arXiv subjects

Fan Wu

Publications and source records attributed to Fan Wu.

At least 343 records · Page 19Linked to original sources

arXiv2021

Game-based Pricing and Task Offloading in Mobile Edge Computing Enabled Edge-Cloud Systems

As a momentous enabling of the Internet of things (IoT), mobile edge computing (MEC) provides IoT mobile devices (MD) with powerful external computing and storage resources. However, a mechanism addressing distributed task offloading and price competition for the open exchange marketplace has not been established properly, which has become a huge obstacle to MEC's application in the IoT market. In this paper, we formulate a distributed mechanism to analyze the interaction between OSPs and IoT MDs in the MEC enabled edge-cloud system by appling multi-leader multi-follower two-tier Stackelberg game theory. We first prove the existence of the Stackelberg equilibrium, and then we propose two distributed algorithms, namely iterative proximal offloading algorithm (IPOA) and iterative Stackelberg game pricing algorithm (ISPA). The IPOA solves the follower non-cooperative game among IoT MDs and ISPA uses backward induction to deal with the price competition among OSPs. Experimental results show that IPOA can markedly reduce the disutility of IoT MDs compared with other traditional task offloading schemes and the price of anarchy is always less than 150\%. Besides, results also demonstrate that ISPA is reliable in boosting the revenue of OSPs.

cs.GT↗

arXiv2021

Optimizing Multiple Performance Metrics with Deep GSP Auctions for E-commerce Advertising

In e-commerce advertising, the ad platform usually relies on auction mechanisms to optimize different performance metrics, such as user experience, advertiser utility, and platform revenue. However, most of the state-of-the-art auction mechanisms only focus on optimizing a single performance metric, e.g., either social welfare or revenue, and are not suitable for e-commerce advertising with various, dynamic, difficult to estimate, and even conflicting performance metrics. In this paper, we propose a new mechanism called Deep GSP auction, which leverages deep learning to design new rank score functions within the celebrated GSP auction framework. These new rank score functions are implemented via deep neural network models under the constraints of monotone allocation and smooth transition. The requirement of monotone allocation ensures Deep GSP auction nice game theoretical properties, while the requirement of smooth transition guarantees the advertiser utilities would not fluctuate too much when the auction mechanism switches among candidate mechanisms to achieve different optimization objectives. We deployed the proposed mechanisms in a leading e-commerce ad platform and conducted comprehensive experimental evaluations with both offline simulations and online A/B tests. The results demonstrated the effectiveness of the Deep GSP auction compared to the state-of-the-art auction mechanisms.

cs.GT↗

arXiv2021

Generating Giant Vortex in a Fermi Superfluid via Spin-Orbital-Angular-Momentum Coupling

Spin-orbital-angular-momentum (SOAM) coupling has been realized in recent experiments of Bose-Einstein condensates [Chen et al., Phys. Rev. Lett. 121, 113204 (2018) and Zhang et al., Phys. Rev. Lett. 122, 110402 (2019)], where the orbital angular momentum imprinted upon bosons leads to quantized vortices. For fermions, such an exotic synthetic gauge field can provide fertile ground for fascinating pairing schemes and rich superfluid phases, which are yet to be explored. Here we demonstrate how SOAM coupling stabilizes vortices in Fermi superfluids through a unique mechanism that can be viewed as the angular analog to that of the spin-orbit-coupling-induced Fulde-Ferrell state under a Fermi surface deformation. Remarkably, the vortex size is comparable with the beam waist of Raman lasers generating the SOAM coupling, which is typically much larger than previously observed vortices in Fermi superfluids. With tunable size and core structure, these giant vortex states provide unprecedented experimental access to topological defects in Fermi superfluids.

cond-mat.quant-gas↗

arXiv2020

Distributed Non-Convex Optimization with Sublinear Speedup under Intermittent Client Availability

Federated learning is a new distributed machine learning framework, where a bunch of heterogeneous clients collaboratively train a model without sharing training data. In this work, we consider a practical and ubiquitous issue when deploying federated learning in mobile environments: intermittent client availability, where the set of eligible clients may change during the training process. Such intermittent client availability would seriously deteriorate the performance of the classical Federated Averaging algorithm (FedAvg for short). Thus, we propose a simple distributed non-convex optimization algorithm, called Federated Latest Averaging (FedLaAvg for short), which leverages the latest gradients of all clients, even when the clients are not available, to jointly update the global model in each iteration. Our theoretical analysis shows that FedLaAvg attains the convergence rate of $O(E^{1/2}/(N^{1/4} T^{1/2}))$, achieving a sublinear speedup with respect to the total number of clients. We implement FedLaAvg along with several baselines and evaluate them over the benchmarking MNIST and Sentiment140 datasets. The evaluation results demonstrate that FedLaAvg achieves more stable training than FedAvg in both convex and non-convex settings and indeed reaches a sublinear speedup.

stat.ML↗

arXiv2020

A Continuous-Time Mirror Descent Approach to Sparse Phase Retrieval

We analyze continuous-time mirror descent applied to sparse phase retrieval, which is the problem of recovering sparse signals from a set of magnitude-only measurements. We apply mirror descent to the unconstrained empirical risk minimization problem (batch setting), using the square loss and square measurements. We provide a convergence analysis of the algorithm in this non-convex setting and prove that, with the hypentropy mirror map, mirror descent recovers any $k$-sparse vector $\mathbf{x}^\star\in\mathbb{R}^n$ with minimum (in modulus) non-zero entry on the order of $\| \mathbf{x}^\star \|_2/\sqrt{k}$ from $k^2$ Gaussian measurements, modulo logarithmic terms. This yields a simple algorithm which, unlike most existing approaches to sparse phase retrieval, adapts to the sparsity level, without including thresholding steps or adding regularization terms. Our results also provide a principled theoretical understanding for Hadamard Wirtinger flow [58], as Euclidean gradient descent applied to the empirical risk problem with Hadamard parametrization can be recovered as a first-order approximation to mirror descent in discrete time.

stat.ML↗

arXiv2020

Exploiting Variable Impedance for Energy Efficient Sequential Movements

Compliant robotics have seen successful applications in energy efficient locomotion and cyclic manipulation. However, exploitation of variable physical impedance for energy efficient sequential movements has not been extensively addressed. This work employs a hierarchical approach to encapsulate low-level optimal control for sub-movement generation into an outer loop of iterative policy improvement, thereby leveraging the benefits of both optimal control and reinforcement learning. The framework enables optimizing efficiency trade-off for minimal energy expenses in a model-free manner, by taking account of cost function weighting, variable impedance exploitation, and transition timing -- which are associated with the skill of compliance. The effectiveness of the proposed method is evaluated using two consecutive reaching tasks on a variable impedance actuator. The results demonstrate significant energy saving by improving the skill of compliance, with an electrical consumption reduction of about 30% measured in a physical robot experiment.

cs.RO↗

arXiv2020

Better Model Selection with a new Definition of Feature Importance

Feature importance aims at measuring how crucial each input feature is for model prediction. It is widely used in feature engineering, model selection and explainable artificial intelligence (XAI). In this paper, we propose a new tree-model explanation approach for model selection. Our novel concept leverages the Coefficient of Variation of a feature weight (measured in terms of the contribution of the feature to the prediction) to capture the dispersion of importance over samples. Extensive experimental results show that our novel feature explanation performs better than general cross validation method in model selection both in terms of time efficiency and accuracy performance.

stat.ML↗

arXiv2020

IEO: Intelligent Evolutionary Optimisation for Hyperparameter Tuning

Hyperparameter optimisation is a crucial process in searching the optimal machine learning model. The efficiency of finding the optimal hyperparameter settings has been a big concern in recent researches since the optimisation process could be time-consuming, especially when the objective functions are highly expensive to evaluate. In this paper, we introduce an intelligent evolutionary optimisation algorithm which applies machine learning technique to the traditional evolutionary algorithm to accelerate the overall optimisation process of tuning machine learning models in classification problems. We demonstrate our Intelligent Evolutionary Optimisation (IEO)in a series of controlled experiments, comparing with traditional evolutionary optimisation in hyperparameter tuning. The empirical study shows that our approach accelerates the optimisation speed by 30.40% on average and up to 77.06% in the best scenarios.

cs.LG↗

arXiv2020

A Deep Prediction Network for Understanding Advertiser Intent and Satisfaction

For e-commerce platforms such as Taobao and Amazon, advertisers play an important role in the entire digital ecosystem: their behaviors explicitly influence users' browsing and shopping experience; more importantly, advertiser's expenditure on advertising constitutes a primary source of platform revenue. Therefore, providing better services for advertisers is essential for the long-term prosperity for e-commerce platforms. To achieve this goal, the ad platform needs to have an in-depth understanding of advertisers in terms of both their marketing intents and satisfaction over the advertising performance, based on which further optimization could be carried out to service the advertisers in the correct direction. In this paper, we propose a novel Deep Satisfaction Prediction Network (DSPN), which models advertiser intent and satisfaction simultaneously. It employs a two-stage network structure where advertiser intent vector and satisfaction are jointly learned by considering the features of advertiser's action information and advertising performance indicators. Experiments on an Alibaba advertisement dataset and online evaluations show that our proposed DSPN outperforms state-of-the-art baselines and has stable performance in terms of AUC in the online environment. Further analyses show that DSPN not only predicts advertisers' satisfaction accurately but also learns an explainable advertiser intent, revealing the opportunities to optimize the advertising performance further.

cs.SI↗

arXiv2020

Retrieve Synonymous keywords for Frequent Queries in Sponsored Search in a Data Augmentation Way

In sponsored search, retrieving synonymous keywords is of great importance for accurately targeted advertising. The semantic gap between queries and keywords and the extremely high precision requirements (>= 95\%) are two major challenges to this task. To the best of our knowledge, the problem has not been openly discussed. In an industrial sponsored search system, the retrieved keywords for frequent queries are usually done ahead of time and stored in a lookup table. Considering these results as a seed dataset, we propose a data-augmentation-like framework to improve the synonymous retrieval performance for these frequent queries. This framework comprises two steps: translation-based retrieval and discriminant-based filtering. Firstly, we devise a Trie-based translation model to make a data increment. In this phase, a Bag-of-Core-Words trick is conducted, which increased the data increment's volume 4.2 times while keeping the original precision. Then we use a BERT-based discriminant model to filter out nonsynonymous pairs, which exceeds the traditional feature-driven GBDT model with 11\% absolute AUC improvement. This method has been successfully applied to Baidu's sponsored search system, which has yielded a significant improvement in revenue. In addition, a commercial Chinese dataset containing 500K synonymous pairs with a precision of 95\% is released to the public for paraphrase study (http://ai.baidu.com/broad/subordinate?dataset=paraphrasing).

cs.IR↗

arXiv2020

Ground-state phase diagram and excitation spectrum of a Bose-Einstein condensate with spin-orbital-angular-momentum coupling

We investigate the ground-state phase diagram and excitation spectrum of an interacting spinor Bose-Einstein condensate with spin-orbital-angular-momentum (SOAM) coupling realized in recent experiments by introducing atomic Raman transition with a pair of copropagating Laguerre-Gaussian laser beams that carry different orbital angular momenta (OAM) [Chen et al., Phys. Rev. Lett. 121, 113204 (2018) and Zhang et al., Phys. Rev. Lett. 122, 110402 (2019)]. Because of the ground-state degeneracy of the single-particle Hamiltonian at vanishing detuning, several angular-stripe phases, which are superposition of states with different angular quantum numbers, appear in the phase diagram. However, these phases normally exist at small detuning, which makes them hard to be probed in experiments. We show that for a large OAM difference of the laser beams, an asymmetric kind of angular-stripe phase can exist even at large detuning. The excitation spectra in different phases exhibit distinct features: In the angular-stripe phase there exist two gapless bands corresponding to the broken U(1) and rotational symmetries, while in the half-skyrmion phase the gapless band exhibits a roton-like structure. Our predictions of the angular-stripe phases and the low-energy excitations can be examined in recently realized BECs with SOAM coupling.

cond-mat.quant-gas↗

arXiv2020

DeGNN: Characterizing and Improving Graph Neural Networks with Graph Decomposition

Despite the wide application of Graph Convolutional Network (GCN), one major limitation is that it does not benefit from the increasing depth and suffers from the oversmoothing problem. In this work, we first characterize this phenomenon from the information-theoretic perspective and show that under certain conditions, the mutual information between the output after $l$ layers and the input of GCN converges to 0 exponentially with respect to $l$. We also show that, on the other hand, graph decomposition can potentially weaken the condition of such convergence rate, which enabled our analysis for GraphCNN. While different graph structures can only benefit from the corresponding decomposition, in practice, we propose an automatic connectivity-aware graph decomposition algorithm, DeGNN, to improve the performance of general graph neural networks. Extensive experiments on widely adopted benchmark datasets demonstrate that DeGNN can not only significantly boost the performance of corresponding GNNs, but also achieves the state-of-the-art performances.

cs.LG↗

arXiv2020

A regularity criterion for three-dimensional micropolar fluid equations in Besov spaces of negative regular indices

In this article, we study regularity criteria for the 3D micropolar fluid equations in terms of one partial derivative of the velocity. It is proved that if \begin{equation*} \int^{T}_{0}\|\partial_{3}u\|^{\frac{2}{1-r}}_{\dot{B}^{-r}_{\infty,\infty}} dt<\infty \quad \text{with} \quad 0< r<1, \end{equation*} then, the solutions of the micropolar fluid equations actually are smooth on $(0, T)$. This improves and extends many previous results.

math.AP↗

arXiv2020

Effective theory for ultracold strongly interacting fermionic atoms in two dimensions

We propose a minimal theoretical model for the description of a two-dimensional (2D) strongly interacting Fermi gas confined transversely in a tight harmonic potential, and present accurate predictions for its equation of state and breathing mode frequency. We show that the minimal model Hamiltonian needs at least two independent interaction parameters, the 2D scattering length and effective range of interactions, in order to quantitatively explain recent experimental measurements at nonzero filling factor $N/N_{2D}$, where $N$ is the total number of atoms and $N_{2D}$ is the threshold number to reach the 2D limit. We therefore resolve in a satisfactory way the puzzling experimental observations of reduced equations of state and reduced quantum anomaly in breathing mode frequency, due to small yet non-negligible $N/N_{2D}$. We argue that a conclusive demonstration of the much-anticipated quantum anomaly is possible at a filling factor of a few percent. Our establishment of the minimal model for 2D ultracold atoms could be crucial to understanding the fermionic Berezinskii-Kosterlitz-Thouless transition in the strongly correlated regime.

cond-mat.quant-gas↗

arXiv2020

UAVs as a Service: Boosting Edge Intelligence for Air-Ground Integrated Networks

The air-ground integrated network is a key component of future sixth generation (6G) networks to support seamless and near-instant super-connectivity. There is a pressing need to intelligently provision various services in 6G networks, which however is challenging. To meet this need, in this article, we propose a novel architecture called UaaS (UAVs as a Service) for the air-ground integrated network, featuring UAV as a key enabler to boost edge intelligence with the help of machine learning (ML) techniques. We envision that the proposed UaaS architecture could intelligently provision wireless communication service, edge computing service, and edge caching service by a network of UAVs, making full use of UAVs' flexible deployment and diverse ML techniques. We also conduct a case study where UAVs participate in the model training of distributed ML among multiple terrestrial users, whose result shows that the model training is efficient with a negligible energy consumption of UAVs, compared to the flight energy consumption. Finally, we discuss the challenges and open research issues in the UaaS.

cs.NI↗

arXiv2020

A Multi-Agent-Based Rolling Optimization Method for Restoration Scheduling of Electrical Distribution Systems with Distributed Generation

Resilience against major disasters is the most essential characteristic of future electrical distribution systems (EDS). A multi-agent-based rolling optimization method for EDS restoration scheduling is proposed in this paper. When a blackout occurs, considering the risk of losing the centralized authority due to the failure of the common core communication network, the agents available after disasters or cyber-attacks identify the communication-connected parts (CCPs) in the EDS with distributed communication. A multi-time interval optimization model is formulated and solved by the agents for the restoration scheduling of a CCP. A rolling optimization process for the entire EDS restoration is proposed. During the scheduling/rescheduling in the rolling process, the CCPs in the EDS are reidentified and the restoration schedules for the CCPs are updated. Through decentralized decision-making and rolling optimization, EDS restoration scheduling can automatically start and periodically update itself, providing effective solutions for EDS restoration scheduling in a blackout event. A modified IEEE 123-bus EDS is utilized to demonstrate the effectiveness of the proposed method.

eess.SP↗

arXiv2020

Distributed Optimization over Block-Cyclic Data

We consider practical data characteristics underlying federated learning, where unbalanced and non-i.i.d. data from clients have a block-cyclic structure: each cycle contains several blocks, and each client's training data follow block-specific and non-i.i.d. distributions. Such a data structure would introduce client and block biases during the collaborative training: the single global model would be biased towards the client or block specific data. To overcome the biases, we propose two new distributed optimization algorithms called multi-model parallel SGD (MM-PSGD) and multi-chain parallel SGD (MC-PSGD) with a convergence rate of $O(1/\sqrt{NT})$, achieving a linear speedup with respect to the total number of clients. In particular, MM-PSGD adopts the block-mixed training strategy, while MC-PSGD further adds the block-separate training strategy. Both algorithms create a specific predictor for each block by averaging and comparing the historical global models generated in this block from different cycles. We extensively evaluate our algorithms over the CIFAR-10 dataset. Evaluation results demonstrate that our algorithms significantly outperform the conventional federated averaging algorithm in terms of test accuracy, and also preserve robustness for the variance of critical parameters.

cs.LG↗

arXiv2020

Energy regenerative damping in variable impedance actuators for long-term robotic deployment

Energy efficiency is a crucial issue towards longterm deployment of compliant robots in the real world. In the context of variable impedance actuators (VIAs), one of the main focuses has been on improving energy efficiency through reduction of energy consumption. However, the harvesting of dissipated energy in such systems remains under-explored. This study proposes a novel variable damping module design enabling energy regeneration in VIAs by exploiting the regenerative braking effect of DC motors. The proposed damping module uses four switches to combine regenerative and dynamic braking, in a hybrid approach that enables energy regeneration without a reduction in the range of damping achievable. A physical implementation on a simple VIA mechanism is presented in which the regenerative properties of the proposed module are characterised and compared against theoretical predictions. To investigate the role of variable regenerative damping in terms of energy efficiency of longterm operation, experiments are reported in which the VIA equipped with the proposed damping module performs sequential reaching to a series of stochastic targets. The results indicate that the combination of variable stiffness and variable regenerative damping is preferable to achieve the optimal trade-off between task performance and energy efficiency. Use of the latter results in a 25% performance improvement on overall performance metrics (incorporating reaching accuracy, settling time, energy consumption and regeneration), over comparable schemes where either stiffness or damping are fixed.

cs.RO↗