arXiv subjects

Tan Su

Publications and source records attributed to Tan Su.

8 recordsLinked to original sources

arXiv2026

MemCorr-DP: Counterfactual Correspondence Conditioning for a Diffusion Policy Guided by a Reference

Behavior-cloned visuomotor policies can remain accurate near their training distribution yet fail when object position and camera viewpoint change together. A successful reference trajectory contains the geometry needed to transfer the same interaction, but the policy must align that geometry with the current scene and remain sensitive to it during denoising. To address these challenges, we present MemCorr-DP, a diffusion policy that lifts frozen RoMa v2 matches into explicit 3D relations between the current scene and the reference trajectory. A counterfactual paired objective assigns opposite behaviors the same physical state and noisy action while retaining reference-specific denoising targets. Mixed-condition fine-tuning then adapts the policy from ground-truth geometry to measured correspondence errors. Our strongest evaluation places the Door in the outermost position bands beyond the training support and changes the query camera by $\pm15^\circ$. Under this combined shift, MemCorr-DP achieves 96.67% closed-loop success, compared with 88.00% for a visual Transformer with the same action architecture. Objective ablations and reference interventions show that behavior responds to the selected reference, while matched controls favor the complete relation set over future motion or centroid geometry alone. These results support explicit 3D reference relations as a robust conditioning interface when spatial and viewpoint changes are compounded in the evaluated task.

cs.RO↗

arXiv2026

A General B\'ezier Tree Encoding Counterfactual Framework for Retinal-Vessel-Mediated Disease Analysis

The geometry of the retinal vessel is a key biomarker of vascular diseases, yet clinical evidence remains primarily observational. Existing generative counterfactuals intervene only at the image-level disease label, failing to isolate explicit anatomical structure. To address this limitation, we propose the B\'ezier Tree Encoding Counterfactual Framework (BTECF). By abstracting vascular networks into interconnected cubic-B\'ezier segments, BTECF establishes a disease-agnostic representation in which structural topology is explicitly preserved and atomically perturbable. Coupling this encoding with a diffusion-based generator enables parameter-level do-interventions on explicit geometric axes (e.g., tortuosity, caliber) while preserving background fundus textures. We validate BTECF on diabetic retinopathy, together with independent cohorts for ischemic stroke and Alzheimer's disease. Isolated counterfactual interventions produce dose-responsive shifts in classifier predictions; a matched pixel-drop control attenuates this response by an order of magnitude or more, ruling out out-of-distribution generation artifacts. By enforcing causal isolation between vessel topology and pixel-level confounders, BTECF provides a unified generative paradigm for hypothesis verification across systemic diseases. To support reproducibility, the code will be publicly released upon acceptance.

eess.IV↗

arXiv2025

Index invariants and Eta invariants determine Differential KO theory in degrees that are multiples of 8

Sullivan--Simons developed a Cheeger--Simons differential character analogue for degree (0 mod 2) differential K-theory, giving a complete set of numerical invariants that determine a complex vector bundle with unitary connection on a base manifold X, up to Chern--Simons equivalence of the connection. In this paper we develop a degree (0 mod 8) differential KO-analogue. Namely, given a real vector bundle with orthogonal connection, we construct R/Z -valued eta-invariants in the context of Atiyah--Patodi--Singer and Z2 Atiyah--Singer index invariants that completely determine differential KO-theory in degree (0 mod 8); we call this the differential KO-character. In the second part, for a Riemannian submersion X to B with closed 8k-dimensional spin fibers, we develop two family index theorems in differential KO -theory: one in the differential KO -character model and one in the structured-bundle model. The fact that these two pushforwards agree follows from the Bismut--Cheeger adiabatic limit theorem, providing a new interpretation of that result.

math.KT↗

arXiv2025

A Systematic Investigation on Deep Learning-Based Omnidirectional Image and Video Super-Resolution

Omnidirectional image and video super-resolution is a crucial research topic in low-level vision, playing an essential role in virtual reality and augmented reality applications. Its goal is to reconstruct high-resolution images or video frames from low-resolution inputs, thereby enhancing detail preservation and enabling more accurate scene analysis and interpretation. In recent years, numerous innovative and effective approaches have been proposed, predominantly based on deep learning techniques, involving diverse network architectures, loss functions, projection strategies, and training datasets. This paper presents a systematic review of recent progress in omnidirectional image and video super-resolution, focusing on deep learning-based methods. Given that existing datasets predominantly rely on synthetic degradation and fall short in capturing real-world distortions, we introduce a new dataset, 360Insta, that comprises authentically degraded omnidirectional images and videos collected under diverse conditions, including varying lighting, motion, and exposure settings. This dataset addresses a critical gap in current omnidirectional benchmarks and enables more robust evaluation of the generalization capabilities of omnidirectional super-resolution methods. We conduct comprehensive qualitative and quantitative evaluations of existing methods on both public datasets and our proposed dataset. Furthermore, we provide a systematic overview of the current status of research and discuss promising directions for future exploration. All datasets, methods, and evaluation metrics introduced in this work are publicly available and will be regularly updated. Project page: https://github.com/nqian1/Survey-on-ODISR-and-ODVSR.

cs.CV↗

arXiv2025

Hierarchical Flow Diffusion for Efficient Frame Interpolation

Most recent diffusion-based methods still show a large gap compared to non-diffusion methods for video frame interpolation, in both accuracy and efficiency. Most of them formulate the problem as a denoising procedure in latent space directly, which is less effective caused by the large latent space. We propose to model bilateral optical flow explicitly by hierarchical diffusion models, which has much smaller search space in the denoising procedure. Based on the flow diffusion model, we then use a flow-guided images synthesizer to produce the final result. We train the flow diffusion model and the image synthesizer end to end. Our method achieves state of the art in accuracy, and 10+ times faster than other diffusion-based methods. The project page is at: https://hfd-interpolation.github.io.

cs.CV↗

arXiv2012

On asymptotic description of passage through a resonance in quasi-linear Hamiltonian systems

We consider a quasi-linear Hamiltonian system with one and a half degrees of freedom. The Hamiltonian of this system differs by a small, $\sim\varepsilon$, perturbing term from the Hamiltonian of a linear oscillatory system. We consider passage through a resonance: the frequency of the latter system slowly changes with time and passes through 0. The speed of this passage is of order of $\varepsilon$. We provide asymptotic formulas that describe effects of passage through a resonance with an accuracy $O(\varepsilon^{\frac32})$. This is an improvement of known results by Chirikov (1959), Kevorkian (1971, 1974) and Bosley (1996). The problem under consideration is a model problem that describes passage through an isolated resonance in multi-frequency quasi-linear Hamiltonian systems.

math.DS↗

arXiv2012

On phenomenon of scattering on resonances associated with discretisation of systems with fast rotating phase

Numerical integration of ODEs by standard numerical methods reduces a continuous time problems to discrete time problems. Discrete time problems have intrinsic properties that are absent in continuous time problems. As a result, numerical solution of an ODE may demonstrate dynamical phenomena that are absent in the original ODE. We show that numerical integration of system with one fast rotating phase lead to a situation of such kind: numerical solution demonstrate phenomenon of scattering on resonances that is absent in the original system.

math.DS↗

arXiv2011

On the accuracy of conservation of adiabatic invariants in slow-fast systems

Let the adiabatic invariant of action variable in slow-fast Hamiltonian system with two degrees of freedom have two limiting values along the trajectories as time tends to infinity. The difference of two limits is exponentially small in analytic systems. An iso-energetic reduction and canonical transformations are applied to transform the slow-fast systems to form of systems depending on slowly varying parameters in a complexified phase space. On the basis of this method an estimate for the accuracy of conservation of adiabatic invariant is given for such systems.

math.DS↗