arXiv subjects

Yi Zhao

Publications and source records attributed to Yi Zhao.

At least 19 recordsLinked to original sources

arXiv2026

Sparsely Supervised Diffusion

Diffusion models have shown remarkable success across a wide range of generative tasks. However, they often suffer from spatially inconsistent generation, arguably due to excessive correlations learned by the model. This can produce samples that are locally plausible but globally inconsistent. We propose a principled method to mitigate this issue. Our method, sparsely supervised diffusion (SSD), is a simple yet effective masking framework that can be implemented with only a few lines of code. We analytically show that SSD fundamentally alters diffusion model training by modifying the spectrum of the data covariance and that it suppresses correlations in the data covariance matrix. Experiments show that our method more accurately approximates the underlying population score function, reduces memorization on small datasets, and promotes the use of essential contextual information during generation. Moreover, even when up to 98\% of pixels are masked, it achieves competitive FID scores across a range of datasets and, importantly, avoids the training instability commonly observed on small datasets.

cs.LG↗

arXiv2026

Number of independent transversals in multipartite graphs

An independent transversal in a multipartite graph is an independent set that intersects each part in exactly one vertex. We show that for every even integer $r\ge 2$, there exist $c_r>0$ and $n_0$ such that every $r$-partite graph with parts of size $n\ge n_0$ and maximum degree at most $rn/(2r-2)-t$, where $t=o(n)$, contains at least $c_r t n^{r-1}$ independent transversals. This is best possible up to the value of $c_r$. Our result confirms a conjecture of Haxell and Szabó from 2006 and partially answers a question raised by Erdős in 1972 and studied by Bollobás, Erdős and Szemerédi in 1975. We also show that, given any integer $s\ge 2$ and even integer $r\ge 2$, there exist $c_{r,s}>0$ and $n_0$ such that every $r$-partite graph with parts of size $n\ge n_0$ and maximum degree at most $rn/(2r-2)- c_{r, s} n^{1-1/s}$ contains an independent set with exactly $s$ vertices in each part. This is best possible up to the value of $c_{r, s}$ if a widely believed conjecture for the Zarankiewicz number holds. Our result partially answers a question raised by Di Braccio and Illingworth recently.

math.CO↗

arXiv2026

BriLLM-MetaPred: A Representation-Prediction Unity Language Model with Adaptive Self-Measurement

We present BriLLM-MetaPred, the first engineering implementation of the MetaPred machine-learning framework established by a mathematical theory of AGI, instantiated for language modeling. The theory argues that MetaPred is both necessary and sufficient as a machine-learning primitive for general cognition. Taking this theoretical conclusion as a design premise, we report an implementation of the primitive for language modeling. A single continuously evolving dynamical state serves as both the context representation and the state from which predictions are read out. Continuous measurement of its own state reduces finite-depth inconsistency, maintains internal consistency, and provides an endogenous dynamical self-supervision signal in fixed-point form. The single dynamical state of MetaPred intrinsically couples training and inference. MetaPred therefore exhibits symmetry across almost all key elements of machine learning, an intuitive and observable property of the mathematical optimum over the space of all machine-learning paradigms. BriLLM-MetaPred is neither another recurrent language model nor an extension of BriLLM. It provides a bridge between two prior contributions: the construction and motivation of the MetaPred primitive provided by AGI theory, and the SiFu learning paradigm provided by BriLLM (Zhao et al., 2025). The latter's node-signal-flow design enables the primitive to run on available software and hardware accelerators. Our implementation of BriLLM-MetaPred achieves predictive performance comparable to a GPT baseline of the same size. This provides initial evidence for the engineering feasibility of the MetaPred framework established by the mathematical theory of AGI. MetaPred is the first machine-learning paradigm derived entirely from mathematical theory rather than assembled and incrementally improved through human design---and perhaps the last such paradigm.

cs.LG↗

arXiv2026

Longitudinal Brain Aging in Covariance Space: Temporally Calibrated Bayesian Inference for Adaptive Covariance Operators

Functional brain aging is often characterized by differences between people of different ages, although aging unfolds within individuals. This distinction is especially important for functional connectivity, where the direction of covariance change may evolve across adulthood. We introduce a Bayesian longitudinal covariance-operator framework that models each visit covariance as a latent positive-definite state informed by repeated resting-state fMRI runs. The approach calibrates run-level information for temporal dependence and replicate variability, separates within-person aging from between-person age differences, and uses an adaptive tangent representation to learn the low-dimensional structure of longitudinal change without relying on pooled covariance variation. The longitudinal operator varies smoothly in magnitude and orientation with age, while exact Fréchet derivatives translate the fitted trajectory to the covariance scale. This framework allows us to examine how brain covariance organization changes within individuals, whether longitudinal aging follows the population age gradient, and which functional systems contribute to their divergence. In the Dallas Lifespan Brain Study, within-person covariance aging differs substantially from the cross-sectional age gradient through early and middle adulthood, with the discrepancy diminishing later in life. Covariance reorganization is strongest around midlife, and the longitudinal signal is concentrated in a few coordinated directions, with most off-diagonal change occurring between functional systems. These findings demonstrate that population-level signatures of brain aging do not necessarily reflect individual aging trajectories and highlight the importance of modeling longitudinal covariance change directly.

stat.ME↗

arXiv2026

Complete tripartite subgraphs of balanced tripartite graphs with large minimum degree

In 1975 Bollobás, Erdős, and Szemerédi asked what minimum degree guarantees an octahedral subgraph $K_3(2)$ in any tripartite graph $G$ with $n$ vertices in each vertex class. We show that $δ(G)\geq n+2n^{\frac{5}{6}}$ suffices thus improving the bound $n+(1+o(1))n^{\frac{11}{12}}$ of Bhalkikar and Zhao obtained by following their approach. Bollobás, Erdős, and Szemerédi conjectured that $n+cn^{\frac{1}{2}}$ suffices and there are many $K_3(2)$-free tripartite graphs $G$ with $δ(G)\geq n+cn^{\frac{1}{2}}$. We confirm this conjecture under the additional assumption that every vertex in $G$ is adjacent to at least $(1/5+\varepsilon)n$ vertices in any other vertex class.

math.CO↗

arXiv2026

Collapse, Repair, and Drift: Dissecting Fine Tuning Dynamics in Speech Language Models

Supervised fine tuning (SFT) is commonly monitored through training and evaluation loss, but loss does not fully characterize how model functionality evolves along the same trajectory. We study this gap through dialect adaptation of Qwen3-ASR, MiMo-Audio, and FireRed-LLM across learning rates from $10^{-5}$ to $10^{-3}$. Character error rate (CER) reveals pronounced transient collapse and recovery whose severity and temporal structure are only weakly reflected by loss. Bidirectional layer patching reveals strong cross checkpoint functional incompatibility during recoverable collapse, with model dependent depth profiles that reorganize during recovery. Pairwise gradient similarity alone does not reliably indicate optimization stability, whereas raw gradient spectral analysis provides a clearer structural distinction between stable and unstable optimization: successfully converged runs approach a broad and temporally stable spectral energy distribution, while failed runs remain spectrally concentrated or exhibit persistent fluctuations. Finally, AdamW update transfer bypasses pronounced early collapse, yet substantial Mandarin degradation still emerges later in training. These results motivate three separable SFT dynamics: collapse, repair, and drift, and show that transient collapse is not necessary for persistent source domain degradation.

cs.SD↗

arXiv2026

PulseBound: Future-Beat State Forecasting Under an Explicit Information Boundary

Predictive representation learning from photoplethysmography (PPG) can violate causal information access even with causal attention, as normalization, nonlocal transforms, or companion views may depend on withheld samples. We introduce PulseBound, a PPG representation learner combining physiologically structured future-beat prediction with an explicit stored-window information boundary. A content-independent cutoff separates the visible prefix from the prediction target. Prefix-only normalization, suffix replacement before derived-view construction, and aligned masking ensure that encoder inputs depend only on the visible prefix and cutoff. This yields stored-suffix invariance: with fixed model state, randomness, prefix, and cutoff, changing the stored suffix cannot change the forecast context. A shared horizon-conditioned head predicts nine rhythm and morphology descriptors for up to four extractor-valid future beats, using elementwise validity masks; optional ECG-derived pulse-arrival-time supervision is restricted to training. On MIMIC and VitalDB groups held out from PulseBound backbone pretraining, PulseBound reduces nine-state transformed-space MAE relative to last-visible-beat persistence by 28.06% and 22.22%, respectively, with gains in MAE, MAE-Skill, and Spearman correlation across all 40 source-cutoff-horizon cells. In a separate comparison of seven models on 13 downstream tasks, PulseBound achieves the best mean on nine frozen linear-probe and seven full-fine-tuning tasks. Stored-suffix interventions cause zero recorded changes in forecast contexts or predictions, with zero suffix-input gradients at audited precision under the stored-window interface. These findings separate three testable aspects of predictive physiological representation learning: information access, supervised future structure, and transfer.

cs.AI↗

arXiv2026

SemiCodec: Bridging Speech Tokenization and High-fidelity Reconstruction with Semi-discrete Representations

Speech representation learning requires a discrete modeling interface while preserving high waveform fidelity. Existing approaches rarely achieve both: discrete tokens ease modeling but lose fine-grained acoustic information, while continuous latents preserve fidelity but require specialized generative objectives. We introduce SemiCodec, a novel Semi-discrete neural speech Codec with synchronized discrete tokens and continuous latents. Direct joint optimization may create a reconstruction shortcut through the continuous pathway, leaving the discrete tokens insufficiently grounded. To address this issue, we propose a Preserve-then-Ground training paradigm that first establishes a high-fidelity synthesis space and then grounds SSL-guided discrete tokens within it. Experiments show that SemiCodec outperforms fully discrete codecs in reconstruction quality and intelligibility while remaining competitive with continuous representations in acoustic fidelity and speaker preservation. Audio samples are available at https://pennyw1-se.github.io/SemiCodec-Demo-Page/

eess.AS↗

arXiv2026

Magnetar X-ray flares as elementary reconnection events in a self-organized critical magnetosphere

Magnetar bursts are commonly attributed to magnetic energy release, yet the physically relevant dissipation unit has remained uncertain. Here we argue that this unit is not the burst as conventionally defined, but the short substructures within bursts, namely magnetar X-ray flares (MXFs). Using 789 MXFs from SGR~J1935+2154, we show that their waiting times separate into two components, interpreted as reflecting clustered intra-episode events and longer inter-episode intervals, that their proxy emission measure and temperature exhibit distinct rising and declining trends, and that their energy release scale (ERS) proxy distribution has a high-end slope in the same statistical regime as solar and stellar flares. Combined with the previously measured MXF duration slope, these results identify MXFs as a three-dimensional fractal-diffusive self-organized critical (FD-SOC) system with $β_{\rm diff}\simeq2$, implying reconnection avalanches that expand linearly through an active magnetosphere. We infer that magnetar active episodes are global unloading states assembled from many elementary flares. Magnetar X-ray flares are therefore the physically relevant units for magnetar outburst statistics and for connecting magnetar high-energy activity to other impulsive phenomena.

astro-ph.HE↗

arXiv2026

The Unexpired Plan: A Free Monitor for Accelerated Diffusion Policies

Training-free acceleration of a diffusion policy is accepted when an internal similarity signal reports that the shortcut changed nothing. We price every monitor a control loop can afford in closed-loop success rather than feature distance, over $114$ accelerator configurations and four policy families: two accelerators each clear their own gate's bar and log every reuse as certified, while on one task one finishes every episode and the other none. A monitor is two designs, not one --- the statistic it reads, and what it does when that statistic fires. Published gates re-arm after every rejection, and under that response even an oracle handed every forward pass and the exact local action error loses twenty points; absorbing the same statistic costs one point, and most of its speed. What makes a response that never forgets affordable is a statistic that rarely fires, and what it must measure is deviation from the policy the accelerator replaced --- which a chunked policy has already paid for, its last plan not yet expired and free to read. Guarding every call this way cuts per-call compute by $1.55$--$3.09\times$, where any reference-requiring check at the same coverage would have to stop accelerating altogether. On three of our four families the schedule alone already holds the pre-stated $\pm2$-point margin. What the monitor is measurably worth shows in three places: on the fourth family, where it rescues the candidate selection landed on; on four configurations it did not select; and on a contact-rich fifth family, chosen where the schedule was expected to fail and run after every design choice was frozen, where no unmonitored arm at its speed holds the margin and the monitored one does.

cs.RO↗

arXiv2026

You Only Align Once: Propagating Cooperative Behaviors in Multi-Agent Systems through Seed Agents

Ensuring aligned agent behaviors in distributed open multi-agent systems remains challenging, especially as populations grow and unaligned agents may exist. We show that a single aligned agent can propagate cooperative behaviors to unmodified agents purely through natural-language interaction, a phenomenon we term Alignment Propagation. We study this in the Red-Black Game, a team-based iterated Prisoner's Dilemma in which teammates deliberate and vote to determine their team's collective action. By distilling the cooperative reasoning and persuasive dialogues of a teacher model into Qwen3-14B, we obtain a seed agent that, when placed among four unmodified teammates, more than doubles the cooperation rate from 24.8% to 62.2%, outperforming the teacher model and a vanilla Gemini-3.1-Pro. Remarkably, a seed trained exclusively on the Red-Black Game transfers zero-shot to Sugarscape, a spatially grounded survival simulation with pairwise trading, achieving a 91.5% trade success rate versus a 21.6% baseline. Our results reframe multi-agent alignment from an exhaustive per-agent training problem to a scalable social capability that can be engineered through strategic seed placement.

cs.MA↗

arXiv2026

NeuroECG: ECGFounder-Based Deep ECG Representation for EEG-Free Neurological Prognostication After Cardiac Arrest

Neurological prognostication after cardiac arrest commonly relies on electroencephalography (EEG). However, EEG demands high clinical resources. Bedside electrocardiography (ECG) is standard and low-cost. Yet, its value for predicting neurological outcomes remains underexplored. In this study, we propose NeuroECG, an ECGFounder-based deep representation framework for EEG-free auxiliary prognostication. NeuroECG adapts a pretrained ECG foundation model via task-specific fine-tuning. We implement a gradual unfreezing strategy on single-channel bedside monitoring ECG. Multiple ECG segments per patient are encoded into segment-level deep features. These embeddings are aggregated via quantile pooling (q = 0.24) and compressed using principal component analysis (PCA). Experiments on 412 ECG-available patients from the multicenter I-CARE database show that the adapted ECGFounder backbone achieves the best performance among ECG-only backbone baselines, with a test AUROC of 0.7333. We further combine the learned deep ECG representation with static clinical covariates. The proposed NeuroECG model achieves a test AUROC of 0.8077 and an AUPRC of 0.8970. These results support deep bedside ECG representations as a useful source of auxiliary prognostic information. Their integration with static clinical covariates improves prediction in an EEG-free setting. The source code is available at https://github.com/goddream66/NeuroECG

cs.LG↗

arXiv2026

Evaluating the accuracy of KV cache reuse techniques

Position-independent KV cache reuse aims to reduce latency in retrieval-augmented generation by reusing chunk-level KV caches across prompts. We show that current evaluations of KV cache reuse techniques rely on measurements that fail to faithfully capture the loss of accuracy attributable to reuse, often artificially inflating the reported effectiveness. We also show that existing datasets do not exhibit the reuse dynamics needed to thoroughly evaluate such techniques. To address these issues, we propose an evaluation methodology that measures this accuracy loss without ambiguity and we introduce Boxoffice, a tool that programmatically generates evaluation datasets that exercise challenging KV cache reuse patterns.

cs.LG↗

arXiv2026

CMoE: Contrastive Mixture of Experts for Motion Control and Terrain Adaptation of Humanoid Robots

For effective deployment in real-world environments, humanoid robots must autonomously navigate a diverse range of complex terrains with abrupt transitions. While the Vanilla mixture of experts (MoE) framework is theoretically capable of modeling diverse terrain features, in practice, the gating network exhibits nearly uniform expert activations across different terrains, weakening the expert specialization and limiting the model's expressive power. To address this limitation, we introduce CMoE, a novel single-stage reinforcement learning framework that integrates contrastive learning to refine expert activation distributions. By imposing contrastive constraints, CMoE maximizes the consistency of expert activations within the same terrain while minimizing their similarity across different terrains, thereby encouraging experts to specialize in distinct terrain types. We validated our approach on the Unitree G1 humanoid robot through a series of challenging experiments. Results demonstrate that CMoE enables the robot to traverse continuous steps up to 20 cm high and gaps up to 80 cm wide, while achieving robust and natural gait across diverse mixed terrains, surpassing the limits of existing methods. To support further research and foster community development, we release our code publicly.

cs.RO↗

arXiv2026

Counting Near-Spanning Matchings in Latin Squares and Steiner Triple Systems

Montgomery recently proved that for sufficiently large $n$, every Latin square of order $n$ has a partial transversal with $n-1$ cells, and every Steiner triple system of order $n$ has a matching with $\lfloor n/3\rfloor-1$ edges, thus confirming the Ryser--Brualdi--Stein conjecture for even $n$ and the conjecture of Brouwer. We prove sharp enumerative refinements of these results: there is an absolute constant $c>0$ such that, for sufficiently large $n$, 1) every Latin square of order $n$ has $ \left((1\pm n^{-c})\frac{n}{\mathrm {e}^2}\right)^n$ partial transversals with $n-1$ cells; 2) every Steiner triple system of order $n$ has $ \left((1\pm n^{-c})\frac{n}{2\mathrm {e}^2}\right)^{\lfloor n/3\rfloor}$ matchings with $\lfloor n/3\rfloor-1$ edges. The first estimate confirms predictions of Montgomery and Kelly.

math.CO↗

arXiv2026

CAP: Continuously Adaptive Perception-Blind Humanoid Locomotion via Learned Denoising

Humanoid locomotion across complex terrain demands forward-looking exteroception to anticipate obstacles, yet this signal is unreliable in real-world deployment, failing partially and intermittently. Existing perceptive policies often assume that depth observations remain clean and in-distribution, while recent attempts to unify perceptive and blind control typically route or switch between separate sub-policies, leaving recoverable information in partially corrupted depth unexploited. We instead propose CAP, a single-stage humanoid locomotion policy that recovers this signal with a perceptive world-model encoder trained as a learned denoiser to reconstruct clean depth from a corrupted input, together with a co-active proprioceptive variational encoder that supplies depth-free body-state information. A coupled training recipe pairs a depth-noise curriculum on the world-model input with world-model feature dropout on the policy-facing latent, exposing the policy to failures across the entire perception-quality spectrum. In simulation, CAP matches or improves upon perceptive baselines when depth remains informative, and degrades more smoothly than a binary-switching baseline as perception worsens. On the Unitree G1, controlled trials and indoor-outdoor deployments demonstrate perception-robust locomotion under intermittent occlusion, real-sensor corruption, and outdoor depth artifacts.

cs.RO↗

arXiv2026

Nonparametric heterogeneous causal mediation with orthogonal machine learning

Causal mediation analysis decomposes the total effect of an intervention on an outcome into a direct pathway and an indirect pathway transmitted through a mediator, but standard methods typically summarize these pathways using population average effects. In many applications, however, the indirect effect may vary substantially across individual profiles. We propose an orthogonal statistical learning framework for estimating heterogeneous causal mediation effects conditional on individual characteristics. The method constructs a class of weighted Neyman orthogonal losses motivated by influence function representations of weighted population average effects. These losses directly target conditional mediation estimands whose minimizers are locally insensitive to nuisance estimation errors. We implement the resulting learners under a two-stage meta-learning framework with regularized linear sieves as second-stage smoothers, and introduce a combination of targeted learning and orthogonal learning designed to improve stability when mediator density ratios are unstable. We establish $L^2$ and uniform limit theory and develop pointwise and uniform confidence bands. Simulation studies show that the proposed orthogonal learners reduce the mean integrated squared error by more than $50\%$ compared with existing model-based methods and provide computationally efficient inference in nonlinear settings. The CARDIA, PSACR, and STAR analyses reveal heterogeneous mediated effects across cardiometabolic, psychological, and educational settings.

stat.ME↗

arXiv2026

SurveyAgent-HKA: A multi-agent framework for scientific survey generation with LLMs and human knowledge augmentation

Automatic scientific survey generation has become an important task in scientific document processing. The common approach of retrieving literature from a single source (e.g., arXiv) and generating surveys through a one-pass large language model (LLM) call often leads to limited reference coverage and, more importantly, fails to replicate the expert-driven revision process that is crucial for writing high-quality surveys. In this paper, we introduce SurveyAgent-HKA, a multi-agent framework that improves end-to-end scientific survey generation by incorporating knowledge derived from published surveys and peer-review comments. The framework decomposes survey generation into well-defined sub-tasks handled by LLM-powered agent. It first retrieves relevant papers from multiple sources and identifies key topics through clustering to construct an initial outline, which is then refined using outlines from related human-written surveys. Based on the refined outline, topic-focused papers are retrieved and re-ranked to select for drafting a well-grounded survey. Then, we identify common issues raised by experts in peer-review comments from published surveys to guide the revisions and finalize the survey. Experiments on two domains show that our approach outperforms mainstream baselines in citation quality, structural consistency, and content quality. Furthermore, our framework is efficient in both time and cost, making it a practical solution for broader AI-assisted scientific writing applications.

cs.CL↗