arXiv:2509.20623v2 Announce Type: replace
Abstract: Reinforcement learning has enabled significant progress in complex domains such as coordinating and navigating multiple quadrotors. However, even well-trained policies remain vulnerable to collisions in obstacle-rich environments. Addressing these infrequent but critical safety failures through retraining or fine-tuning is costly and risks degrading previously learned skills. Inspired by activation steering in large language models and latent editing in computer vision, we introduce a framework for inference-time Latent Activation Editing (LAE) that refines the behavior of pre-trained policies without modifying their weights or architecture. The framework operates in two stages: (i) an online classifier monitors intermediate activations to detect states associated with undesired behaviors, and (ii) an activation editing module that selectively modifies flagged activations to shift the policy towards safer regimes. In this work, we focus on improving safety in multi-quadrotor navigation. We hypothesize that amplifying a policy's internal perception of risk can induce safer behaviors. We instantiate this idea through a latent collision world model trained to predict future pre-collision activations, thereby prompting earlier and more cautious avoidance responses. Extensive simulations and real-world Crazyflie experiments demonstrate that LAE achieves statistically significant reduction in collisions (nearly 90% fewer cumulative collisions compared to the unedited baseline) and substantially increases the fraction of collision-free trajectories, while preserving task completion. More broadly, our results establish LAE as a lightweight paradigm, feasible on resource-constrained hardware, for post-deployment refinement of learned robot policies.
Science Journals
arXiv:2603.00667v3 Announce Type: replace
Abstract: Computational pathology has advanced rapidly in recent years, driven by domain-specific image encoders and growing interest in using vision-language models to answer natural-language questions about diseases. Yet, the core problem behind pathology question-answering remains unsolved, considering that a gigapixel slide contains far more information than necessary for a given question. Pathologists naturally navigate tissue and morphology complexity by scanning broadly, and zooming in selectively according to the clinical questions. Current models, in contrast, rely on uniform patch sampling or broad attention maps, often attending equally to irrelevant regions while overlooking key visual evidence. In this work, we try to bring models closer to how humans actually examine slides. We propose a question-guided, tissue-aware, and coarse-to-fine retrieval framework, HistoSelect, that consists of two key components: a group sampler that identifies question-relevant tissue regions, followed by a patch selector that retrieves the most informative patches within those regions. By selecting only the most informative patches, our method becomes significantly more efficient: reducing visual token usage by 70% on average, while improving accuracy across three pathology QA tasks. Evaluated on 356,000 question-answer pairs, our approach outperforms existing methods and produces answers grounded in interpretable, pathologist-consistent regions. Our results suggest that bringing human-like search and attention patterns into WSI reasoning is a promising direction for building practical and reliable pathology VLMs. Code is available at https://github.com/winston52/HistoSelect.
arXiv:2407.11821v2 Announce Type: replace
Abstract: Statistical information is ubiquitous but drawing valid conclusions from it is prohibitively hard. We explain how knowledge graph embeddings can be used to approximate probabilistic inference efficiently using the example of Statistical EL (SEL), a statistical extension of the lightweight Description Logic EL. We provide proofs for runtime and soundness guarantees, and empirically evaluate the runtime and approximation quality of our approach.
arXiv:2606.03553v1 Announce Type: cross
Abstract: While principal component analysis (PCA) is a fundamental tool for dimensionality reduction, its dense representations make it ill-suited for high-dimensional data. Existing methods address this by promoting sparsity through explicit $\ell_1$-penalties, but these are not obvious to tune due to the unsupervised nature of the task. In contrast, we propose Adversarial PCA (AdvPCA), which leverages robust optimization to achieve sparsity by optimizing the reconstruction objective against bounded, worst-case latent space perturbations. We show that this formulation admits a closed-form reduction, leading to a practical iterative algorithm that alternates between adversarial linear regression-style updates for the sparse encoder and orthogonal updates for the decoder. By theoretically characterizing the solution, we derive a data-adaptive parameterization that allows the algorithm to perform effectively out of the box. We validate these claims through numerical experiments on synthetic and real-world genomics data.
arXiv:2606.02651v1 Announce Type: new
Abstract: Enforcing invariants in safety-critical systems is increasingly urgent as AI-generated code becomes widespread. Unfortunately, the runtimes required to support high-level specification languages are too large for most embedded targets. In this article, we show how formally verified firmware is achievable today. We built Encore!, a bare-metal Continuation Passing Style (CPS) virtual machine (VM) that runs Rocq-extracted Scheme on microcontrollers. We also show how to structure firmware as a pure state-transition function, making its core fully provable in Rocq while keeping the unverified host layer constant regardless of firmware complexity. Large Language Model (LLM)-assisted tactic synthesis fits naturally into this workflow: formal theorem statements replace manual code review, allowing AI-generated firmware to prove itself.
arXiv:2606.02659v1 Announce Type: new
Abstract: Multimodal data fusion involves integrating and analyzing information from multiple modalities to uncover latent correlations and complementary patterns, thereby enhancing data processing and decision-making. While existing methods for structured multimodal inputs are typically designed around specific tasks and assume fully observed modalities, real-world applications often suffer from uncertain or missing modality inputs due to various factors. Some traditional models overly emphasize local interactions within missing modalities, neglecting the global complementary cues embedded in multimodal representations. To overcome these limitations, we propose a Dynamic Multimodal Data Fusion model based on Contrastive Learning (CL-DMDF). CL-DMDF introduces a novel attention mechanism that operates across both feature and modality dimensions to compute reliable attention scores, effectively reflecting importance at each level. The CL-DMDF further incorporates an entity-centroid contrastive learning module that constructs centroid-based positive samples from entity features to enhance discriminative learning. Additionally, an adaptive fusion module is employed to improve the efficiency and accuracy of dynamic fusion strategies. Extensive experiments conducted on three datasets demonstrate the effectiveness of the CL-DMDF across diverse multimodal fusion tasks.
arXiv:2606.02674v1 Announce Type: new
Abstract: The rapid proliferation of multi-cloud and SaaS platforms has transformed Identity Security Posture Management (ISPM) into a fundamentally cross-vendor challenge: critical misconfigurations and privilege escalation paths increasingly span multiple identity providers, infrastructure layers, and authentication systems never designed to interoperate. Existing evaluations focus on isolated single-platform environments and provide no means to assess whether an AI agent can reason across these fragmented boundaries. To address this gap, we introduce the Cross-Vendor Sola ISPM Benchmark, a production-grade benchmark of 50 data-grounded tasks requiring multi-hop entity resolution and cross-system correlation across eight integrated enterprise platforms including AWS, Okta, Azure AD, and Google Workspace. We also contribute an evaluation framework measuring not only final answer correctness but also evidentiary grounding, structural join fidelity, retrieval quality, and SQL equivalence. We evaluate the Sola AI Agent across five context configurations - from no injected metadata to full schema, graph, and retrieval context - using three frontier LLMs. Results show that structured relational context improves answer correctness by approximately 34% relatively and reduces exploration queries by approximately 70% across all tested models, with the largest gains driven by cross-vendor graph topology. Our findings indicate that frontier LLMs possess substantial latent security reasoning capability, but reliable cross-vendor identity analysis is fundamentally constrained by the availability of explicit relational context for entity resolution and evidentiary grounding. Under full context, the best configuration achieves 78% answer correctness while reducing complete failure to 4%.
arXiv:2606.02763v1 Announce Type: new
Abstract: AI chat tools are shifting problem-solving and brainstorming conversations away from colleagues and into private AI interactions, reducing the shared awareness that supports team coordination. We introduce InquiryBits, a system that shares minimal summaries of AI conversations within configurable trust boundaries, separating AI-only analysis from human-visible sharing. In a study with 80 professionals, we find that people are broadly willing to share these traces to support collaboration and avoid duplicating work - but only within bounded groups. Comfort drops sharply as audience expands beyond close teams; the level of detail shared matters less than who can see it, with a preference for more detail over less within trusted groups. These findings suggest that trust boundaries, more than information granularity, may be the most impactful design parameter.
arXiv:2606.02677v1 Announce Type: new
Abstract: Motion planning in dynamic environments requires robots to continuously adapt their paths in response to environmental changes for safe and uninterrupted navigation. While many surveys have reviewed planning in static settings, systematic reviews focused on dynamic environments remain limited. This paper presents a comprehensive survey of 138 works, primarily published between 2015 and 2025, spanning both classical and learning-based approaches. The motion planning methods are grouped into five categories based on the concepts of sampling, graph search, model predictive control, learning, and additional classical local planning approaches, including velocity obstacles, potential fields and dynamic windows. The learning techniques include supervised learning and reinforcement learning. We also discuss the role of dynamic perception in motion planning, covering techniques for detecting and modeling moving obstacles using cameras, LiDAR, and event-based sensors. The survey analyzes the principles, strengths, and limitations of each method, with particular attention to challenges unique to dynamic environments, such as prediction uncertainty, human-robot interaction, and the freezing robot problem. The survey provides researchers with a structured understanding of motion planning methods in dynamic environments.
Representational Capacity: Geometric Limits on Feature Representation in Transformer Language Models
arXiv:2606.02765v1 Announce Type: new
Abstract: Model dimension ($d_{model}$) is a fundamental hyperparameter in transformer language models, yet its role in setting the geometric limits of feature representation remains under-explored. Grounded in the Linear Representation and Superposition Hypotheses - which propose that models encode features as near-orthogonal directions in latent space - we develop a framework for estimating how many such directions a model can support. We first establish the embedding matrix as a measurable proxy for near-orthogonality constraints across the latent space: the boundary between meaningful token relationships and incidental similarity in the pairwise cosine similarity distribution gives a concrete estimate of the model's accepted deviation $\varepsilon$ from perfect orthogonality. Applying this metric across dozens of open-source models reveals two classes: models with high $\varepsilon$ whose embeddings lack near-orthogonal structure, and models with low $\varepsilon$ that maintain it. We then show that the standard Johnson-Lindenstrauss lemma greatly underestimates the packing efficiency of trained representations, and derive an adjusted capacity formula in which the number of near-orthogonal directions depends on the ratio of vectors to dimensions ($k/d$) rather than the raw count - a single modification that cuts prediction error by two orders of magnitude with no extra parameters. Combining these results, we define representational capacity as an upper bound on the number of distinguishable directions available for features and embeddings in a model's latent space. Capacity is exponentially sensitive to $\varepsilon$, and larger models favor tighter orthogonality constraints over maximizing raw capacity - a pattern compatible with several explanations (a stability-capacity trade-off, a ceiling on usable concepts, or confounds with model scale) that we leave to future work.
arXiv:2407.05312v2 Announce Type: replace
Abstract: Diffusion models have demonstrated impressive image generation capabilities. Personalized approaches, such as textual inversion and Dreambooth, enhance model individualization using specific images. These methods enable generating images of specific objects based on diverse textual contexts. Our proposed approach aims to retain the model's original knowledge during new information integration, resulting in superior outcomes while necessitating less training time compared to Dreambooth and textual inversion.
arXiv:2606.03468v1 Announce Type: cross
Abstract: Recently, industrial pioneers like Amazon, Tencent, ByteDance, and Huawei have been adopting BBR as their congestion control algorithm for live-streaming applications, including TikTok Live. However, BBR, originally crafted for bulk data transmission, faces multiple challenges in live-streaming scenarios. In this paper, we first explore two key issues associated with BBR due to inaccurate bandwidth estimation in live-streaming scenarios: (i) BBR cannot easily exit its startup phase, resulting in a fierce self-inflicted loss. (ii) BBR sends data at a lower rate than the available bandwidth during its stable phase. We then propose BBR-Copilot, an auxiliary congestion control component that cooperates with BBR, making BBR better adapt to live-streaming scenarios. BBR-Copilot allows for proactively generating accurate bandwidth measurement samples by smartly creating and sending extra data. We implement the BBR-Copilot prototype upon QUIC and evaluate it via testbed. Experimental evaluation results show that BBR-Copilot effectively enhances BBR's performance in live-streaming scenarios.
arXiv:2606.02745v1 Announce Type: new
Abstract: Vision-language-action models (VLAs) are promising general-purpose robot policies, but adapting them to new tasks typically requires costly task-specific teleoperation data. As an alternative, we study one-shot demo-conditioned VLAs, where a robot policy is conditioned on a single demonstration video of an unseen task. We find that existing end-to-end approaches often struggle when successful execution requires precisely localizing small target regions. To address this limitation, we propose SeeTraceAct, a demo-conditioned VLA framework that encourages precise spatial grounding through visibility-aware prediction of future end-effector traces. To enable reproducible evaluation with cross-embodiment demonstrations, we introduce and release RoboCasa-DC, a demo-conditioned extension of RoboCasa with episode-paired humanoid videos. Experiments on RoboCasa-DC and a real-world benchmark, where a Franka Panda arm is conditioned on human demonstrations, show that SeeTraceAct outperforms baselines, achieving the best success rate across all four RoboCasa-DC settings and improving real-world average success by 12.5 percentage points.
arXiv:2606.02757v1 Announce Type: new
Abstract: Advanced Air Mobility (AAM) uses electric vertical take-off and landing (eVTOL) vehicles to address urban congestion and emissions. However, corridor design, operation management, and separation standards remain underexamined for safe high-density operations. This paper applies the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to systematically review relevant literature from IEEE Xplore and Web of Science, focusing on publications from 2010 to 2024. A Context, Intervention, Mechanism, and Outcome (CIMO) framework guided the development of research questions. After screening 2,039 journal and conference papers, 62 articles met the inclusion criteria. The findings reveal a lack of integrated corridor design approaches, limited operational strategies, and reliance on standards originally designed for conventional aviation. A unified corridor design and separation definition frameworks and taxonomies are proposed to address these shortcomings, informing future investigations and operational frameworks for safe, efficient eVTOL operation deployment in urban settings.
arXiv:2606.03407v1 Announce Type: cross
Abstract: We give detailed analysis and circuit design of structure-preserving quantum algorithms for second-order linear evolutionary PDEs, including parabolic equations and hyperbolic equations with mixed Dirichlet, Neumann, and periodic boundary conditions and source terms. While prior quantum algorithms usually neglect the stability problem from the PDE-to-ODE reduction, our method-of-lines approach investigates the boundary lifting via Coons interpolation and boundary-aware discretization, so that the resulting semi-discrete systems are stable and compatible with efficient quantum ODE primitives. For the parabolic problem, we use a diagonal similarity transform to ensure the semi-discrete generator must have a positive semi-definite Hermitian part, and then solve the resulting ODE system by the optimal linear combination of Hamiltonian simulation (LCHS). For the hyperbolic problem, we rewrite the semi-discrete equation as an equivalent first-order system and solve it by Hamiltonian simulation. We implement our quantum algorithms with explicit block-encoding constructions and circuit implementations, as well as demonstrating the end-to-end complexity bounds together with spatial and quadrature error estimates. We conduct classical numerical experiments on the convection-diffusion equation, inhomogeneous heat equation, and Klein-Gordon equation to validate our structure-preserving analysis and algorithmic constructions.
arXiv:2606.03292v1 Announce Type: cross
Abstract: Many systems used in real-world environments require adding new categories and incorporating new information without forgetting what was previously learnt by the classification model. This is known as class-incremental continual learning, and in the case of multivariate time-series, is further complicated by the temporal structure of the data. In this paper, we present a novel approach for performing class incremental continual learning for the classification of multivariate time series data based upon the construction of a dual-stream feature extraction pipeline (using both deep temporal embedding features generated via a pre-trained frozen foundation model and application of statistical features). Evaluated on five benchmark datasets, the proposed system achieves competitive average accuracy across all datasets while maintaining low forgetting rates across all experimental configurations.
arXiv:2606.02881v1 Announce Type: new
Abstract: A theoretical model is developed to compute electronic transport coefficients extending from warm and dense to hot and dilute plasma conditions. This kinetic theory-based approach models strong Coulomb correlations by treating interactions using the potential of mean force, electron degeneracy using the Uehling-Uhlenbeck equation, and diffraction by computing cross sections quantum mechanically. The result provides a fast and accurate means to compute electrical conductivity,thermal conductivity and electrothermal coefficients, including contributions from electron-electron interactions. The model enables accurate calculation of materials properties in many warm dense matter systems, including inertial confinement fusion, stellar evolution, and high energy density plasma experiments.
arXiv:2606.03245v1 Announce Type: cross
Abstract: Concepts of calibration formalize the compatibility between probabilistic predictions and the respective outcomes. In a nutshell, the outcomes ought to be indistinguishable from random draws from the predictive distributions. In this paper, we review, extend, and bridge notions of calibration that have been proposed for classification and regression tasks. Particular emphasis is given to hierarchical relations between the various notions, as they apply to general real-valued data, continuous outcomes, count data, nominal classes, and binary outcomes. To highlight a number of contributions, we introduce the notion of modal calibration for nominal outcomes, we distinguish full, partial, and average calibration in this setting, and we show that double probability integral transform (PIT) calibration is logically independent of previously proposed concepts of calibration for discrete outcomes. Furthermore, we generalize extant results on concepts of calibration that are expressed in terms of properties or functionals of the predictive distributions, such as means, quantiles, or event probabilities. Throughout the paper, we illustrate the concepts and their hierarchical relations in worked examples, and we provide algorithmic tools that support the construction of instructive examples and counterexamples.
arXiv:2606.02878v1 Announce Type: new
Abstract: A novel power delivery framework, comprising a package-embedded inductor topology and an inductance-island methodology, is introduced to maximize both inductance and current densities in vertical power delivery (VPD). The framework leverages multiple multi-phase converters, a common strategy in high-performance computing systems, to enhance efficiency and scalability. The proposed topology employs an array of tightly coupled spiral square inductors sharing a common magnetic rod, serving multiple converters operating in the same conversion phase. The array is optimized to maximize coupling and minimize conversion losses, achieving superior inductance and current densities of 250 nH/mm^2 and 10 A/mm^2, respectively. At the system level, the inductance-island methodology partitions the power delivery network into multiple islands, each dedicated to a converter phase and supplying a portion of the load current, thereby enabling scalable and efficient distribution. To validate the framework, the inductor array is designed and simulated in ANSYS Maxwell 3D and Mechanical, exhibiting an average quality factor of 23.6 and efficiency of 97.4% at 2 A load current, 6 V input, and 10 MHz switching frequency. The inductor array netlist is extracted from ANSYS and co-designed in Cadence Virtuoso with a distributed dual-phase power conversion system, ensuring joint optimization of passive and active components. The co-designed converter achieves a significant efficiency gain of 5.65% on average and up to 11.04% at 40 A load over a similar converter with uncoupled inductors, demonstrating the practical benefits of the approach.
arXiv:2606.02887v1 Announce Type: new
Abstract: Symmetric nonnegative matrix factorization (Symmetric NMF) approximates a matrix as $WW^T$ with nonnegative rectangular factor $W$. It has broad applications in graph clustering and machine learning. In contrast to the NMF, projected gradient methods for the symmetric problem had been associated with slow convergence. To address this, we introduce SNMPBB, the first adaptation of nonmonotone projected Barzilai-Borwein methods to Symmetric NMF, demonstrating that gradient algorithms are significantly more effective than previously understood. We further extend SNMPBB to graph clustering using the graph Laplacian regularization (Graph-SNMPBB) and to large problems with low-rank approximations (LAI-SNMPBB). For all variants we prove global convergence to first-order stationary points and also that Barzilai-Borwein curvature information is preserved with randomized approximations. On synthetic data, SNMPBB achieves 6 times speedup over the alternative SymANLS for similar residuals, with advantages growing at higher ranks. Across six real-world clustering benchmarks, Graph-SNMPBB matches or exceeds SymANLS accuracy. Lastly, LAI-SNMPBB outperforms state-of-the-art LAI-SymPGNCG on 34 SuiteSparse matrices in both runtime and residual quality.
arXiv:2606.02888v1 Announce Type: new
Abstract: The effectiveness of a soft wearable back-support device in enhancing postural stability was investigated under trip-like perturbations using two experimental paradigms: perturbed standing and perturbed walking. Healthy subjects completed trials under three different back-support conditions: no device, device worn with low stiffness, and device activated with high stiffness. Whole-body stability was quantified using the minimum Margin of Stability (MOS) at the point of maximal instability. Results demonstrated increased MOS during device use, indicating enhanced postural stability. In standing, MOS increased significantly with device stiffness, whereas in walking, both device conditions improved MOS relative to no device but did not differ significantly from each other. These findings highlight the potential of soft wearable back-support devices with adjustable stiffness to improve reactive balance control against external perturbations, with important implications for fall prevention. Future research should explore personalized stiffness optimization and evaluate efficacy in populations at elevated risk of falls.
arXiv:2606.03050v1 Announce Type: new
Abstract: Remote photoplethysmography (rPPG) enables non-contact extraction of blood volume pulse (BVP) signals using consumer-grade cameras. Recent unsupervised rPPG methods learn BVP representations without requiring ground-truth physiological annotations, yet their optimization is often hindered by noisy and unstable gradients, resulting in slow convergence and limited cross-domain generalization. In this paper, we propose FCUS-rPPG, a fast-converging unsupervised rPPG framework with strong generalization capability. Motivated by the observation that BVP representations exhibit both multi-spectral covariation and low-dimensional manifold structure, we design a spectrally shared backbone that facilitates BVP feature disentanglement while improving optimization efficiency. To jointly enhance convergence stability and generalization performance, we further develop a unified optimization framework operating at the gradient, loss-landscape, and feature-representation levels. Specifically, a post-verification masking mechanism filters out misleading gradients according to the weak-amplitude physiological prior of BVP signals; a perturbation-based loss landscape smoothing strategy steers optimization toward more generalizable flat minima; and a noise-aware null-space regularization constrains feature updates to the orthogonal complement of the noise subspace, thereby mitigating noise-induced representation drift. Extensive experiments on five datasets demonstrate that FCUS-rPPG requires only one training epoch, whereas existing methods typically require tens to hundreds of epochs. Notably, FCUS-rPPG consistently achieves state-of-the-art (SOTA) performance in cross-dataset evaluations. This study provides an efficient and robust solution to the real-world deployment of unsupervised rPPG. The source code will be publicly available at https://github.com/JiaJieLee/FCUS-rPPG.
arXiv:2606.02905v1 Announce Type: new
Abstract: Public broadcasts are at the center of civic discourse: Traditional television talk shows, alongside emerging podcast and web video formats, capture and guide the attention of our societies, shaping how citizens encounter politics, science, and societal issues. Yet, systematic or even simple analyses of these formats face similar challenges: guest and content metadata are scarce, fleeting, fragmented, and not standardized. Research conducted and questions answered are based on extensive, laborious, yet isolated data-curation efforts that capture only a fraction of the relevant landscape. This work seeks to address this issue using a scaling-oriented framework for FAIR data curation in public broadcasting. Evaluated on 15 broadcasting programs, the pipeline aggregates ZDF Archive PDFs, fernsehserien.de, and Wikidata into a unified knowledge graph. Of the 31,817 candidate guest mentions from these three sources, 17,729 could be automatically disambiguated, further 5,958 via 64 hours of manual reconciling using OpenRefine. Results are published at speakermining.wikibase.cloud and linked to Wikidata, enabling SPARQL-based question answering based on gender, age, occupation, or institutional affiliation across 8,436 canonical persons with 23,527 appearances in 6,469 aligned episodes. Our iterative experience reveals that correctly disambiguating and deduplicating speaker data from heterogeneous sources demands dedicated effort on sustainable infrastructure. For scalable and reliable question answering on public broadcasts to be accessible to everyone, we recommend fostering the potential of linked open data: Advancing alignment and utilization approaches like this work, particularly towards crowdsourced development and curation, but also more FAIR data interfaces from public broadcast service providers.
arXiv:2606.02913v1 Announce Type: cross
Abstract: In this study, we conduct a comprehensive comparative analysis of generative and discriminative deep learning-based speech enhancement methods, specifically in noise reduction tasks. Our investigation focuses on evaluating their effectiveness under high and low signal-to-noise ratio conditions, considering both matched and mismatched training scenarios. We further investigate the impact of training data volume, model convergence speed, and interpret the performance differences in terms of objective results for the considered training paradigms. Additionally, we compare the complexity-performance trade-off and the practical viability of these approaches. To further strengthen the evaluation, we study the hallucination characteristics of generative approaches in terms of word error rate and phoneme similarity. The insights derived from this study provide empirical evidence to assist researchers and practitioners in understanding whether the perceptual gains of different approaches justify their computational cost in practical applications.
arXiv:2606.02909v1 Announce Type: cross
Abstract: Gradient observations can substantially improve Gaussian process (GP) surrogates, particularly in high-dimensional settings where function evaluations are expensive. However, exact inference with $n$ function values and $n$ full gradients in $d$ dimensions scales cubically in the joint state size, imposing an intractable $\mathcal{O}(n^3 d^3)$ computational bottleneck. We introduce TERA, a highly scalable derivative GP method based on target-specific exact gradient reduction. We prove that for stationary kernels, the gradient components orthogonal to the directions connecting the target and conditioning points are conditionally independent of the target function value; consequently, the exact conditional density is fully characterized by at most $m^2$ directional derivatives once a conditioning set of size $m$ is specified. By using these reduced, dimension-free conditionals as local factors in a Vecchia approximation, TERA effectively decouples $n$ and $d$ from the dense matrix inversion. This reduces the per-target evaluation cost to $\mathcal{O}(dm^2 + m^6)$ time and $\mathcal{O}(dm^2 + m^4)$ memory, leaving the underlying derivative GP model mathematically unchanged. Empirical evaluations demonstrate that TERA achieves state-of-the-art predictive accuracy while operating orders of magnitude faster than standard derivative GPs. Crucially, both computation time and peak GPU memory remain essentially flat with respect to $d$, enabling highly scalable inference in high-dimensional spaces.