arXiv subjects

Ziqi Ma

Publications and source records attributed to Ziqi Ma.

At least 19 recordsLinked to original sources

arXiv2026

Linear adaptive variable time step schemes for the anisotropic phase field dendritic crystal growth model

In this paper, we propose a class of efficient, linear and decoupled adaptive variable time step schemes for the anisotropic phase field dendritic crystal growth model. The main difficulty is twofold: the gradient dependent anisotropic coefficient and the strongly nonlinear terms make unconditional energy stability difficult to achieve, and on nonuniform meshes the discrete energy dissipation law, which is straightforward to establish on uniform meshes, has to be reestablished under varying step sizes. To overcome these difficulties, two linear stabilizers and a time dependent auxiliary variable with relaxation are introduced and incorporated into a second-order generalized BDF-type operator, with the spatial derivatives discretized by the Fourier spectral method. The proposed schemes are energy stable for arbitrary positive adjacent step ratios, so that no restriction on the step ratio is imposed, and only two linear equations together with a few algebraic equations, which are solved independently in Fourier space, need to be solved at each time step. Moreover, the schemes are naturally compatible with energy based time adaptive strategies. The associated adaptive strategy reduces the number of time steps by $24.6\%$--$86.6\%$ in comparison with uniform time stepping, and numerical experiments on fourfold and sixfold dendritic growth in two dimensions and on three-dimensional growth confirm the accuracy, efficiency and stability of the proposed schemes.

math.NA↗

arXiv2026

C2GA: A Class-Controllable Generative Augmentation Framework for Respiratory Sound Classification

Background: Respiratory sound classification plays a critical role in the clinical identification of pulmonary pathologies. However, its performance is often hindered by the limited size, severe noise, and class imbalance of real-world auscultation datasets. Although conventional audio augmentation techniques are easy to implement, they may inadvertently distort subtle pathological characteristics. Meanwhile, existing Variational Autoencoder (VAE)- or Generative Adversarial Network (GAN)-based generative approaches often suffer from limited sample fidelity and insufficient controllability over class semantics, particularly under conditions of scarce supervision. Methods: To overcome these limitations, we propose C2GA, a class-controllable generative augmentation framework. C2GA first constructs a semantically rich discrete latent space using a conditional Vector-Quantized Variational Autoencoder (VQ-VAE), in which local acoustic tokens are explicitly decoupled from global class prototypes. Subsequently, a Transformer-based autoregressive prior is trained to generate label-consistent token sequences. These generated tokens are then fused with the corresponding class prototypes and decoded into high-fidelity Mel-spectrograms for data augmentation. Conclusion: These results indicate that C2GA provides an effective and semantically reliable augmentation strategy for respiratory sound analysis. By enabling controllable and high-quality data generation, the proposed framework offers a promising solution for improving the robustness and generalization of respiratory sound classification in realistic clinical scenarios.

cs.SD↗

arXiv2026

Balancing Privacy and Compliance in DeFi: A Zero-Knowledge-Based Auditable Cross-Chain Framework

The rapid emergence of decentralized finance (DeFi) has introduced complex challenges for cross-chain transactions, which involve transferring assets across disparate blockchain networks. A fundamental dilemma arises between user privacy and ensuring regulatory compliance. Unlike single-chain systems, cross-chain environments must address privacy and auditability across heterogeneous architectures. Existing approaches,ranging from transparent ledgers to anonymous cryptocurrencies, fail to reconcile these two requirements, hindering regulatory adoption. This research presents an auditable cross-chain framework that integrates three core components. First, zero-knowledge proofs (ZKPs) enable compliance verification (e.g., amount non-negativity, signature validity) without revealing transaction details. Second, a light-client mechanism enables trust-minimized cross-chain verification without dependence on third-party relayers. Third, a threshold view-key mechanism based on distributed key generation (DKG) restricts that audit access to authorized entities under legal triggers such as the FATF Travel Rule and MiCA Regulation. For cross-border investigations, the framework adheres to national laws and the EU Directive on Mutual Legal Assistance. This work systematically integrates ZKPs, threshold cryptography, and light-client verification into an auditable, privacy-preserving cross-chain protocol, and evaluates a prototype on the Ethereum Sepolia testnet. The results demonstrate the practical feasibility of privacy-preserving compliance verification for cross-chain DeFi. At the same time, broader interoperability and Regulatory Technology (RegTech) impact require further validation on additional chains and with real regulatory workflows.

cs.CR↗

arXiv2026

LoGo: Local-Global Rewards for Consistent Long-Horizon Video Generation

Camera-controlled video models are rapidly advancing toward long generation horizons and complex camera control. A key failure mode is 3D inconsistency: as the camera moves, objects lose permanence and scene structures shift. Existing post-training techniques, which assign a single scalar reward to the entire generation, are poorly suited to correcting these inconsistencies over long horizons. We introduce LoGo, which blends global and spatially localized rewards for camera-controlled video models. The local reward provides fine-grained credit assignment, which substantially improves 3D consistency, while the global reward preserves camera following and video quality. Across three base models, LoGo shows a clear advantage on DL3DV and TrajectoryBench, a new benchmark for long-horizon, complex-camera-control generation that current evaluations lack. LoGo effectively reduces local object shifts, artifacts, and global scene changes, illustrating the importance of credit assignment in post-training video models. Project website: https://ziqi-ma.github.io/logo-website/

cs.CV↗

arXiv2026

DynaTokens: Teaching Dynamics to Camera-Controlled Video Models at Test Time

Video generation must account for two sources of motion, one induced by the observer's camera path and the other caused by scene dynamics. An ideal camera-controlled video model should account for both motions: let users move the camera while evolving the scene dynamics. While current models handle camera-induced motion well in static settings, they struggle for dynamic scenes: objects are static, move incorrectly, or degrade in generation quality. We introduce DynaTokens, a lightweight set of learnable scene-specific tokens that teach dynamics to an existing camera-controlled world model. Our method is motivated by a simple asymmetry between the two sources of motion: whereas camera motion affects the generated view globally, object dynamics are spatially localized. Through cross-attention, DynaTokens trains the learnable tokens from a few example trajectories for a scene while keeping the base model frozen, and enables dynamics under new query camera paths. DynaTokens achieves a better simultaneous dynamics-camera tradeoff on VBench2 and WorldScore evaluations than LoRA, block finetuning, and specialized trainable-layer baselines. Analyses of token attention, ablations, and motion temporality suggest that matching the trainable interface to the structure of the learning target is important for effective adaptation. Project website: https://glab-caltech.github.io/dynatokens/

cs.CV↗

arXiv2026

Adaptive World Memory 3D Foundation Model for Scalable 3D Mapping, Localization, and Rendering

Recent 3D foundation models enable generalizable geometric reasoning from RGB images but remain limited in persistent memory, scalability, and renderable scene modeling. We present a memory-centric 3D foundation model for scalable robotic localization, reconstruction, and Gaussian rendering. Its core is an adaptive world memory mechanism that combines transformer-based gated updates with test-time temporal-spatial regulation. Learned gates control recurrent memory propagation, while temporal state evolution and spatial observation-state consistency regulate token-wise updates and forgetting over long image sequences. To support large-scale mapping, we organize memory into local submaps and integrate progressive mapping and tracking, loop closure, and SL(4)-based global refinement to maintain local accuracy and global consistency. A Gaussian reconstruction head decodes memory-enhanced features into renderable primitives, unifying camera pose estimation, dense point-cloud reconstruction, and photorealistic rendering within a single model. Experiments on public benchmarks and self-collected datasets from diverse robotic platforms demonstrate improved trajectory accuracy, reconstruction completeness, and rendering quality over existing 3D foundation reconstruction and SLAM baselines. These results support adaptive memory as a foundation for persistent robotic world modeling. The dataset and code will be made publicly available at \href{https://github.com/dtc111111/AWM-3DFM}{https://github.com/dtc111111/AWM-3DFM}.

cs.CV↗

arXiv2026

Multi-Submap Implicit Neural SLAM with Local-to-Global Loop Closure for Large-Scale Scene Reconstruction

Neural Radiance Fields (NeRF)-based SLAM has demonstrated impressive results in small-scale scene reconstruction, yet scaling these methods to extensive, complex environments remains challenging due to catastrophic forgetting and accumulated trajectory drift. This paper presents a robust, large-scale neural SLAM system featuring a multi-submap architecture and a dual-tier loop closure mechanism. Specifically, we propose a progressive mapping strategy that dynamically allocates neural submaps to maintain high-fidelity representations without memory explosion. For robust pose estimation, an optical-flow-based tracking module is integrated to handle aggressive motions. To address global consistency, we introduce a local-to-global loop closure framework leveraging the foundation model for high-performance global descriptor extraction, significantly enhancing relocalization accuracy under varying viewpoints. Furthermore, an inter-submap online distillation algorithm is designed during back-end optimization to enforce geometric and appearance consistency across overlapping submap boundaries. To validate the system, we developed a customized handheld mechatronic platform and conducted extensive evaluations on both public benchmarks and our large-scale indoor-outdoor datasets. Experimental results, including direct deployment on an onboard computing unit, demonstrate that our approach outperforms state-of-the-art neural SLAM methods in reconstruction quality and localization robustness, providing a scalable solution for real-world robotic perception and digital twinning. We will release the code publicly on \href{https://github.com/dtc111111/MSN-SLAM}{https://github.com/dtc111111/MSN-SLAM} .

cs.CV↗

arXiv2026

SAM 3D: 3Dfy Anything in Images

We present SAM 3D, a generative model for visually grounded 3D object reconstruction, predicting geometry, texture, and layout from a single image. SAM 3D excels in natural images, where occlusion and scene clutter are common and visual recognition cues from context play a larger role. We achieve this with a human- and model-in-the-loop pipeline for annotating object shape, texture, and pose, providing visually grounded 3D reconstruction data at unprecedented scale. We learn from this data in a modern, multi-stage training framework that combines synthetic pretraining with real-world alignment, breaking the 3D "data barrier". We obtain significant gains over recent work, with at least a 5:1 win rate in human preference tests on real-world objects and scenes. We will release our code and model weights, an online demo, and a new challenging benchmark for in-the-wild 3D object reconstruction.

cs.CV↗

arXiv2026

CSST Preparations: Galaxy Completeness and Sérsic Profile Fitting across the Wide, Deep, and Extreme Fields

The upcoming imaging survey of the Chinese Space-station Survey Telescope (CSST) will deliver high-resolution imaging of an unprecedented number of galaxies for galaxy studies. To understand CSST's capability, and to support the preparation of early-science programs, we generate 470,526 mock CSST images for 22,406 simulated galaxies with $M_*>10^9 M_\odot$, whose parameters are calibrated to match real HST observations spanning photometric redshift $0<z\lesssim7$, across seven CSST filters and three planned survey depths: wide, deep, and extreme. We then perform source detection and Sérsic fitting. For point sources, we found that the 95% completeness magnitude in the g band reaches 26.3, 27.4, and 28.5 mag for the wide, deep, and extreme fields, respectively. For extended galaxies, their spatial extent dilutes the surface brightness, leading to brighter 95% completeness magnitudes of 24.4, 25.9, and 27.1 mag. The detection completeness remains above 95% at $z\lesssim3-4$ in the extreme field, while the corresponding redshift limits are $z\approx1$ in the deep field and $z\approx0.5$ in the wide field. Using three fitting codes, GALFIT, AstroPhot, and SourceXtractor++, we quantify measurement biases and uncertainties in galaxy magnitude ($m$), effective radius ($R_e$), effective surface brightness ($μ_e$), Sérsic index ($n$), and axis ratio ($q$). On average, for fainter galaxies, the reduced signal-to-noise ratio leads to systematic overestimates in $m$, $R_e$, and $μ_e$, and underestimates in $n$ and $q$. These biases, as well as the associated scatter, become progressively smaller in deeper fields. Overall, our results provide quantitative constraints on sample selection and the robustness of morphological measurements in CSST early-science and legacy surveys.

astro-ph.GA↗

arXiv2026

Out of Sight, Out of Mind? Evaluating State Evolution in Video World Models

Evolutions in the world, such as water pouring or ice melting, happen regardless of being observed. Video world models generate "worlds" via 2D frame observations. Can these generated "worlds" evolve regardless of observation? To probe this question, we design a benchmark to evaluate whether video world models can decouple state evolution from observation. Our benchmark, STEVO-Bench, applies observation control to evolving processes via instructions of occluder insertion, turning off the light, or specifying camera "lookaway" trajectories. By evaluating video models with and without camera control for a diverse set of naturally-occurring evolutions, we expose their limitations in decoupling state evolution from observation. STEVO-Bench proposes an evaluation protocol to automatically detect and disentangle failure modes of video world models across key aspects of natural state evolution. Analysis of STEVO-Bench results provide new insight into potential data and architecture bias of present-day video world models. Project website: https://glab-caltech.github.io/STEVOBench/. Blog: https://ziqi-ma.github.io/blog/2026/outofsight/

cs.CV↗

arXiv2025

Feedforward 3D Editing via Text-Steerable Image-to-3D

Recent progress in image-to-3D has opened up immense possibilities for design, AR/VR, and robotics. However, to use AI-generated 3D assets in real applications, a critical requirement is the capability to edit them easily. We present a feedforward method, Steer3D, to add text steerability to image-to-3D models, which enables editing of generated 3D assets with language. Our approach is inspired by ControlNet, which we adapt to image-to-3D generation to enable text steering directly in a forward pass. We build a scalable data engine for automatic data generation, and develop a two-stage training recipe based on flow-matching training and Direct Preference Optimization (DPO). Compared to competing methods, Steer3D more faithfully follows the language instruction and maintains better consistency with the original 3D asset, while being 2.4x to 28.5x faster. Steer3D demonstrates that it is possible to add a new modality (text) to steer the generation of pretrained image-to-3D generative models with 100k data. Project website: https://glab-caltech.github.io/steer3d/

cs.CV↗

arXiv2025

Can LLMs Translate Human Instructions into a Reinforcement Learning Agent's Internal Emergent Symbolic Representation?

Emergent symbolic representations are critical for enabling developmental learning agents to plan and generalize across tasks. In this work, we investigate whether large language models (LLMs) can translate human natural language instructions into the internal symbolic representations that emerge during hierarchical reinforcement learning. We apply a structured evaluation framework to measure the translation performance of commonly seen LLMs -- GPT, Claude, Deepseek and Grok -- across different internal symbolic partitions generated by a hierarchical reinforcement learning algorithm in the Ant Maze and Ant Fall environments. Our findings reveal that although LLMs demonstrate some ability to translate natural language into a symbolic representation of the environment dynamics, their performance is highly sensitive to partition granularity and task complexity. The results expose limitations in current LLMs capacity for representation alignment, highlighting the need for further research on robust alignment between language and internal agent representations.

cs.CL↗

arXiv2025

Manipulate as Human: Learning Task-oriented Manipulation Skills by Adversarial Motion Priors

In recent years, there has been growing interest in developing robots and autonomous systems that can interact with human in a more natural and intuitive way. One of the key challenges in achieving this goal is to enable these systems to manipulate objects and tools in a manner that is similar to that of humans. In this paper, we propose a novel approach for learning human-style manipulation skills by using adversarial motion priors, which we name HMAMP. The approach leverages adversarial networks to model the complex dynamics of tool and object manipulation, as well as the aim of the manipulation task. The discriminator is trained using a combination of real-world data and simulation data executed by the agent, which is designed to train a policy that generates realistic motion trajectories that match the statistical properties of human motion. We evaluated HMAMP on one challenging manipulation task: hammering, and the results indicate that HMAMP is capable of learning human-style manipulation skills that outperform current baseline methods. Additionally, we demonstrate that HMAMP has potential for real-world applications by performing real robot arm hammering tasks. In general, HMAMP represents a significant step towards developing robots and autonomous systems that can interact with humans in a more natural and intuitive way, by learning to manipulate tools and objects in a manner similar to how humans do.

cs.RO↗

arXiv2025

Reconstruction of Antarctic sea ice thickness from sparse satellite laser altimetry data using a partial convolutional neural network

The persistent lack of spatially complete Antarctic sea ice thickness (SIT) data at sub-monthly resolution has fundamentally constrained the quantitative understanding of large-scale sea ice mass balance processes. In this study, a pan-Antarctic SIT dataset at 5-day and 12.5 km resolution was developed based on sparse Ice, Cloud and Land Elevation Satellite (ICESat: 2003-2009) and ICESat-2 (2018-2024) along-track laser altimetry SIT retrievals using a deep learning approach. The reconstructed SIT was quantitatively validated against independent upward-looking sonar (ULS) observations and showed higher accuracy than the other four satellite-derived and reanalyzed Antarctic SIT datasets. The temporal evolution of the reconstructed SIT was further validated by ULS and ICESat-2 observations. Consistent seasonal cycles and intra-seasonal tendencies across these datasets confirm the reconstruction's reliability. Beyond advancing the mechanistic understanding of Antarctic sea ice variability and climate linkages, this reconstruction dataset's near-real-time updating capability offers operational value for monitoring and forecasting the Antarctic sea ice state.

physics.geo-ph↗

arXiv2025

Find Any Part in 3D

Why don't we have foundation models in 3D yet? A key limitation is data scarcity. For 3D object part segmentation, existing datasets are small in size and lack diversity. We show that it is possible to break this data barrier by building a data engine powered by 2D foundation models. Our data engine automatically annotates any number of object parts: 1755x more unique part types than existing datasets combined. By training on our annotated data with a simple contrastive objective, we obtain an open-world model that generalizes to any part in any object based on any text query. Even when evaluated zero-shot, we outperform existing methods on the datasets they train on. We achieve 260% improvement in mIoU and boost speed by 6x to 300x. Our scaling analysis confirms that this generalization stems from the data scale, which underscores the impact of our data engine. Finally, to advance general-category open-world 3D part segmentation, we release a benchmark covering a wide range of objects and parts. Project website: https://ziqi-ma.github.io/find3dsite/

cs.CV↗

arXiv2024

Calibrated Uncertainty Quantification for Operator Learning via Conformal Prediction

Operator learning has been increasingly adopted in scientific and engineering applications, many of which require calibrated uncertainty quantification. Since the output of operator learning is a continuous function, quantifying uncertainty simultaneously at all points in the domain is challenging. Current methods consider calibration at a single point or over one scalar function or make strong assumptions such as Gaussianity. We propose a risk-controlling quantile neural operator, a distribution-free, finite-sample functional calibration conformal prediction method. We provide a theoretical calibration guarantee on the coverage rate, defined as the expected percentage of points on the function domain whose true value lies within the predicted uncertainty ball. Empirical results on a 2D Darcy flow and a 3D car surface pressure prediction task validate our theoretical results, demonstrating calibrated coverage and efficient uncertainty bands outperforming baseline methods. In particular, on the 3D problem, our method is the only one that meets the target calibration percentage (percentage of test samples for which the uncertainty estimates are calibrated) of 98%.

cs.LG↗

arXiv2024

Unsupervised Motion Retargeting for Human-Robot Imitation

This early-stage research work aims to improve online human-robot imitation by translating sequences of joint positions from the domain of human motions to a domain of motions achievable by a given robot, thus constrained by its embodiment. Leveraging the generalization capabilities of deep learning methods, we address this problem by proposing an encoder-decoder neural network model performing domain-to-domain translation. In order to train such a model, one could use pairs of associated robot and human motions. Though, such paired data is extremely rare in practice, and tedious to collect. Therefore, we turn towards deep learning methods for unpaired domain-to-domain translation, that we adapt in order to perform human-robot imitation.

cs.RO↗

arXiv2023

Long-term and Real-time High-speed Underwater Wireless Optical Communications in Deep Sea

Seafloor observation network can perform all-weather, long-term, continuous, real-time, and in-situ observation of the ocean by combing various observation methods including cabled seafloor nodes, self-contained nodes, as well as mobile platforms, where reliable and long-term high-speed underwater wireless communication becomes an essential demand. Recently, underwater wireless optical communication (UWOC) has emerged as a highly promising solution and is rapidly becoming a research hotspot for meeting this requirement. In this article, we demonstrate the experiment and application of high-speed UWOC system for deep sea seafloor observation network. To the best of our knowledge this is the first long-term real-time deep-sea UWOC link with bitrate as high as 125 Mbps. Between 30 m distance and at a depth of 1650 m, two-way Ethernet UWOC links are realized with 125 Mbps direction-adjustable green light link and 6.25 Mbps non-line-of-sight (NLOS) blue light link. High quality video transmission of 8K 30 FPS and 4K 120 FPS are realized through high-speed 125 Mbps green light link, with 100% peak signal-to-noise ratio (PSNR) agreement, showing the capability of transmitting high-quality videos lossless. The 30-day long-term measurement results show that the BER performance of both 125 Mbps and 6.25 Mbps links is lower than 10-5, proving the stability and reliability of this UWOC system at depth of 1650 m. The maximum transmission distance for the green and blue light links are estimated to be 117.7 and 128.3 m with considering the geometry loss, which can be extended to 231.6 and 337.5 m without geometry loss. As the first long-term and real-time UWOC system in deep sea, we believe this demonstration can provide valuable experience for further UWOC studies and converged ocean observation networking with cabled and cable-less observation platforms.

cs.NI↗