arXiv subjects

Yining Ding

Publications and source records attributed to Yining Ding.

8 recordsLinked to original sources

arXiv2026

Choice of Collateral Currency in Differential Swaps

The role of collateral in derivative pricing has evolved beyond credit risk mitigation, particularly following the global financial crisis, when funding costs and basis spreads became central to valuation practices. This development coincided with the transition from the London Interbank Offered Rate (LIBOR) to risk-free rates (RFRs) and the increasing standardization of collateralised trading. We study the valuation and hedging of a class of differential swaps referencing backward-looking averages of overnight rates, with SOFR swaps appearing as a particular instance. The focus is on the impact of the collateral currency. Extending earlier results Ding et al. [Math. Finance 36 (2026), pp.~180--202], we allow the collateral account to be denominated in a currency different from that of the contractual cash flows and derive explicit pricing and hedging strategies using a futures-based replication approach. We show that the choice of collateral currency can have a non-trivial effect on both valuation and risk management. In particular, foreign-currency collateral can introduce additional risk exposures even when contractual cash flows are entirely denominated in the domestic currency. Numerical study demonstrates that collateral effects can lead to significant valuation adjustments and therefore need to be properly incorporated in modern multi-currency modelling frameworks.

q-fin.MF↗

arXiv2025

Estimating Fog Parameters from a Sequence of Stereo Images

We propose a method which, given a sequence of stereo foggy images, estimates the parameters of a fog model and updates them dynamically. In contrast with previous approaches, which estimate the parameters sequentially and thus are prone to error propagation, our algorithm estimates all the parameters simultaneously by solving a novel optimisation problem. By assuming that fog is only locally homogeneous, our method effectively handles real-world fog, which is often globally inhomogeneous. The proposed algorithm can be easily used as an add-on module in existing visual Simultaneous Localisation and Mapping (SLAM) or odometry systems in the presence of fog. In order to assess our method, we also created a new dataset, the Stereo Driving In Real Fog (SDIRF), consisting of high-quality, consecutive stereo frames of real, foggy road scenes under a variety of visibility conditions, totalling over 40 minutes and 34k frames. As a first-of-its-kind, SDIRF contains the camera's photometric parameters calibrated in a lab environment, which is a prerequisite for correctly applying the atmospheric scattering model to foggy images. The dataset also includes the counterpart clear data of the same routes recorded in overcast weather, which is useful for companion work in image defogging and depth reconstruction. We conducted extensive experiments using both synthetic foggy data and real foggy sequences from SDIRF to demonstrate the superiority of the proposed algorithm over prior methods. Our method not only produces the most accurate estimates on synthetic data, but also adapts better to real fog. We make our code and SDIRF publicly available\footnote{https://github.com/SenseRoboticsLab/estimating-fog-parameters} to the community with the aim of advancing the research on visual perception in fog.

cs.CV↗

arXiv2025

Quadratic Mean-Field BSDEs and Exponential Utility Maximization

In this paper, we study a class of real-valued mean-field backward stochastic differential equations (BSDEs) with generators of quadratic growth in the control variable and the mean-field term. Under this assumption, together with a bounded terminal condition, we establish the existence and uniqueness of solutions. Our approach departs from classical fixed-point arguments and instead combines Malliavin calculus with refined BMO and stability estimates. The result bridges the gap between the quadratic BSDE results of [Ann. Probab. 45 (2017), pp.~3795--3828] and Hao et al. [Ann. Appl. Probab. 35 (2025), pp.~2128--2174]. Moreover, motivated by the structure of the mean-field exponential utility maximization problem introduced in our paper, we extend our framework to terminal conditions without continuity or the Markovian assumption. We establish the existence and uniqueness of solutions under a smallness terminla value on the terminal conditions. We then apply this extended theory to solve a mean-field exponential utility maximization problem, which developing the classical framework of Hu et al. [Ann. Appl. Probab. 15 (2005), pp.~1691--1712] to a fully coupled quadratic mean-field setting.

math.OC↗

arXiv2025

DPMT: Dual Process Multi-scale Theory of Mind Framework for Real-time Human-AI Collaboration

Real-time human-artificial intelligence (AI) collaboration is crucial yet challenging, especially when AI agents must adapt to diverse and unseen human behaviors in dynamic scenarios. Existing large language model (LLM) agents often fail to accurately model the complex human mental characteristics such as domain intentions, especially in the absence of direct communication. To address this limitation, we propose a novel dual process multi-scale theory of mind (DPMT) framework, drawing inspiration from cognitive science dual process theory. Our DPMT framework incorporates a multi-scale theory of mind (ToM) module to facilitate robust human partner modeling through mental characteristic reasoning. Experimental results demonstrate that DPMT significantly enhances human-AI collaboration, and ablation studies further validate the contributions of our multi-scale ToM in the slow system.

cs.LG↗

arXiv2025

CURL-SLAM: Continuous and Compact LiDAR Mapping

This paper studies 3D LiDAR mapping with a focus on developing an updatable and localizable map representation that enables continuity, compactness and consistency in 3D maps. Traditional LiDAR Simultaneous Localization and Mapping (SLAM) systems often rely on 3D point cloud maps, which typically require extensive storage to preserve structural details in large-scale environments. In this paper, we propose a novel paradigm for LiDAR SLAM by leveraging the Continuous and Ultra-compact Representation of LiDAR (CURL) introduced in [1]. Our proposed LiDAR mapping approach, CURL-SLAM, produces compact 3D maps capable of continuous reconstruction at variable densities using CURL's spherical harmonics implicit encoding, and achieves global map consistency after loop closure. Unlike popular Iterative Closest Point (ICP)-based LiDAR odometry techniques, CURL-SLAM formulates LiDAR pose estimation as a unique optimization problem tailored for CURL and extends it to local Bundle Adjustment (BA), enabling simultaneous pose refinement and map correction. Experimental results demonstrate that CURL-SLAM achieves state-of-the-art 3D mapping quality and competitive LiDAR trajectory accuracy, delivering sensor-rate real-time performance (10 Hz) on a CPU. We will release the CURL-SLAM implementation to the community.

cs.RO↗

arXiv2024

Cross-Currency Basis Swaps Referencing Backward-Looking Rates

The financial industry has undergone a significant transition from the London Interbank Offered Rates (LIBORs) to Risk Free Rates (RFRs) such as, e.g., the Secured Overnight Financing Rate (SOFR) in the U.S. and the Cash Rate (AONIA) in Australia, as primary benchmark rates for borrowing costs. The paper examines the pricing and hedging method for financial products in a cross-currency framework with the special emphasis on the Compound SOFR vs Average AONIA cross-currency basis swap (CCBS) where both reference rates are backward-looking and the swap is collateralized. While the SOFR and AONIA are used as particular instances of RFRs in a cross-currency basis swap, the proposed approach is able to handle backward-looking rates for any two currencies. We give explicit pricing and hedging results for a constant notional cross-currency basis swap with either domestic or foreign collateralization using interest rate futures and currency futures as hedging instruments within an arbitrage-free cross-currency multi-curve setting.

q-fin.MF↗

arXiv2021

Pricing and hedging of SOFR derivatives

The LIBOR has served since the 1970s as a fundamental measure for floating term rates across multiple currencies and maturities. However, in 2017 the Financial Conduct Authority announced the discontinuation of LIBOR from the end of 2021 and the New York Fed declared the Treasury repo financing rate, called the Secured Overnight Financing Rate (SOFR), as a candidate for a new reference rate for interest rate swaps denominated in U.S. dollars. We examine arbitrage-free pricing and hedging of swaps referencing SOFR without and with collateral backing. As hedging instruments, we take SOFR futures and idiosyncratic funding rates for the hedge and margin account. For simplicity, a one-factor model based on Vasicek's equation is used to specify the joint dynamics of several overnight interest rates, including the SOFR and unsecured funding rate.

q-fin.MF↗

arXiv2021

CPT: Efficient Deep Neural Network Training via Cyclic Precision

Low-precision deep neural network (DNN) training has gained tremendous attention as reducing precision is one of the most effective knobs for boosting DNNs' training time/energy efficiency. In this paper, we attempt to explore low-precision training from a new perspective as inspired by recent findings in understanding DNN training: we conjecture that DNNs' precision might have a similar effect as the learning rate during DNN training, and advocate dynamic precision along the training trajectory for further boosting the time/energy efficiency of DNN training. Specifically, we propose Cyclic Precision Training (CPT) to cyclically vary the precision between two boundary values which can be identified using a simple precision range test within the first few training epochs. Extensive simulations and ablation studies on five datasets and eleven models demonstrate that CPT's effectiveness is consistent across various models/tasks (including classification and language modeling). Furthermore, through experiments and visualization we show that CPT helps to (1) converge to a wider minima with a lower generalization error and (2) reduce training variance which we believe opens up a new design knob for simultaneously improving the optimization and efficiency of DNN training. Our codes are available at: https://github.com/RICE-EIC/CPT.

cs.LG↗