Machine Learning (Statistics)
Ordinary Nonconvex SGD under Distance-Dependent Moments: Finite-Horizon Stationarity and Nagaev Bounds
Wei Biao Wu
Uniform noise-moment bounds exclude
stochastic gradients whose variability increases with the iterate. We study ordinary, single-sample
stochastic gradient descent for smooth, lower-bounded, possibly nonconvex objectives under distance-dependent conditional moments. Under second moments alone, a direct descent–displacement argument yields $T^{-1/3}$ expected average squared-gradient stationarity with a horizon-dependent stepsize. An explicit oracle-complexity corollary matches the known smooth Blum–Gladyshev (BG-0) lower bound, including the $Lb_2\Delta^3\varepsilon^{-6}$ and $L\Delta\sigma^2\varepsilon^{-4}$
stochastic terms, where $\Delta$ is the initial objective gap and $\sigma^2+b_2\|x-x_1\|^2$ bounds the variance. Thus unchanged SGD attains the minimax
stochastic complexity in this second-moment class. For $p>2$, predictable localization and a
Hilbert-space Fuk–Nagaev inequality yield a high-probability bound separating logarithmic variance and
polynomial rare-shock contributions. The localization radius is derived from the recursion: no bounded-iterate assumption, clipping, normalization, momentum, or increasing batch size is needed. We also give increasing-confidence rates, an objective-gap-growth refinement recovering root-$T$ stationarity, and
stochastic $L^p$-Lipschitz examples. The broad BG-0 optimality statement is distinguished from the smaller mean-square-smooth class, in which additional oracle structure permits faster
algorithms.
Read the full abstract ▾
Machine Learning
Learning to Bias: Machine Learning-Enhanced Particle Filters
Apoorv Srivastava and Eric Darve
Sequential inference estimates latent states from noisy and incomplete observations. Particle Filters (PFs), a class of Monte Carlo methods based on importance sampling, provide a flexible framework for this task, but often suffer from poor sample efficiency and unfavorable scaling with dimension, partly due to suboptimal proposal distributions. We address these challenges by integrating learned proposals into the PF framework. We introduce Neural Optimal Particle Filters (NOPFs), which learn an amortized approximation to the optimal proposal from offline simulated one-step conditioning tuples. The learned proposal is used as a drop-in replacement in standard PF updates, with samples corrected by standard importance weights so that the method asymptotically targets the same filtering distribution under standard support and density-evaluation assumptions. Across
stochastic nonlinear
benchmarks of varying inference complexity, NOPFs improve sample efficiency and distributional accuracy over standard PF baselines with modest computational overhead. The approach integrates data-driven proposal learning into classical inference without altering the underlying filtering objective.
Read the full abstract ▾
Distributed Computing
Unifying In-Memory Data Analytics through Sparse Compilation
Anand Jayarajan and Gennady Pekhimenko
As modern data analytics workloads become increasingly heterogeneous and hardware-intensive, achieving efficient multi-core performance across diverse applications remains an open challenge. We present Reffine, a compiler-based in-
memory analytics engine that delivers high performance across a broad range of data analytics workloads. Reffine introduces a novel intermediate representation (IR), grounded in relational
algebra and sparse iteration theory, that provides a unified abstraction for data and computation. This representation enables workload-agnostic, end-to-end optimizations such as operator fusion and automatic parallelization across diverse analytics applications. We further develop a sparse compiler backend that translates Reffine IR into hardware-efficient imperative code, achieving high multi-core performance without domain-specific implementations. On the TPC-H
benchmark, Reffine outperforms the in-
memory analytical database DuckDB by up to $24.9\times$ and the state-of-the-art compilation-based database Umbra by up to $3.2\times$. Reffine also achieves average speedups of $18.3\times$ and $47.9\times$ over Polars and NetworkX on streaming and graph analytics workloads, respectively. Source code: https://github.com/ampersand-projects/reffine
Read the full abstract ▾
Quantum Physics
Disentangling Expressibility, Symmetry Protection, and Hardware Noise in Variational Quantum Simulation of the Two-Flavor Schwinger Model
Karthikeya Machiraju, Krishna Sujith and Kaustav Bhowmick
Existing
quantum simulations of the two-
flavor Schwinger model have run at a single
lattice size, and it is not known how far the variational approach can be pushed or which weakness stops it first. Following the model from N = 2 to 6 staggered
lattice sites, we find that the binding constraint at reachable sizes is hardware noise rather than
circuit expressibility or trainability, and identify N = 3 as the immediately viable extension of existing trapped-ion experiments. The energy error of a charge-conserving ansatz collapses onto one function of p/d, the ratio of variational parameters to physical-sector dimension, and falls by more than two orders of magnitude as p/d rises through order unity, giving the expressibility condition L(4N - 1) >= binom(2N,N) for L
circuit layers. The condition is local in chemical potential: at N = 3 the layer count sufficient at zero chemical potential leaves a 74.38% error near the first-order boundary, while one further layer reaches 0.08%. Charge conservation also protects trainability and prevents charge-sector leakage: as the
qubit count doubles from 4 to 8, the normalized gradient variance falls to 1/3.56 of its starting value for the constrained ansatz, versus 1/13.57 for an unconstrained
circuit. Comparing a global contraction with per-gate local noise, a fixed-parameter control shows that the noise model, not whether the optimizer runs inside the noisy loop, sets how strongly noise degrades the first-order transition. At N = 4, a noiseless control reaches 0.12% mean error, whereas the same
circuit at 1.00% depolarizing noise reaches 25.69 to 52.81%.
Read the full abstract ▾
Robotics
Memory-Aware Multi-Sensor Perception for Efficient and Safe Navigation in Dynamic Environments
Jingshuo Li, Yifan Xue, Yifei Li et al. (+2 more)
Autonomous
navigation in previously unseen environments requires effective perception, persistent environmental representation, and collision avoidance while maintaining progress toward a goal. Existing perception-based methods often rely on prior maps or short-horizon observations, limiting their ability to exploit previously observed structure. We propose a
memory-aware multi-sensor
navigation framework that integrates
LiDAR and RGB perception, online distance-field representation learning, and a stage-adaptive Modulated Control Barrier Function Quadratic Program (MCBF-QP). The framework persistently represents static infrastructure while tracking dynamic obstacles, enabling the MCBF-QP controller to exploit previously observed geometry for obstacle circumvention and adapt its safety constraints and guidance to local conditions. Experiments in complex indoor and outdoor environments demonstrate improved
navigation efficiency and goal-reaching performance while maintaining collision avoidance in narrow passages and around dynamic obstacles.
Read the full abstract ▾
All authors Jingshuo Li, Yifan Xue, Yifei Li, Shubhodeep Shiv Aditya, Nadia Figueroa
High Energy Astrophysical Phenomena
Spectral and Polarization Properties of Coherent Radio Emission from Relativistic Magnetized Pair Plasma Shocks
Yuanhong Qu, Joonas Nättilä and Lorenzo Sironi
Fast
radio bursts (FRBs) are millisecond-duration
radio transients whose emission mechanism remains an open question. The so-called synchrotron maser instability in relativistic magnetized shocks is a leading candidate for FRBs produced outside the compact object magnetosphere, yet its spectral and polarization predictions have not been systematically explored. We present three-dimensional particle-in-
cell simulations of relativistic magnetized pair
plasma shocks with upstream magnetization $\sigma > 1$ and bulk Lorentz factor $\gamma_0 = 10$, assuming a plane-parallel shock geometry and neglecting radiative cooling, and compute the spectra and polarization of the precursor maser emission for an arbitrary line of sight. For on-axis observers aligned with the shock-propagation direction, the X-mode dominates the emission. The O-mode power increases at off-axis viewing angles and is primarily associated with currents parallel to the upstream background
magnetic field. Assuming a spherical shock, we find full-width-at-half-maximum fractional bandwidths of $\Delta\nu/\nu_0\simeq0.57$ and $0.82$ for $\sigma=3$ and $6$, respectively. For highly magnetized shocks, the intrinsic emission bandwidth accounts for most of the spectral width, while high-latitude contributions provide additional broadening. The integrated emission remains highly linearly polarized near the spectral peak, with a nearly constant polarization angle due to the dominance of the X-mode. Our results suggest that synchrotron maser emission can produce spectra with bandwidths and high linear polarization degrees compatible with observations of some non-repeating FRBs, and may provide testable signatures for distinguishing shock-powered FRBs from other emission mechanisms.
Read the full abstract ▾
Databases
Batched Feedback and the Random-Access Wall in Search-Based Graph Construction
Édgar Chávez
Navigable graphs for nearest-neighbor search are built either incrementally, each inserted point searching a graph that mutates as construction proceeds, or in batch over a fixed substrate, which buys determinism and parallelism at a price in build time. We measure that price and find where it comes from. Instrumenting a tuned Vamana and PiPNN and building every system repeatedly in a paired design on one 64-thread machine, we separate build time into work (distance evaluations) and cost per evaluation. Letting the batch builder's substrate mutate in $B$ synchronous blocks recovers the feedback loop of incremental construction while the graph stays a function of (data, seed, parameters, $B$): eight trees with batched feedback match thirty-two frozen trees, and the build does $0.88\times$ Vamana's distance work. Yet it takes $1.55\times$ Vamana's wall-clock, because the mutating substrate costs more per evaluation. Pushing further, we find a wall that no search-based builder crosses. Every one of them, incremental or batched, evaluates distances adaptively, one dependent random access at a time, and runs $20$-$27\times$ below a dense kernel on the same machine at $d \approx 100$, five to six times of it with the data resident in cache. The gap is not a low-dimension artifact: an adaptive evaluation costs $d^{1.04}$ and a blocked one $d^{0.63}$, so the wall grows as $d^{0.4}$ and is twice as high at $d = 960$ as at $d = 128$. PiPNN's order-of-magnitude build advantage is that kernel: it evaluates as many distances per point, but as fixed-in-advance dense blocks. We show the beam cannot be batched after the fact (the useful density of a lockstep block is 2-3%), state the wall as a two-ceiling roofline, and delimit it: it binds whenever the distance is a black box or the evaluation order is data-dependent.
Read the full abstract ▾
Computation and Language
Breaking Homogeneity: Diversifying Persona Sets for Creative LLM Outputs
Sang Bin Moon, Nicole Cho, Daniel Borrajo et al. (+2 more)
Language models often produce homogeneous responses to open-ended tasks; such homogeneity can spawn groupthink-the convergence of ideas toward a singular and potentially suboptimal decision. We formulate persona diversification as a set-level conditioning problem and study two orthogonal design choices: selecting versus generating personas, and space-filling versus frontier-seeking diversity. We instantiate this design space with four methods spanning coverage and dispersion subset selections, uniform-coverage sampling, and evolutionary persona generation. Evaluations on the Alternative Uses Task (AUT), Infinity-Chat, and Divergent Association Task (DAT) show the benefits of the proposed methods across tasks and creativity objectives. On AUT, evolutionary persona generation increases response diversity by 78.8%, originality by 26.1%, flexibility by 49.5%, and holistic creativity by 13.9% over task-only prompting, while maintaining 98.5% validity; on Infinity-Chat, it nearly doubles persona-induced response separation relative to random personas. Moreover, evolutionary personas compose with creativity-optimized prompting, further increasing its response diversity by 18.6% and creativity by 6.3%. These results establish persona-set geometry as a task-agnostic mechanism for eliciting divergent LLM outputs, and support persona diversification as a reusable complement to
prompt optimization.
Read the full abstract ▾
All authors Sang Bin Moon, Nicole Cho, Daniel Borrajo, Sumitra Ganesh, Abolfazl Hashemi
Materials Science
Anomalous Pressure-Enhanced Polarization in Sliding Ferroelectrics
Lujin Min, Sahas Kamat, Ameer Mustafa et al. (+8 more)
Hydrostatic pressure commonly suppresses polarization in conventional displacive ferroelectrics by weakening the off-center ionic distortion. Sliding ferroelectrics may provide a contrasting case because their polarization arises from stacking-dependent interlayer charge transfer rather than intra-unit-
cell ionic displacement. Here, hydrostatic pressure is shown to enhance, rather than suppress, the polarization of parallel-stacked
bilayer boron nitride, a prototypical sliding ferroelectric. A
graphene layer placed on top of a ferroelectric boron nitride
bilayer is used for sensing the
evolution of polarization under pressure. The pressure enhances both the interlayer potential and the ferroelectric polarization, with the latter increasing approximately linearly by 80% up to 1.66GPa. These results establish hydrostatic compression as a tuning parameter for sliding-ferroelectric polarization, with potential implications for ultrathin slidetronic
memory devices.
Read the full abstract ▾
All authors Lujin Min, Sahas Kamat, Ameer Mustafa, Nguyen The Duy, Kyle Nadel, Junhao Lin, Kenji Watanabe, Takashi Taniguchi, Daniel Bennett, Brad J. Ramshaw, Kenji Yasuda
Quantum Physics
Construction of Partial Join Graphs with Perfect State Transfer in Shunt Decomposition-Based Quantum Walks
Banita Katuwal, Y Lakshmi Naidu, Srinath M S et al. (+1 more)
In this paper, we define directed partial join graphs with signed couplings and construct discrete-time
quantum-walk transition operators for these graphs using the shunt-decomposition framework. The resulting transition operators apply to several important graph families, including complete graphs with loops, circulant partial joins, complete bipartite graphs, tensor powers of complete bipartite graphs, all with signed couplings. For each family, we identify the corresponding structure of the transition operator and derive necessary and sufficient conditions for periodicity and perfect state transfer (PST), when one of the two directed regular graphs admits PST. Based on these results, we identify two types of state transfer; internal PST, which occurs between vertices within the same graph, and coupling PST, which occurs between two components of join graphs. We further develop a double-cover construction for directed partial join graphs and derive conditions for periodicity and PST when the associated transition operators do not necessarily commute. Using this construction, we establish PST results for double covers of complete graphs with loops. In particular, we provide an example in which the complete graph \(K_n\) does not exhibit PST for \(n\geq4\), whereas a suitable partial join of \(K_n\) exhibits PST when \(n=2^m\), \(m\geq2\). Hence, these results extend the class of graph families admitting PST in shunt-decomposition-based
quantum walks and provide a unified framework for studying
quantum state transfer in graph joins, products, and covers.
Read the full abstract ▾
All authors Banita Katuwal, Y Lakshmi Naidu, Srinath M S, Supriyo Dutta
Artificial Intelligence
BioEVAL: A global, multi-institutional benchmark of large language and multimodal models for bioengineering
Shun Ye, Vinny Chandran Suja, Chenlong Li et al. (+60 more)
Large
Language Models (LLMs) have demonstrated historic breakthroughs in general
reasoning with early successes in biomedical science. However, existing LLM benchmarking emphasizes factual recall, offering limited insight into model performance on frontier and
multimodal tasks. We assembled BioEVAL (BioEngineering Validation of AI and LLMs), a global, multi-institutional initiative designed to assess experimental
reasoning capability across bioengineering (BE) subfields. BioEVAL spans 11 major BE subfields plus a set of uncategorized items, bringing together 22 research groups to create a PhD-level
benchmark comprising 608 evaluation items: 1) 380 multiple-choice questions (MCQs, 359 retained after audit), 2) 218 literature synthesis tasks, and 3) 10
multimodal problems with experimental image interpretation.
Benchmark items underwent authoring-group expert review and centralized quality control before evaluation. Following evaluation, a blinded cross-group consensus audit of the highest- and lowest-accuracy MCQ items flagged 21 questions for revision or removal; these were withheld, and all reported MCQ results are computed on the 359 retained items. We evaluated diverse cloud-scale foundation/
multimodal models (e.g., ChatGPT, Gemini, and Grok) and locally deployable models suitable for inference on consumer-grade GPUs. Models achieved the highest accuracy of up to 90% on MCQs, similarity score of 0.72 on literature synthesis, and accuracy of 80% on a small sample of
multimodal reasoning questions, with substantial performance variation across subfields. Leaderboard rankings characterize current capabilities, limitations, and development priorities across the evaluated BE task categories. BioEVAL is maintained as an extensible
benchmark with standardized protocols for continuing expert item contribution and model evaluation.
Read the full abstract ▾
All authors Shun Ye, Vinny Chandran Suja, Chenlong Li, Chongming Jiang, Reza Zamani, Xiang Li, Christopher Bain, Yuqi Zhou, Walker Peterson, Huidong Wang, Chenglang Hu, Jongchan Park, Xiao Cheng, Benjamin Swedlund, Sandra Murillo, Anjali Sivanandan, Shiyu Sun, Liang Lanfeng, Mohammad Tariqul Islam, Baju C. Joy, Ishaq N. Khan, Sreedhar S. Kumar, Gabriel Mercado-Vásquez, James V. Vizzard, Jonathan M. Matthews, Helen Huang, Xiaolu Guo, Ethan Nicklow, Guorui Chen, Ryan A. Neff, Surjendu Maity, Hyeonjin Park, Han-ho Joo, Katherine Dong, Yuyan Cai, Weihang Huang, Yichen Zou, Rui Yan, Raphael Figueroa, Artem Goncharov, Bella Rose Schremmer, Lian Elsa Linton, Keisuke Goda, Liang Gao, Ke Cheng, Leonardo Morsut, Jennifer L. Wilson, Jianping Fu, Lim Chwee Teck, Deblina Sarkar, Andreas Hierlemann, Savaş Tay, Alexander Hoffmann, Donald Richieri Griffin, Jun Chen, Shana O. Kelley, Shyni Varghese, Jinwoo Cheon, Wilbur A. Lam, James J. Moon, Wilson W. Wong, Samir Mitragotri, Dino Di Carlo
High Energy Physics (Experiment)
Direct Neutrino Communication Through the Earth: NuMI-Calibrated Simulation and Far-Field Sensitivity
Sven-Patrik Hallsjö
Neutrinos can pass through substantial matter, potentially enabling direct links where electromagnetic paths are obstructed, but their weak interactions make reception difficult. We present two separate calculations. First, a repeated OOK/CRC simulation calibrated to the published NuMI-MINERvA selected-event mean correctly accepts 65.7% of simulated packets with a 40-bit payload and an eight-bit CRC after five repetitions, with about 587 s mean latency under assumed external synchronization. Varying the input mean across its approximate Poisson counting interval gives 60-72% correct acceptance; this is a sensitivity range, not a joint confidence interval. Second, a distinct idealized far-field model gives 24.4 MW peak
neutrino-carried on-slot power for a 10 kt receiver at 5,000 km, conditional on 3 GeV
neutrinos, 1 mrad divergence, 50% selection efficiency, unit
flavor survival, zero background and one-second slots. Neither result establishes a practical long-baseline source or measured CRC performance.
Read the full abstract ▾
Machine Learning
Geometric Feature Learning for Functional Data Valued on the Symmetric Positive Definite Manifold
Samuel V. Singh and Mimi Zhang
We here develop a functional neural network, termed MatFAE, for learning trajectories on the Riemannian
manifold of symmetric positive definite (SPD)
matrices. MatFAE features intrinsic layers that map
manifold-valued functions to Euclidean vector-valued functions, followed by a functional layer that projects them into a finite-dimensional Euclidean space. Unlike most neural networks for discrete-time sequences, MatFAE treats each sequence as a continuous function and can therefore encode trajectory dynamics (e.g., first-order derivatives) in its latent representations. Additionally, the morphology of the functional weights in the functional layer offers interpretability by revealing the regions of the input functional data that contribute most to the latent representations. We justify the design principles and properties of each intrinsic layer and detail how
matrix factorization is handled during backpropagation. We apply MatFAE to a range of fMRI
datasets, demonstrating its ability to efficiently learn informative representations from high-dimensional SPD trajectories and its practical value for real-world neuroimaging analysis.
Read the full abstract ▾
Materials Science
Revealing Polar Walls in CsPbBr3: Herringbone Structure and Formation Pathway
Weilun Li, Qimu Yuan, Michael B. Johnston et al. (+1 more)
Domain walls in ferroic materials can host nanoscale functionalities absent from the periodic
crystal, offering new opportunities to control electronic transport and polarization in
semiconductor devices. Metal halide perovskites demonstrate important photophysical properties; however, their ferroic properties remain unclear. Here we discover a hierarchical herringbone network of polar walls within the non-polar halide perovskite CsPbBr3 comprising ferroelastic twin walls and antiphase walls. Atomic-scale imaging reveals that both wall types comprise nanoscale regions whose inversion-
symmetry is broken compared to the surrounding bulk
lattice. In-situ heating experiments show that these interfaces form sequentially during
symmetry-lowering phase transitions, revealing a different formation pathway compared with 'conventional'
oxide perovskites. These polar nano-walls, together with their enclosed
topology, may impact carrier transport. This reveals a previously unrecognized structural motif in CsPbBr3, providing new insights into their photophysical and optoelectronic behavior.
Read the full abstract ▾
All authors Weilun Li, Qimu Yuan, Michael B. Johnston, Joanne Etheridge
Atmospheric and Oceanic Physics
Toward Actionable Predictability of Tropical Mesoscale Convective Systems through Large-scale Recurrent Environments in Northern South America
Vanessa Robledo, John F. Mejía, Jhayron S. Pérez-Carrasquilla et al. (+2 more)
Mesoscale convective systems (MCSs) produce much of the extreme precipitation, floods, and landslides that affect northern South America (NOSA), yet remain difficult to anticipate because they are strongly governed by local processes. On the night of 31 March 2017, an MCS triggered a catastrophic flash flood in Mocoa, Colombia —an event whose severity was not foreseen with sufficient lead time for emergency action, which typically requires several hours to a couple of days. Motivated by this disaster, this study investigates whether such events emerge within recurrent environments that provide a source of large-scale predictability, rather than as isolated and inherently unpredictable occurrences. We combine a 21-year (2001–2021) MCS tracking
dataset, ERA5 reanalysis, equatorial Rossby (ER) wave diagnostics, and a Self-Organizing Map (SOM) to identify recurrent environments associated with MCS occurrence in the Orinoco–Amazon basin. The Mocoa environment maps onto one of nine recurrent configurations characterized by enhanced low-level moisture, neutral Orinoco low-level jet conditions, favorable upper-level circulation, and strong ER-wave activity. We further trace convective precursors to an upstream source region over the Guiana Highlands up to 24~h before arrival in the Mocoa region. Together, these results show that the Mocoa event was not climatologically exceptional but a high-impact expression of a recurrent, partly predictable large-scale regime, demonstrating that portion of MCS predictability arise from recurring large-scale conditions that can be monitored hours to days in advance, and support hazardous weather
forecasting in NOSA and other tropical regions.
Read the full abstract ▾
All authors Vanessa Robledo, John F. Mejía, Jhayron S. Pérez-Carrasquilla, Mohamed Abdelkader, Humberto Vergara
Artificial Intelligence
Do LLMs Understand Context? A Knowledge Graph-Based Evaluation Framework
Subavarshana Arumugam, Mamta Nallaretnam, Kithuni Wickramasinghe et al. (+4 more)
While large
language models (LLMs) have achieved remarkable linguistic capabilities, a profound question lingers at their core: do these models truly comprehend context or simply excel at pattern matching on an unprecedented scale? Contextual understanding in LLMs refers to the ability to correctly extract relevant information from a given context, integrate it into a coherent internal representation, and reason over it to produce factually consistent and contextually grounded responses. However, traditional methods such as BiLingual Evaluation Understudy (BLEU) and perplexity simply measure surface-level performance. This reveals a critical gap in question answering (QA), where responses must be contextually grounded rather than simply being memorized associations. To fill this void, we propose a novel knowledge graph (KG) based evaluation framework for LLM contextual understanding in QA. Central to this is Semantic Structural Similarity for KGs (S3KG), a hybrid similarity measure combining structural and semantic signals into a single score. In addition, a diagnostic analysis framework is developed to identify and categorize
reasoning errors at the triplet level, enabling fine-grained analysis of model failures. Together, across nine
benchmarks, S3KG achieves F1 gains of up to $+7.6$ points over the strongest baseline and AUROC up to $0.973$.
Read the full abstract ▾
All authors Subavarshana Arumugam, Mamta Nallaretnam, Kithuni Wickramasinghe, Chamath Gunapala, Pragatheeswaran Vipulanandan, Kamal Premaratne, Uthayasanker Thayasivam
Sound
AcoustiClaim: A Numeric Claim Benchmark with Instrument Ground Truth
Sheng-Tse Lin, Siyuan Zhai, Chien-Liang Kuo et al. (+2 more)
Audio
language models state numbers for acoustic quantities, and neither human opinion nor a judge model says whether such a number is true of the signal. AcoustiClaim extracts each numeric claim from free text, scores it against the instrument that defines the quantity, and classes each quantity by where its reference can be read. Four open-weight systems and one closed model, asked for ten quantities five ways on two corpora, fill 207
cells. Of these, 49 emit fewer than five distinct values, and eight of the 158
cells that can be ranked exceed a rank correlation of 0.3, the bar we set, three with an interval clear of it, five of them one closed model reading pitch. Error sits at or above a constant-predictor floor in every ranked
cell but three. The reference decoder we train declines the five voice quantities in prose on 95% of mixtures, with nothing withheld, and states them on the clean twins, reproducing its targets' rule from audio alone. With a calibrated threshold, withholding lowers error on all ten quantities on the mixtures in the mean and on eight at every split, against at most 0.6% from a random selector. A linear baseline orders errors at least as well as ours. F0 s.d. and shimmer stay above the constant floor.
Read the full abstract ▾
All authors Sheng-Tse Lin, Siyuan Zhai, Chien-Liang Kuo, Massa Baali, Bhiksha Raj
Algebraic Geometry
Demazure Models of Algebraic Tori in Dimension 4
Nicole Lemire
A Demazure model of an algebraic $k$-
torus $T$ split by a finite Galois extension of fields $L/k$ and splitting group $G=\text{Gal}(L/k)$ is a smooth projective $T_L$-toric variety $X_{\Sigma}$ such that the natural action of $G$ on $T_L$ extends to $X_{\Sigma}$. A smooth projective model of the algebraic $k$-
torus $T$ can then be taken to be the quotient of a Demazure model by its splitting group. The existence of Demazure models of algebraic $k$-
tori was determined by Brylinski and further refined by Colliot-Thélène, Harari and Skorobogatov. Voskresenskii and Kunyavskii's birational classifications of algebraic $k$-
tori in dimensions 2 and 3 involved constructions of Demazure models of the algebraic $k$-
tori corresponding to maximal finite subgroups of $\text{GL}(r,\mathbb{Z})$ for $r=2,3$. We discuss some explicit constructions of Demazure models of algebraic $k$-
tori, making connections with the defining integral representations of their splitting groups. The constructions determine smooth projective Demazure models of the algebraic
tori corresponding to maximal finite subgroups of $\text{GL}(4,\mathbb{Z})$. Since projective toric varieties are determined by
lattice polytopes, our constructions focus on some highly symmetric families of polytopes, such as root polytopes and central transportation polytopes.
Read the full abstract ▾
Econometrics
Do Third-Party Web Traffic Estimates Preserve Causal Variation?
Mehrzad Khosravi and Hema Yoganarasimhan
Researchers increasingly rely on third-party platforms such as Similarweb and Semrush to measure web traffic when first-party analytics are unavailable. Yet these platforms report model-generated estimates rather than raw data, raising questions about whether their measures preserve the temporal and cross-source variation required for causal inference. As a motivating diagnostic, we examine reported referral traffic around two documented search-engine outages; the absence of visible discontinuities illustrates why preservation of identifying variation cannot be taken for granted. We then characterize three mechanisms, within-source smoothing, cross-source leakage, and treatment-induced calibration error, through which platform processing can generate nonclassical outcome measurement error. Analytical results and a stylized difference-in-differences simulation show that this error can attenuate, amplify, or reverse estimated treatment effects. Our findings caution against using third-party traffic measures based on black-box proprietary models for causal inference.
Read the full abstract ▾
Analysis of PDEs
Continuous Data Assimilation for Memristive FitzHugh–Nagumo Neural Networks
Jing Tian
In this study, we investigate continuous data assimilation for a class of partly diffusive FitzHugh–Nagumo neural networks with memristive feedback and linear synaptic coupling. The nudging
algorithm implemented here only adds a feedback term to the membrane-potential equations, while the recovery and memductance variables remain unobserved. We obtain weighted energy estimates that give sufficient conditions for exponential convergence of the data-assimilated solution to the reference solution under full or coarse membrane potential observations. Moreover, we prove energy-class uniqueness in dimensions one and two. In dimension three, the uniqueness criterion is an \(L^4\) membrane-potential initial data. We also establish an observational-noise estimate for locally square-integrable forcing, yielding an asymptotic error bound when the noise energy is eventually bounded.
Read the full abstract ▾
Robotics
Learning-Based Pressure Predictive Control of a Vertebraic Soft Robotic Tail
Wenjian Yang, Nan Huang, Yukang Nie et al. (+4 more)
Soft
robots have attracted much
attention for their safe human-
robot interaction and flexibility, but the typical continuum structure and nonlinear material behavior make the kinematics modelling complex, especially in non-static motions. In this work, we proposed an LSTM-based pressure predictive control (PPC) for the motion control of a vertebraic soft robotic tail and the coordination with a quadruped
robot. The PPC consists of an inverse kinematics (IK) model, a forward kinematics (FK) model and a pressure compensation (P-comp) model, and achieves non-static and quasi-static motion control of the tail. Compared with the IK-only model, the average RMSE of the PPC's simulation trajectories reduces by 69.8%, when executing target trajectories. In the coordinated motions of the soft tail quadruped, using a prediction data set to train the PPC enables next-moment action prediction and reduces computation time by 60.9%, which enhances the real-time response of the tail to match the quadruped torso's moving rate. The PPC provides a simple and effective method to model the soft tail for both non-static and quasi-static motion control, and grants the soft tail quadruped with the functionality of interacting with the environment.
Read the full abstract ▾
All authors Wenjian Yang, Nan Huang, Yukang Nie, Fang Chen, Wanchao Chi, Jiansheng Dai, Sicong Liu
Computer Vision
The Shape of Events: Edge-Based Inductive Biases via Cross-Domain Distillation
Soshun Kihara, Shunsuke Yasuki and Masato Taki
Convolutional neural networks trained on ImageNet are known to exhibit a strong preference for local high-frequency texture, an inductive bias that translates into fragile robustness against distribution shifts in real-world environments. Event cameras, in contrast, record only changes in scene brightness and are therefore well suited to capturing contour information; however, due to the absence of diagnostic
benchmarks in the event domain, the inductive bias that event-camera data instills in
vision models has remained underexplored. In this work, we use knowledge
distillation from the event domain to the RGB domain so as to exploit the rich evaluation toolkit available in the RGB domain and systematically dissect this inductive bias. Our experiments show that
distillation from the event domain induces, in the RGB domain, color invariance, shape bias, and robustness to high-frequency noise. We identify the underlying mechanism as the model suppressing its dependence on high-frequency texture while acquiring a stronger dependence on edge-based object shape. This hypothesis is supported by changes in how color and spatial information are processed at the early layers, together with a spectral trade-off in which robustness to the absence of high-frequency components coexists with vulnerability to contamination of the relied-upon frequency bands and to disruption of
geometric structure. We further show that this inductive bias differs from existing robustification methods and that it functions as a useful prior for diverse downstream tasks in which shape and contour information contribute alongside other cues. The code is available at https://github.com/snskysk/event2rgb-
distillation .
Read the full abstract ▾
Statistics Theory
Scaffold-Constrained Subset Dynamic Programming for Exact SSE Clustering
Yordan P. Raykov and Max A. Little
Exact Euclidean \(K\)-means partitions \(n\) observations into \(K\) unlabelled clusters, but the unrestricted search is generally exponential. We use data-derived
geometric graphs to precondition an exact subset dynamic program: as a result only connected vertex subsets are admitted as clusters, while sum-of-squared-errors (SSE) loss is unchanged. A remaining-set recurrence minimises fixed-\(K\) or penalised SSE, with exact factorisation over the connected components of each remaining set. The central question we study is how much computational support can be removed while preserving an unrestricted optimum. Graph inclusion gives monotone coverage and support relations, and a bottleneck threshold identifies the first covering graph in a nested hierarchy. For fixed \(K\) and dimension, under compact ball support and density bounds, retaining \(q=O(\log n)\) nearest neighbours per observation preserves an empirical SSE optimum with probability tending to one, using an \(O(\log n/n)\) fraction of complete-graph edges. Truncated Gaussian mixtures with unequal weights and covariances satisfy these conditions. The rate we provide is a sufficient upper bound rather than a result implying
polynomial optimisation complexity. Objective-matched synthetic and full-data comparisons assess coverage, compression, and reference-label agreement. As a secondary application, we illustrate how the proposed scaffold preconditioning can be utilized to improve the efficiency of split-merge proposals that preserve unrestricted mixture posteriors.
Read the full abstract ▾
Audio and Speech Processing
Asymmetric Classifier-Free Guidance for Target-Speaker ASR
Yiwen Guan and Jacob Whitehill
Target-speaker automatic
speech recognition (TS-ASR) must identify and transcribe a desired speaker under varying overlap and noise conditions. These changes alter the acoustic evidence for the target speaker in the
speech mixture, motivating inference-time calibration of speaker conditioning. We introduce asymmetric classifier-free guidance (CFG) for TS-ASR using Whisper: the speaker-conditioned branch predicts the target transcript, while the speaker-unconditioned branch predicts serialized multi-speaker transcripts. CFG adjusts the contribution of speaker conditioning during decoding through a single guidance scale. We select a global guidance scale on target-domain development data and train a lightweight encoder-based predictor to adjust it for each utterance, keeping the recognition model fixed. Under domain shifts, our full system achieves relative word error rate (WER) reductions of up to 21.8% over the condition-only baseline, and 5.6% over standard conditional decoding of the same CFG-trained model. Oracle analysis shows that substantially larger WER reductions are possible through utterance-level scale selection and identifies how beneficial adjustments vary with domain shifts.
Read the full abstract ▾
Human-Computer Interaction
Redesigning Trust: Replacing Dark Patterns with Fair Choice Architecture in Financial Interfaces
Oluwadamilola Awakan, Tawan Aroonwechkul, Roshan Gunjoor et al. (+2 more)
Digital financial platforms make enrollment effortless and cancellation laborious. This asymmetry is a dark pattern that manipulates users who have already decided to leave. Existing work identifies such patterns after deployment, and regulators sanction them after harm, yet neither provides designers with a criterion for building interfaces that avoid manipulation. We model the provider as an adversary whose instrument is effort and express
fairness as a constraint requiring that leaving never cost more than joining. Defining interaction cost over
navigation steps, mandatory inputs, and confirmation
prompts, we prove that this constraint holds for every assignment of effort weights if and only if no component of the exit flow exceeds its counterpart at entry.
Fairness is therefore verifiable by counting rather than by estimating cognitive effort, and exact equivalence is unnecessary because exit legitimately requires fewer inputs than entry. We instantiate the model, together with invariants for visual parity and linguistic neutrality, in a mobile credit card prototype with parallel sign-up and cancellation workflows.
Read the full abstract ▾
All authors Oluwadamilola Awakan, Tawan Aroonwechkul, Roshan Gunjoor, Supriya Khadka, Sanchari Das
Machine Learning
Mentored Decoding: Faster Inference meets Boosting
Vivien Tran-Thien and Richard Nock
Speculative decoding is a successful technique speeding up inference of a target autoregressive
language model via a fast drafter model. Lossy speculative decoding allows a drift with respect to the target to further improve speed. Interestingly, it has been observed experimentally that the resulting model can $\textit{also}$ beat the target $\textit{quality-wise}$. Our paper formally proves how such a feat is possible with a formal approach to lossy speculative decoding called $\textit{mentored decoding}$. To get there, we connect inference to a celebrated ML training theory, $\textit{boosting}$, and proceed via the generalization of mentored decoding to the whole set of $f$-divergences. We uncover key properties of mentored decoding, among which (i) the particularly appealing
geometric nature of the total variation case, (ii) simple approximations for any $f$-divergence in direct relation with boosting compliance, and (iii) a $\textit{divergence independent}$ $O(n)$ space and $O(\mathrm{sort}(n))$ time data structure built on drafter and target outputs, which allows to query the optimal parameters of the dual problem in $O(\log n)$ time and constructing optimal mentored distributions in $O(n)$ time for any $f$-divergence.
Read the full abstract ▾
High Energy Physics (Theory)
Divergences in Nonpolynomial Quantum Field Theories
Galib Hoq and Thomas Curtright
The structure of divergences in several nonpolynomial scalar
quantum field theories are investigated. Various $n$-point Green functions for nonpolynomial theories are calculated and their renormalization is analyzed, with special regard for the nonpolynomial $\left|\phi\right|^{3}$ theory in $4$ spacetime dimensions.
Read the full abstract ▾
Machine Learning
Moment-guided edge sampling
Weibin Cai and Reza Zafarani
Edge sampling makes local decisions to achieve graph-level objectives, such as preserving structural properties. This creates a fundamental challenge: how can the effect of a local edge edit (i.e., edge addition or removal) on global graph structure be quantified and controlled? We address this challenge with a moment-guided edge sampling framework based on spectral moments of the random-walk transition
matrix. We compute exact moment changes through two complementary methods: a combinatorial method with closed-form updates for low-order moments, and a low-rank method that exploits locality and cyclic trace invariance to compress computations to edited endpoints, supporting arbitrary moment orders and batched edits. For single-edge edits at fixed moment orders, the low-rank method reduces the cost from $O(mn)$ to $O(m)$, while the combinatorial method evaluates low-order changes in constant time given maintained local statistics. These moment changes provide interpretable structural signatures of local edge motifs that aggregate into graph-level fingerprints. This structural meaning motivates us to ask whether preserving moments also preserves the graph properties. We further derive and validate that moment-preserving sampling can retain related structural properties, including triangle-weighted clustering coefficient. These structural insights enable analysis and improvement of graph learning: different edge structures have distinct effects on supervised node classification, while moment-guided augmentation is competitive for graph contrastive learning. Together, these findings establish moments as an interpretable and controllable bridge from local edge edits to global graph structure and learning.
Read the full abstract ▾
Computation and Language
CARGO: Context-Aware Retrieval-Gated Evaluation of Agentic AI in Production
Mukul Chhabra, Shail Patel and Luigi Medrano
Reference-based LLM-as-a-judge evaluation assumes the reference answer is the target. In deployed agentic systems that operate over dynamic entities (support cases, assets, accounts), the closest available reference typically applies the correct procedure to a different entity, so a literal judge penalizes different identifiers, dates, and statuses as errors or
hallucinations. We name this failure mode reference-instance divergence (RID). We propose CARGO, a framework that (i) treats retrieved references as procedural exemplars and grounds factual judgments in the live instance's observed context, (ii) assigns each claim a three-way status (supported, contradicted, unverifiable) and penalizes only contradictions, and (iii) gates evaluation by
retrieval confidence, casting production evaluation as selective prediction. We introduce CARGO-Bench, a perturbation-based diagnostic suite with ground truth by construction that separates leniency from discrimination. On CARGO-Bench (246 items, two judge models, 7,872 judgments), the standard reference-based judge penalizes 100% of correct entity-transplanted answers and is uninformative (discrimination index DI ~ 0); supplying the live facts without reframing changes nothing. CARGO eliminates these false penalties (0/50) while retaining near-complete contradiction recall (50/50 and 49/50), raising DI to 0.58 [0.48, 0.68]; a rubric-swap control attributes most of the effect to context-grounded dimension definitions. CARGO also exposes a limitation of its own design: the leniency that protects entity values suppresses detection of procedural corruptions (20% recall). A post-hoc fix does not close the gap, and an LLM-as-annotator study with written guidelines and adjudication shows the same blind spot. We release a preregistered protocol for extending the evaluation to expert agreement, risk-coverage, and cost on production traffic.
Read the full abstract ▾
Machine Learning
Reliability-aware Cross-sample Enhancement for Robust Multimodal Sentiment Analysis
Menghua Jiang, Haokai Gao, Xiangui Kang et al. (+2 more)
Multimodal Sentiment Analysis (MSA) aims to infer human
emotions from multiple modalities such as text, audio, and
vision. In practice, inputs are often corrupted by noise and missing modalities, which degrades performance. Existing methods typically address these challenges in isolation, limiting their effectiveness in realistic settings. To address this limitation, we propose a Reliability-aware Cross-sample Enhancement (RCE) framework. Specifically, RCE first introduces an adaptive variational information bottleneck to model modality-wise uncertainty and perform quality-aware information compression, thereby suppressing redundant noise in unreliable modalities. Furthermore, we design a reliability-aware cross-sample enhancement strategy that retrieves high-confidence, semantically consistent neighbors from a large candidate pool to enrich and calibrate current representations, effectively alleviating information deficiency caused by missing modalities. Building upon this, RCE integrates cross-modal interactions with a multilevel reliability-aware fusion mechanism to adaptively aggregate information across modalities and enhancement stages, leading to more robust
multimodal representations. Extensive experiments demonstrate that RCE consistently outperforms state-of-the-art methods across full, noisy, and missing-modality settings.
Read the full abstract ▾
All authors Menghua Jiang, Haokai Gao, Xiangui Kang, Haifeng Hu, Sijie Mai