces. Finally, we formalize this behavior as an information-theoretic hypothesis for RHI's implicit optimization objective, suggesting RHI as a practical algorithm for continual learning within the paradigm of model--harness co-evolution.
arXiv:2607.15987v1 Announce Type: new
Abstract: The ODRL policy language is emerging as the de-facto standard for policy modelling data access and usage preferences, AI governance policies and data workflows in European dataspaces. The current standard has no mathematical formal semantics to describe how a system should implement policy evaluation. This has resulted in a variety of systems and tools that implement their own interpretation of the language, which limits interoperability and cannot guarantee consistent results. Based on an existing semantic model of ODRL, we formalise the problems of ODRL evaluation for the access control and monitoring scenarios, in both static and streaming settings, and we provide a novel, efficient algorithm and implementation. We present the first ODRL Evaluator with transparent formal semantics and supporting all rule types. We experimentally measure its performance, analysing different scalability dimensions related to policy complexity and size of the data on which a policy is evaluated. We compare our system with the state-of-the-art by providing a comparative review of existing ODRL evaluators, which highlights the differences in supported ODRL features and evaluation modes.
arXiv:2607.15545v1 Announce Type: new
Abstract: LLM-based agents excel at writing articles, coding and information retrieval. However, they fail to form strong collaborations within the scientific community due to the bidirectional, dynamic nature of the problem and a high demand of decision interpretability. We proposed COWEAVER, a bidirectional, learnable and explainable algorithm to match scientists and form strong collaborations within a human-agent network. COWEAVER matches candidates and requesters through filling capability gaps and filters candidates through a two-stage ranking step. Finally, the model explores newcomers by maintaining uncertainty-aware capability estimates and updating them through requester's feedback. We show that the selection mechanism of combining both exploration (UCB) and greedy of COWEAVER exceeds the greedy-only mechanism - the analytical best solution - on 6 out of the 20 tasks and performed on par with the greedy-only mechanism in terms of selecting the best candidate. We compared COWEAVER baselines in terms of matching quality and efficiency. COWEAVER outperforms baselines on all metrics.
arXiv:2607.15561v1 Announce Type: new
Abstract: In this work, we present the comprehensive rounding error analysis of RCLUPPr proposed in \cite{RCLUPP}, which is a novel randomized CholeskyQR-type algorithm performing LU decomposition with partial pivoting (LUPP decomposition) directly on the tall-skinny $X\in\mathbb{R}^{m\times n}$ with $m \ge n$ and $\mbox{rank}(X)=n$. In contrast to the existing RCLUPP in \cite{RCLUPP}, which applies matrix sketching before LUPP decomposition, RCLUPPr places LUPP decomposition as a preconditioning step first, significantly reducing error propagation. Our analysis rigorously proves that RCLUPPr enjoys markedly better applicability to the ill-conditioned matrices than the existing CholeskyQR-type algorithms and remains stable and accurate in the mixed-precision arithmetic. We further propose practical acceleration strategies in the real implementations of RCLUPPr. Extensive numerical experiments on the real-world problems confirm the theoretical results in this work, demonstrating the robustness and practicality of RCLUPPr in the single, double, and the mixed-precision architecture.
arXiv:2607.15576v1 Announce Type: new
Abstract: In the era of Large Language Models (LLMs), efficient vector updates are critical for capturing real-time information from rapidly evolving data. However, it is not trivial to process frequent vector insert and delete updates and maintain a high recall of the search results simultaneously. Specifically, the cluster-based vector indexing methods have high update throughput but low search result quality. Existing out-of-place graph-based vector indexing update approaches suffer from poor update throughput due to the need to periodically merge update batches into the underlying graph index.
Building a vector data system that supports efficient and effective in-place updates is inherently challenging. In this work, we propose Yi to achieve it. In particular, Yi supports in-place graph-based vector indexing updates with consistently high update throughput and good search result quality. The key idea of Yi is decomposition facilitates consolidation. In particular, we introduce a vector-level update mechanism and architect Yi with three core components: (i) a tasklet-based execution engine, (ii) an asynchronous buffer manager, and (iii) a vector file system. Experimental results demonstrate that Yi achieves 1.75x higher update throughput and 1.8x higher concurrent search throughput than the state-of-the-art systems on the 800M dataset, while using only 73% of the peak memory and fewer CPU cores.
arXiv:2607.15592v1 Announce Type: new
Abstract: Multimodal Knowledge Graph Completion (MKGC) requires inferring missing entities from structural, textual, and visual cues. Existing diffusion-based MKGC methods usually denoise directly on raw multimodal features. Such a design forces the denoiser to simultaneously perform relation-dependent cue selection, cross-modal semantic alignment, and structure-aware entity generation, which introduces noisy and semantically inconsistent conditions for diffusion and consequently leads to suboptimal completion performance. To address this limitation, we propose MGDT: MLLM-Guided Diffusion Transformer with Relation-Adaptive Mixture-of-Experts (MGDT), a novel MKGC framework built on an align-then-diffuse paradigm. MGDT first employs a Relation-Adaptive Semantic Routing Mixture-of-Experts (RASR-MoE) module to select relation-relevant multimodal semantic transformation paths and suppress irrelevant modality interference. MGDT then uses a frozen Multimodal Large Language Model (MLLM) as a semantic anchor to align the routed multimodal representations into a unified latent space and reduce cross-modal semantic heterogeneity. Finally, a Knowledge Graph Diffusion Transformer (KGDT) performs graph-conditioned denoising generation in the aligned space to produce the missing entity representation. Experiments on three benchmark datasets show that MGDT consistently outperforms strong baselines.
arXiv:2607.15614v1 Announce Type: new
Abstract: Geometric phase in metasurfaces is conventionally realized using spatially rotated linearly birefringent meta-atoms under circularly polarized illumination. Here we demonstrate that full $\sim2\pi$ geometric phase modulation can be achieved through circular birefringence and observed under linear polarization excitation. We introduce a plasmonic metasurface composed of spatially rotated chiral spiral unit cells designed to produce space-variant circular retardance. This generates a geometric phase ramp equivalent to a blazed grating, leading to $\pm1$ diffraction orders in momentum space. Using leakage radiation microscopy, we directly resolve these orders and show how their intensities depend upon the input linear polarization confirming the geometric origin of the phase. We theoretically analyze the phenomenon using a rotated Poincar\'{e} space and confirm our results by direct Stokes parameters measurement. These results establish basis-independent approach to geometric phase accumulation in circularly birefringent plasmonic metasurfaces.
arXiv:2607.15855v1 Announce Type: new
Abstract: This paper introduces a generative model for temporal networks that jointly controls community evolution and dynamic node sets. The model represents community structure as a sequence of partitions and uses a genetic search guided by a similarity measure based on mutual information to regulate changes between snapshots. This allows explicit control of community evolution including splits and merges while handling node additions and removals. Temporal edges are then generated using intra- and inter-community probabilities derived from data or theoretical bounds to ensure connectivity. Simulation experiments on real-world datasets demonstrate the ability of the generative model to model the evolution of real dynamic communities. The model is used as a benchmark to study the impact of the rate at which nodes join/leave the network on the performance of dynamic community detection algorithms.
arXiv:2607.15644v1 Announce Type: new
Abstract: Partially coherent beams have attracted considerable attention due to their intrinsic resilience against complex environmental perturbations. However, the intrinsic wavefront fluctuations make it fundamentally challenging to preserve well-defined orbital angular momentum during propagation. In this work, we propose and experimentally demonstrate a class of radially correlated, partially coherent beams that carry deterministic vortex structures, generated via optical conformal mapping from Cartesian to log-polar coordinates. The resulting beams exhibit a ring-shaped coherence distribution, characterized by low coherence in the radial direction and high coherence in the azimuthal direction. This unique feature of such a beam supports a well-defined deterministic vortex phase, thereby enabling the beam to preserve its ring-shaped coherence distribution during propagation through a focusing system. Our results provide new insights into the design of new partially coherent beams and may facilitate the development of applications in optical encoding, free-space information transmission, and ultrafast light-matter interactions.
arXiv:2607.15622v1 Announce Type: new
Abstract: Radiative heat transfer at subwavelength distances is generally understood as enhanced energy exchange mediated by photon tunnelling between neighboring bodies. While near-field interactions can dramatically increase mutual heat transfer, whether they also modify the thermal radiation emitted by the bodies themselves remains an open question. Here we experimentally show that near-field electromagnetic coupling reshapes far-field thermal emission through a distance-dependent dressed emissivity. Using a dual-probe calorimetric platform, we independently monitor the radiative balance of two borosilicate microspheres over separations ranging from 120 micrometers to a few hundred nanometers, spanning the transition from the far field to the near field. Nanowatt-resolved differential radiometry reveals asymmetric heat fluxes and a non-monotonic response of the hotter sphere, demonstrating that thermal radiation is governed not only by emitter-bath interactions but also by coupling to the surrounding photonic environment. By analyzing the total power exchanged between the coupled system and the external thermal bath, we directly extract a dressed emissivity and show that near-field interactions renormalize the far-field thermal emission of the pair through a redistribution of the electromagnetic modes available to thermal fluctuations. These observations provide direct experimental evidence that thermal emitters are dressed by their electromagnetic environment, establishing a thermal analogue of the Purcell effect.
arXiv:2607.15788v1 Announce Type: new
Abstract: Flying-focus laser pulses can extend the effective interaction length in inverse Compton sources by controlling the trajectory of the focal intensity. Their practical advantage, however, depends on tolerance to shot-to-shot electron--laser alignment and synchronization errors. We develop a semi-analytical model for the shot-averaged total photon yield in head-on inverse Compton scattering of an axisymmetric Gaussian electron bunch with a flying-focus laser pulse. The model includes finite electron-beam emittance, laser diffraction, transverse laser-centroid jitter, longitudinal focus-position jitter, and laser arrival-time jitter at the nominal interaction point. The ensemble averaging over these independent Gaussian errors and the integration over the longitudinal electron and laser coordinates are performed analytically, reducing the overlap problem to a single positive numerical quadrature. This formulation enables rapid evaluation of jitter-robust operating points and provides a compact tool for defining alignment and synchronization tolerances in flying-focus inverse Compton sources.
arXiv:2607.15641v1 Announce Type: new
Abstract: Humans combine reasoning and motor control to solve complex manipulation tasks under diverse constraints. They build an understanding of the physical world that helps them convert reasoning into actions and quickly adapt to new scenes, tasks, and rules. We refer to this capability as intuitive manipulation. Existing benchmarks fail to capture this integration: they evaluate physical reasoning in isolation from execution, or measure policy performance without requiring explicit reasoning. We introduce IMBENCH, a benchmark designed to evaluate intuitive manipulation as an integrated capability spanning perception, physical reasoning, action generation, and iterative execution. Our tasks require models to infer task-relevant physical structure and generate feasible action sequences under explicit constraints, including contact-rich manipulation, tool use, and multi-stage dependencies. We introduce a benchmark of 35 tasks, 14K filtered trajectories, and scalable tools for generating diverse scenarios. Experiments reveal a consistent gap: vision language models show partial physical reasoning ability but fail to produce executable plans, while state-of-the-art vision-language-action models struggle to satisfy task constraints and generalize across scenarios. These results identify intuitive manipulation as a missing axis in current foundation models and generalist robot policies, and position IMBENCH as a step toward evaluating and enabling more integrated, adaptive physical intelligence.
arXiv:2607.15657v1 Announce Type: new
Abstract: Multimodal AI agents increasingly rely on persistent long-term memory to ground generation in past visual and textual episodes. We show that unconditional trust in visual data creates a critical vulnerability. We propose Lucid, a black-box adversarial framework that compromises multimodal memory pipelines under a strictly image-bounded threat model, requiring no access to the target MLLM, target retrieval encoder, or the text channel. Lucid crafts imperceptible perturbations to enable two distinct failure modes based on the availability of historical context: (1) Memory poisoning, an in-context attack where the adversarial image replaces a benign one whose content is reinforced by prior textual context, reliably corrupting visual recall and steering the agent toward attacker-chosen narratives; (2) Memory injection, an out-of-context attack where the adversarial image replaces a benign one in a conversation turn devoid of prior textual grounding, causing the agent to generate attacker-influenced responses with no corrective signal from memory. We evaluate Lucid across various conversation domains and five black-box memory architectures, including graph-structured, LLM-summarized, and commercially deployed systems. Lucid achieves 61.6% ASR on poisoning and 58.4% ASR on injection, exposing a structural vulnerability in multimodal memory pipelines.
arXiv:2406.01514v5 Announce Type: replace
Abstract: We introduce a training-free safety enhancement method for aligning large language models (LLMs) without the need for supervised fine-tuning or reinforcement learning from human feedback. Our main idea is to provide a robust plug-and-play approach to prevent shadow alignment when models are adapted to downstream tasks. Specifically, we leverage knowledge distillation to extract alignment signals from well-aligned LLMs and inject them into shadow-aligned models via model fusion, enabling plug-and-play alignment correction. In our methodology, we employ delta debugging to identify the critical components of knowledge necessary for effective distillation. On the harmful question dataset, our method significantly enhances the average defense success rate by approximately 14.42%, reaching as high as 51.39% across 17 influenced LLMs, without compromising performance. Our code is available at https://github.com/NWULIST/DAPA.
arXiv:2607.15695v1 Announce Type: new
Abstract: Iterating the Dyson equation with the static part of the self-energy leads to a concise and possibly improved expression of the one-body reduced density matrix from any self-energy approximation. Here we apply the procedure to Hedin's $GW$ approximation. The non-iterated $GW$ based density matrix was already known to yield accurate density matrices for molecular systems. We show that the Dyson-equation-based procedure is equivalent to the so-called variational Z-vector approach applied to the Random-Phase approximation energy functional, but only in the case of a Hartree-Fock mean-field starting point. When a generalized Kohn-Sham scheme is employed instead, the two approaches differ. By comparing the density matrix for a benchmark set of 34 small molecules to coupled-cluster reference values, we conclude that the iterated Dyson equation indeed produces improved density matrices for molecular systems. Interestingly, we observe that the excitation rank of the reference coupled-cluster matters much and that the inclusion of triple excitations (CCSDT) quantitatively changes the conclusions of the benchmark as compared to single and double excitations coupled-cluster (CCSD).
arXiv:2607.15698v1 Announce Type: new
Abstract: We propose and evaluate three hierarchical ensemble setups for zebrafish phenotype classification from embryo images. In all setups, stage 1 uses a single four-class classifier to assign images to one of the exclusive phenotypes: Normal, Chorion, Dead, or Other. Images classified as Other are then processed in stage 2, where the ensemble design differs across setups: a single multi-label classifier, two specialized multi-label classifiers, or an ensemble of binary classifiers. We compare these setups using three backbone architectures: ResNet18, ViT, and ConvNeXt. Overall, ConvNeXt achieves the best performance across setups, while the specialized hierarchical ensemble in setup 2 provides the best balance in terms of F1-score. The results show that the proposed specialised hierarchical ensembles are effective for zebrafish phenotype recognition, and suggest that ConvNeXt is particularly useful backbone model.
arXiv:2607.15724v1 Announce Type: new
Abstract: With the rapid development of deep learning, its vulnerability has gradually emerged in recent years. This work focuses on backdoor attacks on speech recognition systems. We adopt sounds that are ordinary in nature or in our daily life as triggers for natural backdoor attacks. We conduct experiments on two datasets and three models to validate the performance of natural backdoor attacks and explore the effects of poisoning rate, trigger duration and blend ratio on the performance of natural backdoor attacks. Our results show that natural backdoor attacks have a high attack success rate without compromising model performance on benign samples, even with short or low-amplitude triggers. It requires only 5% of poisoned samples to achieve a near 100% attack success rate. In addition, the backdoor will be automatically activated by the corresponding sound in nature, which is not easy to be detected and will bring severer harm.