arXiv subjects

Yuan Hu

Publications and source records attributed to Yuan Hu.

At least 37 records · Page 2Linked to original sources

arXiv2022

PolyBuilding: Polygon Transformer for End-to-End Building Extraction

We present PolyBuilding, a fully end-to-end polygon Transformer for building extraction. PolyBuilding direct predicts vector representation of buildings from remote sensing images. It builds upon an encoder-decoder transformer architecture and simultaneously outputs building bounding boxes and polygons. Given a set of polygon queries, the model learns the relations among them and encodes context information from the image to predict the final set of building polygons with fixed vertex numbers. Corner classification is performed to distinguish the building corners from the sampled points, which can be used to remove redundant vertices along the building walls during inference. A 1-d non-maximum suppression (NMS) is further applied to reduce vertex redundancy near the building corners. With the refinement operations, polygons with regular shapes and low complexity can be effectively obtained. Comprehensive experiments are conducted on the CrowdAI dataset. Quantitative and qualitative results show that our approach outperforms prior polygonal building extraction methods by a large margin. It also achieves a new state-of-the-art in terms of pixel-level coverage, instance-level precision and recall, and geometry-level properties (including contour regularity and polygon complexity).

cs.CV↗

arXiv2022

ESG-valued discrete option pricing in complete markets

We consider option pricing using replicating binomial trees, with a two fold purpose. The first is to introduce ESG valuation into option pricing. We explore this in a number of scenarios, including enhancement of yield due to trader information and the impact of the past history of a market driver. The second is to emphasize the use of discrete dynamic pricing, rather than continuum models, as the natural model that governs actual market practice. We further emphasize that discrete option pricing models must use discrete compounding (such as risk-free rate compounding of $1+r_f Δt$) rather than continuous compounding (such as $e^{r_f Δt})$.

q-fin.PR↗

arXiv2022

SwinVRNN: A Data-Driven Ensemble Forecasting Model via Learned Distribution Perturbation

Data-driven approaches for medium-range weather forecasting are recently shown extraordinarily promising for ensemble forecasting for their fast inference speed compared to traditional numerical weather prediction (NWP) models, but their forecast accuracy can hardly match the state-of-the-art operational ECMWF Integrated Forecasting System (IFS) model. Previous data-driven attempts achieve ensemble forecast using some simple perturbation methods, like initial condition perturbation and Monte Carlo dropout. However, they mostly suffer unsatisfactory ensemble performance, which is arguably attributed to the sub-optimal ways of applying perturbation. We propose a Swin Transformer-based Variational Recurrent Neural Network (SwinVRNN), which is a stochastic weather forecasting model combining a SwinRNN predictor with a perturbation module. SwinRNN is designed as a Swin Transformer-based recurrent neural network, which predicts future states deterministically. Furthermore, to model the stochasticity in prediction, we design a perturbation module following the Variational Auto-Encoder paradigm to learn multivariate Gaussian distributions of a time-variant stochastic latent variable from data. Ensemble forecasting can be easily achieved by perturbing the model features leveraging noise sampled from the learned distribution. We also compare four categories of perturbation methods for ensemble forecasting, i.e. fixed distribution perturbation, learned distribution perturbation, MC dropout, and multi model ensemble. Comparisons on WeatherBench dataset show the learned distribution perturbation method using our SwinVRNN model achieves superior forecast accuracy and reasonable ensemble spread due to joint optimization of the two targets. More notably, SwinVRNN surpasses operational IFS on surface variables of 2-m temperature and 6-hourly total precipitation at all lead times up to five days.

cs.CV↗

arXiv2022

A unified stochastic particle method based on the Bhatnagar-Gross-Krook model for polyatomic gases and its combination with DSMC

Simulating hypersonic flow around a space vehicle is challenging because of the multiscale and nonequilibrium nature inherent in these flows. To effectively deal with such flows, a novel particle particle hybrid scheme combining the stochastic particle Bhatnagar-Gross-Krook (BGK) method with Direct Simulation Monte Carlo (DSMC) was developed recently, but only for monatomic gases [Fei et. al., J. Comput. Phys. 2021]. Here this work is extended to the particle-particle hybrid method for polyatomic gases. In the near continuum regime, employing the Ellipsoidal Statistical BGK model proposed by Y. Dauvois, et. al. [Eur. J. Mech. B Fluids, 2021] with discrete levels of vibrational energy, the stochastic particle BGK method is first established following the idea of the unified stochastic particle BGK (USPBGK) scheme. In the fluid limit, it has been proven to be of second order temporal and spatial accuracy. Then, the USPBGK scheme with rotational and vibrational energies is combined with DSMC to construct a hybrid method for polyatomic gases. The present hybrid scheme is validated with numerical tests of homogenous relaxation, 1D shock structure and 2D hypersonic flows past a wedge and a cylinder. Compared to traditional stochastic particle methods, the proposed hybrid method can achieve higher accuracy at a much lower computational cost. Therefore, it is a more efficient tool to study multiscale hypersonic flows.

physics.flu-dyn↗

arXiv2021

Market Complete Option Valuation using a Jarrow-Rudd Pricing Tree with Skewness and Kurtosis

Applying the Cherny-Shiryaev-Yor invariance principle, we introduce a generalized Jarrow-Rudd (GJR) option pricing model with uncertainty driven by a skew random walk. The GJR pricing tree exhibits skewness and kurtosis in both the natural and risk-neutral world. We construct implied surfaces for the parameters determining the GJR tree. Motivated by Merton's pricing tree incorporating transaction costs, we extend the GJR pricing model to include a hedging cost. We demonstrate ways to fit the GJR pricing model to a market driver that influences the price dynamics of the underlying asset. We supplement our findings with numerical examples.

q-fin.MF↗

arXiv2021

TextFlint: Unified Multilingual Robustness Evaluation Toolkit for Natural Language Processing

Various robustness evaluation methodologies from different perspectives have been proposed for different natural language processing (NLP) tasks. These methods have often focused on either universal or task-specific generalization capabilities. In this work, we propose a multilingual robustness evaluation platform for NLP tasks (TextFlint) that incorporates universal text transformation, task-specific transformation, adversarial attack, subpopulation, and their combinations to provide comprehensive robustness analysis. TextFlint enables practitioners to automatically evaluate their models from all aspects or to customize their evaluations as desired with just a few lines of code. To guarantee user acceptability, all the text transformations are linguistically based, and we provide a human evaluation for each one. TextFlint generates complete analytical reports as well as targeted augmented data to address the shortcomings of the model's robustness. To validate TextFlint's utility, we performed large-scale empirical evaluations (over 67,000 evaluations) on state-of-the-art deep learning models, classic supervised methods, and real-world systems. Almost all models showed significant performance degradation, including a decline of more than 50% of BERT's prediction accuracy on tasks such as aspect-level sentiment classification, named entity recognition, and natural language inference. Therefore, we call for the robustness to be included in the model evaluation, so as to promote the healthy development of NLP technology.

cs.CL↗

arXiv2021

Portfolio Optimization Constrained by Performance Attribution

This paper investigates performance attribution measures as a basis for constraining portfolio optimization. We employ optimizations that minimize expected tail loss and investigate both asset allocation (AA) and the selection effect (SE) as hard constraints on asset weights. The test portfolio consists of stocks from the Dow Jones Industrial Average index; the benchmark is an equi-weighted portfolio of the same stocks. Performance of the optimized portfolios is judged using comparisons of cumulative price and the risk-measures maximum drawdown, Sharpe ratio, and Rachev ratio. The results suggest a positive role in price and risk-measure performance for the imposition of constraints on AA and SE, with SE constraints producing the larger performance enhancement.

q-fin.RM↗

arXiv2020

Option Pricing Incorporating Factor Dynamics in Complete Markets

Using the Donsker-Prokhorov invariance principle we extend the Kim-Stoyanov-Rachev-Fabozzi option pricing model to allow for variably-spaced trading instances, an important consideration for short-sellers of options. Applying the Cherny-Shiryaev-Yor invariance principles, we formulate a new binomial path-dependent pricing model for discrete- and continuous-time complete markets where the stock price dynamics depends on the log-return dynamics of a market influencing factor. In the discrete case, we extend the results of this new approach to a financial market with informed traders employing a statistical arbitrage strategy involving trading of forward contracts. Our findings are illustrated with numerical examples employing US financial market data. Our work provides further support for the conclusion that any option pricing model must preserve valuable information on the instantaneous mean log-return, the probability of the stock's upturn movement (per trading interval), and other market microstructure features.

q-fin.MF↗

arXiv2020

Option Pricing in Markets with Informed Traders

The objective of this paper is to introduce the theory of option pricing for markets with informed traders within the framework of dynamic asset pricing theory. We introduce new models for option pricing for informed traders in complete markets where we consider traders with information on the stock price direction and stock return mean. The Black-Scholes-Merton option pricing theory is extended for markets with informed traders, where price processes are following continuous-diffusions. By doing so, the discontinuity puzzle in option pricing is resolved. Using market option data, we estimate the implied surface of the probability for a stock upturn, the implied mean stock return surface, and implied trader information intensity surface.

q-fin.MF↗

arXiv2020

Option Pricing with Mixed Levy Subordinated Price Process and Implied Probability Weighting Function

It is essential to incorporate the impact of investor behavior when modeling the dynamics of asset returns. In this paper, we reconcile behavioral finance and rational finance by incorporating investor behavior within the framework of dynamic asset pricing theory. To include the views of investors, we employ the method of subordination which has been proposed in the literature by including business (intrinsic, market) time. We define a mixed Levy subordinated model by adding a single subordinated Levy process to the well-known log-normal model, resulting in a new log-price process. We apply the proposed models to study the behavioral finance notion of "greed and fear" disposition from the perspective of rational dynamic asset pricing theory. The greedy or fearful disposition of option traders is studied using the shape of the probability weighting function. We then derive the implied probability weighting function for the fear and greed deposition of option traders in comparison to spot traders. Our result shows the diminishing sensitivity of option traders. Diminishing sensitivity results in option traders overweighting the probability of big losses in comparison to spot traders.

q-fin.MF↗

arXiv2020

GBCNs: Genetic Binary Convolutional Networks for Enhancing the Performance of 1-bit DCNNs

Training 1-bit deep convolutional neural networks (DCNNs) is one of the most challenging problems in computer vision, because it is much easier to get trapped into local minima than conventional DCNNs. The reason lies in that the binarized kernels and activations of 1-bit DCNNs cause a significant accuracy loss and training inefficiency. To address this problem, we propose Genetic Binary Convolutional Networks (GBCNs) to optimize 1-bit DCNNs, by introducing a new balanced Genetic Algorithm (BGA) to improve the representational ability in an end-to-end framework. The BGA method is proposed to modify the binary process of GBCNs to alleviate the local minima problem, which can significantly improve the performance of 1-bit DCNNs. We develop a new BGA module that is generic and flexible, and can be easily incorporated into existing DCNNs, such asWideResNets and ResNets. Extensive experiments on the object classification tasks (CIFAR, ImageNet) validate the effectiveness of the proposed method. To highlight, our method shows strong generalization on the object recognition task, i.e., face recognition, facial and person re-identification.

cs.CV↗

arXiv2019

RBCN: Rectified Binary Convolutional Networks for Enhancing the Performance of 1-bit DCNNs

Binarized convolutional neural networks (BCNNs) are widely used to improve memory and computation efficiency of deep convolutional neural networks (DCNNs) for mobile and AI chips based applications. However, current BCNNs are not able to fully explore their corresponding full-precision models, causing a significant performance gap between them. In this paper, we propose rectified binary convolutional networks (RBCNs), towards optimized BCNNs, by combining full-precision kernels and feature maps to rectify the binarization process in a unified framework. In particular, we use a GAN to train the 1-bit binary network with the guidance of its corresponding full-precision model, which significantly improves the performance of BCNNs. The rectified convolutional layers are generic and flexible, and can be easily incorporated into existing DCNNs such as WideResNets and ResNets. Extensive experiments demonstrate the superior performance of the proposed RBCNs over state-of-the-art BCNNs. In particular, our method shows strong generalization on the object tracking task.

cs.CV↗

arXiv2019

Modelling Crypto Asset Price Dynamics, Optimal Crypto Portfolio, and Crypto Option Valuation

Despite being described as a medium of exchange, cryptocurrencies do not have the typical attributes of a medium of exchange. Consequently, cryptocurrencies are more appropriately described as crypto assets. A common investment attribute shared by the more than 2,500 crypto assets is that they are highly volatile. An investor interested in reducing price volatility of a portfolio of crypto assets can do so by constructing an optimal portfolio through standard optimization techniques that minimize tail risk. Because crypto assets are not backed by any real assets, forming a hedge to reduce the risk contribution of a single crypto asset can only be done with another set of similar assets (i.e., a set of other crypto assets). A major finding of this paper is that crypto portfolios constructed via optimizations that minimize variance and Conditional Value at Risk outperform a major stock market index (the S$\&$P 500). As of this writing, options in which the underlying is a crypto asset index are not traded, one of the reasons being that the academic literature has not formulated an acceptable fair pricing model. We offer a fair valuation model for crypto asset options based on a dynamic pricing model for the underlying crypto assets. The model was carefully backtested and therefore offers a reliable model for the underlying crypto assets in the natural world. We then obtain the valuation of crypto options by passing the natural world to the equivalent martingale measure via the Esscher transform. Because of the absence of traded crypto options we could not compare the prices obtained from our valuation model to market prices. Yet, we can claim that if such options on crypto assets are introduced, they should follow closely our theoretical prices after adjusting for market frictions and design feature nuances.

q-fin.RM↗

arXiv2019

Panoptic Edge Detection

Pursuing more complete and coherent scene understanding towards realistic vision applications drives edge detection from category-agnostic to category-aware semantic level. However, finer delineation of instance-level boundaries still remains unexcavated. In this work, we address a new finer-grained task, termed panoptic edge detection (PED), which aims at predicting semantic-level boundaries for stuff categories and instance-level boundaries for instance categories, in order to provide more comprehensive and unified scene understanding from the perspective of edges.We then propose a versatile framework, Panoptic Edge Network (PEN), which aggregates different tasks of object detection, semantic and instance edge detection into a single holistic network with multiple branches. Based on the same feature representation, the semantic edge branch produces semantic-level boundaries for all categories and the object detection branch generates instance proposals. Conditioned on the prior information from these two branches, the instance edge branch aims at instantiating edge predictions for instance categories. Besides, we also devise a Panoptic Dual F-measure (F2) metric for the new PED task to uniformly measure edge prediction quality for both stuff and instances. By joint end-to-end training, the proposed PEN framework outperforms all competitive baselines on Cityscapes and ADE20K datasets.

cs.CV↗

arXiv2019

Dynamic Feature Fusion for Semantic Edge Detection

Features from multiple scales can greatly benefit the semantic edge detection task if they are well fused. However, the prevalent semantic edge detection methods apply a fixed weight fusion strategy where images with different semantics are forced to share the same weights, resulting in universal fusion weights for all images and locations regardless of their different semantics or local context. In this work, we propose a novel dynamic feature fusion strategy that assigns different fusion weights for different input images and locations adaptively. This is achieved by a proposed weight learner to infer proper fusion weights over multi-level features for each location of the feature map, conditioned on the specific input. In this way, the heterogeneity in contributions made by different locations of feature maps and input images can be better considered and thus help produce more accurate and sharper edge predictions. We show that our model with the novel dynamic feature fusion is superior to fixed weight fusion and also the naïve location-invariant weight fusion methods, via comprehensive experiments on benchmarks Cityscapes and SBD. In particular, our method outperforms all existing well established methods and achieves new state-of-the-art.

cs.CV↗

arXiv2016

Influence of intrinsic decoherence on tripartite entanglement and bipartite fidelity of polar molecules in pendular states

An array of ultracold polar molecules trapped in an external electric field is regarded as a promising carrier of quantum information. Under the action of this field, molecules are compelled to undergo pendular oscillations by the Stark effect. Particular attention has been paid to the influence of intrinsic decoherence on the model of linear polar molecular pendular states, thereby we evaluate the tripartite entanglement with negativity, as well as fidelity of bipartite quantum systems for input and output signals using electric dipole moments of polar molecules as qubits. According to this study, we consider three typical initial states for both systems respectively, and investigate the temporal evolution with variable values of the external field intensity, the intrinsic decoherence factor and the dipole-dipole interaction. Thus we demonstrate the sound selection of these three main parameters to obtain the best entanglement degree and fidelity.

quant-ph↗

arXiv2014

Emergency Route Selection for D2D Cellular Communications During an Urban Terrorist Attack

Device-to-Device (D2D) communications is a technology that allows mobile users to relay information to each other, without access to the cellular network. In this paper, we consider how to dynamically select multi-hop routes for D2D communications in spectrum co-existence with a fully loaded cellular network. The modelling scenario is that of a real urban environment, when the cellular network is congested during an unexpected event, such as a terrorist attack. We use D2D to relay data across the urban terrain, in the presence of conventional cellular (CC) communications. We consider different wireless routing algorithms, namely: shortest-path-routing (SPR), interference-aware-routing (IAR), and broadcast-routing (BR). In general, there is a fundamental trade-off between D2D and CC outage performances, due to their mutual interference relationship. For different CC outage constraints and D2D end-to-end distances, the paper recommends different D2D routing strategies. The paper also considers the effects of varying user density and urban building material properties on overall D2D relaying feasibility. Over a distance of a kilometre, it was found that the success probability of D2D communications can reach 91% for a moderate participating user density (400 per square km) and a low wall penetration loss (<10dB).

cs.NI↗

arXiv2013

Optimizing Device-to-Device Communications in Cellular Networks

In this report, we consider how to dynamically select transmission bands and multi-hop routes for device-to-device (D2D) communications in co-existence with a cellular network overlay. Firstly, we consider different wireless routing algorithms, i.e. broadcasting-routing (BR) method, and shortest-path-routing (SPR) method. The results show that depending on the co-existence cellular users' outage constraint, different routing strategies have different merits. BR is acceptable at the low D2D user density but is terrible at high density. We also consider the channel band performance (Uplink band and Downlink band). The results show that the multi-hop D2D can achieve a low outage probability using the uplink band (approximately 5%), and D2D in the downlink band performs a little poorly (approximately 12%) outage with SPR.

cs.NI↗