arXiv subjects

Xiuqiang He

Publications and source records attributed to Xiuqiang He.

At least 55 records · Page 3Linked to original sources

arXiv2024

Training Agents with Weakly Supervised Feedback from Large Language Models

Large Language Models (LLMs) offer a promising basis for creating agents that can tackle complex tasks through iterative environmental interaction. Existing methods either require these agents to mimic expert-provided trajectories or rely on definitive environmental feedback for reinforcement learning which limits their application to specific scenarios like gaming or code generation. This paper introduces a novel training method for LLM-based agents using weakly supervised signals from a critic LLM, bypassing the need for expert trajectories or definitive feedback. Our agents are trained in iterative manner, where they initially generate trajectories through environmental interaction. Subsequently, a critic LLM selects a subset of good trajectories, which are then used to update the agents, enabling them to generate improved trajectories in the next iteration. Extensive tests on the API-bank dataset show consistent improvement in our agents' capabilities and comparable performance to GPT-4, despite using open-source models with much fewer parameters.

cs.CL↗

arXiv2024

Fusion Matters: Learning Fusion in Deep Click-through Rate Prediction Models

The evolution of previous Click-Through Rate (CTR) models has mainly been driven by proposing complex components, whether shallow or deep, that are adept at modeling feature interactions. However, there has been less focus on improving fusion design. Instead, two naive solutions, stacked and parallel fusion, are commonly used. Both solutions rely on pre-determined fusion connections and fixed fusion operations. It has been repetitively observed that changes in fusion design may result in different performances, highlighting the critical role that fusion plays in CTR models. While there have been attempts to refine these basic fusion strategies, these efforts have often been constrained to specific settings or dependent on specific components. Neural architecture search has also been introduced to partially deal with fusion design, but it comes with limitations. The complexity of the search space can lead to inefficient and ineffective results. To bridge this gap, we introduce OptFusion, a method that automates the learning of fusion, encompassing both the connection learning and the operation selection. We have proposed a one-shot learning algorithm tackling these tasks concurrently. Our experiments are conducted over three large-scale datasets. Extensive experiments prove both the effectiveness and efficiency of OptFusion in improving CTR model performance. Our code implementation is available here\url{https://github.com/kexin-kxzhang/OptFusion}.

cs.IR↗

arXiv2024

Cross-Forming Control and Fault Current Limiting for Grid-Forming Inverters

This article proposes a "cross-forming" control concept for grid-forming inverters operating against grid faults. Cross-forming refers to voltage angle forming and current magnitude forming. It differs from classical grid-forming and grid-following paradigms that feature voltage magnitude-and-angle forming and voltage magnitude-and-angle following (or current magnitude-and-angle forming), respectively. The cross-forming concept addresses the need for inverters to remain grid-forming (particularly voltage angle forming, as required by grid codes) while managing fault current limitation. Simple and feasible cross-forming control implementations are proposed, enabling inverters to quickly limit fault currents to a prescribed level while preserving voltage angle forming for grid-forming synchronization and providing dynamic ancillary services, during symmetrical or asymmetrical fault ride-through. Moreover, the cross-forming control yields an equivalent system featuring a constant virtual impedance and a "normal form" representation, allowing for the extension of previously established transient stability results to include scenarios involving current saturation. Simulations and experiments validate the efficacy of the proposed cross-forming control implementations.

eess.SY↗

arXiv2024

Grid-Forming Control of Modular Dynamic Virtual Power Plants

This article explores a flexible and coordinated control design for an aggregation of heterogeneous distributed energy resources (DERs) in a dynamic virtual power plant (DVPP). The control design aims to provide a desired aggregate grid-forming (GFM) response based on the coordination of power contributions between different DERs. Compared to existing DVPP designs with an AC-coupled AC-output configuration, a more generic modular DVPP design is proposed in this article, which comprises four types of basic DVPP modules, involving AC- or DC-coupling and AC- or DC-output, adequately accommodating diverse DER integration setups, such as AC, DC, AC/DC hybrid microgrids and renewable power plants. The control design is first developed for the four basic modules by the aggregation of DERs and the disaggregation of the control objectives, and then extended to modular DVPPs through a systematic top-down approach. The control performance is comprehensively validated through simulation. The modular DVPP design offers scalable and standardizable advanced grid interfaces (AGIs) for building and operating AC/DC hybrid power grids.

eess.SY↗

arXiv2024

Mixed-Precision Embeddings for Large-Scale Recommendation Models

Embedding techniques have become essential components of large databases in the deep learning era. By encoding discrete entities, such as words, items, or graph nodes, into continuous vector spaces, embeddings facilitate more efficient storage, retrieval, and processing in large databases. Especially in the domain of recommender systems, millions of categorical features are encoded as unique embedding vectors, which facilitates the modeling of similarities and interactions among features. However, numerous embedding vectors can result in significant storage overhead. In this paper, we aim to compress the embedding table through quantization techniques. Given that features vary in importance levels, we seek to identify an appropriate precision for each feature to balance model accuracy and memory usage. To this end, we propose a novel embedding compression method, termed Mixed-Precision Embeddings (MPE). Specifically, to reduce the size of the search space, we first group features by frequency and then search precision for each feature group. MPE further learns the probability distribution over precision levels for each feature group, which can be used to identify the most suitable precision with a specially designed sampling strategy. Extensive experiments on three public datasets demonstrate that MPE significantly outperforms existing embedding compression methods. Remarkably, MPE achieves about 200x compression on the Criteo dataset without comprising the prediction accuracy.

cs.IR↗

arXiv2024

Masked Random Noise for Communication Efficient Federated Learning

Federated learning is a promising distributed training paradigm that effectively safeguards data privacy. However, it may involve significant communication costs, which hinders training efficiency. In this paper, we aim to enhance communication efficiency from a new perspective. Specifically, we request the distributed clients to find optimal model updates relative to global model parameters within predefined random noise. For this purpose, we propose Federated Masked Random Noise (FedMRN), a novel framework that enables clients to learn a 1-bit mask for each model parameter and apply masked random noise (i.e., the Hadamard product of random noise and masks) to represent model updates. To make FedMRN feasible, we propose an advanced mask training strategy, called progressive stochastic masking (PSM). After local training, each client only need to transmit local masks and a random seed to the server. Additionally, we provide theoretical guarantees for the convergence of FedMRN under both strongly convex and non-convex assumptions. Extensive experiments are conducted on four popular datasets. The results show that FedMRN exhibits superior convergence speed and test accuracy compared to relevant baselines, while attaining a similar level of accuracy as FedAvg.

cs.LG↗

arXiv2024

Dynamic Ancillary Services: From Grid Codes to Transfer Function-Based Converter Control

Conventional grid-code specifications for dynamic ancillary services provision such as fast frequency and voltage regulation are typically defined by means of piece-wise linear step-response capability curves in the time domain. However, although the specification of such time-domain curves is straightforward, their practical implementation in a converter-based generation system is not immediate, and no customary methods have been developed yet. In this paper, we thus propose a systematic approach for the practical implementation of piece-wise linear time-domain curves to provide dynamic ancillary services by converter-based generation systems, while ensuring grid-code and device-level requirements to be reliably satisfied. Namely, we translate the piece-wise linear time-domain curves for active and reactive power provision in response to a frequency and voltage step change into a desired rational parametric transfer function in the frequency domain, which defines a dynamic response behavior to be realized by the converter. The obtained transfer function can be easily implemented e.g. via a PI-based matching control in the power loop of standard converter control architectures. We demonstrate the performance of our method in numerical grid-code compliance tests, and reveal its superiority over classical droop and virtual inertia schemes which may not satisfy the grid codes due to their structural limitations.

eess.SY↗

arXiv2024

Optimal Dynamic Ancillary Services Provision Based on Local Power Grid Perception

In this paper, we propose a systematic closed-loop approach to provide optimal dynamic ancillary services with converter-interfaced generation systems based on local power grid perception. In particular, we structurally encode dynamic ancillary services such as fast frequency and voltage regulation in the form of a parametric transfer function matrix, which includes several parameters to define a set of different feasible response behaviors, among which we aim to find the optimal one to be realized by the converter system. Our approach is based on a so-called "perceive-and-optimize" (P&O) strategy: First, we identify a grid dynamic equivalent at the interconnection terminals of the converter system. Second, we consider the closed-loop interconnection of the identified grid equivalent and the parametric transfer function matrix, which we optimize for the set of transfer function parameters, resulting in a stable and optimal closed-loop performance for ancillary services provision. In the process, we ensure that grid-code and device-level requirements are satisfied. Finally, we demonstrate the effectiveness of our approach in different numerical case studies based on a modified Kundur two-area test system.

eess.SY↗

arXiv2024

End-to-End Cost-Effective Incentive Recommendation under Budget Constraint with Uplift Modeling

In modern online platforms, incentives are essential factors that enhance user engagement and increase platform revenue. Over recent years, uplift modeling has been introduced as a strategic approach to assign incentives to individual customers. Especially in many real-world applications, online platforms can only incentivize customers with specific budget constraints. This problem can be reformulated as the multi-choice knapsack problem. This optimization aims to select the optimal incentive for each customer to maximize the return on investment. Recent works in this field frequently tackle the budget allocation problem using a two-stage approach. However, this solution is confronted with the following challenges: (1) The causal inference methods often ignore the domain knowledge in online marketing, where the expected response curve of a customer should be monotonic and smooth as the incentive increases. (2) An optimality gap between the two stages results in inferior sub-optimal allocation performance due to the loss of the incentive recommendation information for the uplift prediction under the limited budget constraint. To address these challenges, we propose a novel End-to-End Cost-Effective Incentive Recommendation (E3IR) model under budget constraints. Specifically, our methods consist of two modules, i.e., the uplift prediction module and the differentiable allocation module. In the uplift prediction module, we construct prediction heads to capture the incremental improvement between adjacent treatments with the marketing domain constraints (i.e., monotonic and smooth). We incorporate integer linear programming (ILP) as a differentiable layer input in the allocation module. Furthermore, we conduct extensive experiments on public and real product datasets, demonstrating that our E3IR improves allocation performance compared to existing two-stage approaches.

cs.IR↗

arXiv2024

Dynamic Complex-Frequency Control of Grid-Forming Converters

Complex droop control, alternatively known as dispatchable virtual oscillator control (dVOC), stands out for its unique capabilities in synchronization and voltage stabilization among existing control strategies for grid-forming converters. Complex droop control leverages the novel concept of ``complex frequency'', thereby establishing a coupled connection between active and reactive power inputs and frequency and rate-of-change-of voltage outputs. However, its reliance on static droop gains limits its ability to exhibit crucial dynamic response behaviors required in future power systems. To address this limitation, this paper introduces dynamic complex-frequency control, upgrading static droop gains with dynamic transfer functions to enhance the richness and flexibility in dynamic responses for frequency and voltage control. Unlike existing approaches, the complex-frequency control framework treats frequency and voltage dynamics collectively, ensuring small-signal stability for frequency synchronization and voltage stabilization simultaneously. The control framework is validated through detailed numerical case studies on the IEEE nine-bus system, also showcasing its applicability in multi-converter setups.

eess.SY↗

arXiv2024

Touch the Core: Exploring Task Dependence Among Hybrid Targets for Recommendation

As user behaviors become complicated on business platforms, online recommendations focus more on how to touch the core conversions, which are highly related to the interests of platforms. These core conversions are usually continuous targets, such as \textit{watch time}, \textit{revenue}, and so on, whose predictions can be enhanced by previous discrete conversion actions. Therefore, multi-task learning (MTL) can be adopted as the paradigm to learn these hybrid targets. However, existing works mainly emphasize investigating the sequential dependence among discrete conversion actions, which neglects the complexity of dependence between discrete conversions and the final continuous conversion. Moreover, simultaneously optimizing hybrid tasks with stronger task dependence will suffer from volatile issues where the core regression task might have a larger influence on other tasks. In this paper, we study the MTL problem with hybrid targets for the first time and propose the model named Hybrid Targets Learning Network (HTLNet) to explore task dependence and enhance optimization. Specifically, we introduce label embedding for each task to explicitly transfer the label information among these tasks, which can effectively explore logical task dependence. We also further design the gradient adjustment regime between the final regression task and other classification tasks to enhance the optimization. Extensive experiments on two offline public datasets and one real-world industrial dataset are conducted to validate the effectiveness of HTLNet. Moreover, online A/B tests on the financial recommender system also show that our model has improved significantly. Our implementation is available here\footnote{\url{https://github.com/fuyuanlyu/HTLNet}}.

cs.IR↗

arXiv2024

OptDist: Learning Optimal Distribution for Customer Lifetime Value Prediction

Customer Lifetime Value (CLTV) prediction is a critical task in business applications. Accurately predicting CLTV is challenging in real-world business scenarios, as the distribution of CLTV is complex and mutable. Firstly, there is a large number of users without any consumption consisting of a long-tailed part that is too complex to fit. Secondly, the small set of high-value users spent orders of magnitude more than a typical user leading to a wide range of the CLTV distribution which is hard to capture in a single distribution. Existing approaches for CLTV estimation either assume a prior probability distribution and fit a single group of distribution-related parameters for all samples, or directly learn from the posterior distribution with manually predefined buckets in a heuristic manner. However, all these methods fail to handle complex and mutable distributions. In this paper, we propose a novel optimal distribution selection model OptDist for CLTV prediction, which utilizes an adaptive optimal sub-distribution selection mechanism to improve the accuracy of complex distribution modeling. Specifically, OptDist trains several candidate sub-distribution networks in the distribution learning module (DLM) for modeling the probability distribution of CLTV. Then, a distribution selection module (DSM) is proposed to select the sub-distribution for each sample, thus making the selection automatically and adaptively. Besides, we design an alignment mechanism that connects both modules, which effectively guides the optimization. We conduct extensive experiments on both two public and one private dataset to verify that OptDist outperforms state-of-the-art baselines. Furthermore, OptDist has been deployed on a large-scale financial platform for customer acquisition marketing campaigns and the online experiments also demonstrate the effectiveness of OptDist.

cs.IR↗

arXiv2024

FedBAT: Communication-Efficient Federated Learning via Learnable Binarization

Federated learning is a promising distributed machine learning paradigm that can effectively exploit large-scale data without exposing users' privacy. However, it may incur significant communication overhead, thereby potentially impairing the training efficiency. To address this challenge, numerous studies suggest binarizing the model updates. Nonetheless, traditional methods usually binarize model updates in a post-training manner, resulting in significant approximation errors and consequent degradation in model accuracy. To this end, we propose Federated Binarization-Aware Training (FedBAT), a novel framework that directly learns binary model updates during the local training process, thus inherently reducing the approximation errors. FedBAT incorporates an innovative binarization operator, along with meticulously designed derivatives to facilitate efficient learning. In addition, we establish theoretical guarantees regarding the convergence of FedBAT. Extensive experiments are conducted on four popular datasets. The results show that FedBAT significantly accelerates the convergence and exceeds the accuracy of baselines by up to 9\%, even surpassing that of FedAvg in some cases.

cs.LG↗

arXiv2024

Saturation-Informed Current-Limiting Control for Grid-Forming Converters

In this paper, we investigate the transient stability of a state-of-the-art grid-forming complex-droop control (i.e., dispatchable virtual oscillator control, dVOC) under current saturation. We quantify the saturation level of a converter by introducing the concept of degree of saturation (DoS), and we propose a provably stable current-limiting control with saturation-informed feedback, which feeds the degree of saturation back to the inner voltage-control loop and the outer grid-forming loop. As a result, although the output current is saturated, the voltage phase angle can still be generated from an internal virtual voltage-source node that is governed by an equivalent complex-droop control. We prove that the proposed control achieves transient stability during current saturation under grid faults. We also provide parametric stability conditions for multi-converter systems under grid-connected and islanded scenarios. The stability performance of the current-limiting control is validated with various case studies.

eess.SY↗

arXiv2024

Complex-Frequency Synchronization of Converter-Based Power Systems

In this paper, we study phase-amplitude multivariable dynamics in converter-based power systems from a complex-frequency perspective. Complex frequency represents the rate of change of voltage amplitude and phase angle by its real and imaginary parts, respectively. This emerging notion is of significance as it accommodates the multivariable characteristics of power networks where active and reactive power are inherently coupled with both voltage amplitude and phase. We propose the notion of complex-frequency synchronization to study the phase-amplitude multivariable stability issue in a power system with dispatchable virtual oscillator-controlled (dVOC) converters. To achieve this, we separate the system into linear fast dynamics and approximately linear slow dynamics. The linearity property makes it tractable to analyze fast complex-frequency synchronization and slower voltage stabilization. From the perspective of complex frequency and complex-frequency synchronization, we provide novel insights into the equivalence of dVOC and complex-power-frequency droop control, stability analysis methods, and stability criteria. Our study offers a practical solution to address challenging stability issues in converter-based power systems.

eess.SY↗

arXiv2024

Rankability-enhanced Revenue Uplift Modeling Framework for Online Marketing

Uplift modeling has been widely employed in online marketing by predicting the response difference between the treatment and control groups, so as to identify the sensitive individuals toward interventions like coupons or discounts. Compared with traditional \textit{conversion uplift modeling}, \textit{revenue uplift modeling} exhibits higher potential due to its direct connection with the corporate income. However, previous works can hardly handle the continuous long-tail response distribution in revenue uplift modeling. Moreover, they have neglected to optimize the uplift ranking among different individuals, which is actually the core of uplift modeling. To address such issues, in this paper, we first utilize the zero-inflated lognormal (ZILN) loss to regress the responses and customize the corresponding modeling network, which can be adapted to different existing uplift models. Then, we study the ranking-related uplift modeling error from the theoretical perspective and propose two tighter error bounds as the additional loss terms to the conventional response regression loss. Finally, we directly model the uplift ranking error for the entire population with a listwise uplift ranking loss. The experiment results on offline public and industrial datasets validate the effectiveness of our method for revenue uplift modeling. Furthermore, we conduct large-scale experiments on a prominent online fintech marketing platform, Tencent FiT, which further demonstrates the superiority of our method in real-world applications.

cs.LG↗

arXiv2024

Benchmarking for Deep Uplift Modeling in Online Marketing

Online marketing is critical for many industrial platforms and business applications, aiming to increase user engagement and platform revenue by identifying corresponding delivery-sensitive groups for specific incentives, such as coupons and bonuses. As the scale and complexity of features in industrial scenarios increase, deep uplift modeling (DUM) as a promising technique has attracted increased research from academia and industry, resulting in various predictive models. However, current DUM still lacks some standardized benchmarks and unified evaluation protocols, which limit the reproducibility of experimental results in existing studies and the practical value and potential impact in this direction. In this paper, we provide an open benchmark for DUM and present comparison results of existing models in a reproducible and uniform manner. To this end, we conduct extensive experiments on two representative industrial datasets with different preprocessing settings to re-evaluate 13 existing models. Surprisingly, our experimental results show that the most recent work differs less than expected from traditional work in many cases. In addition, our experiments also reveal the limitations of DUM in generalization, especially for different preprocessing and test distributions. Our benchmarking work allows researchers to evaluate the performance of new models quickly but also reasonably demonstrates fair comparison results with existing models. It also gives practitioners valuable insights into often overlooked considerations when deploying DUM. We will make this benchmarking library, evaluation protocol, and experimental setup available on GitHub.

cs.LG↗

arXiv2024

Quantitative Stability Conditions for Grid-Forming Converters With Complex Droop Control

In this paper, we analytically study the transient stability of grid-connected converters with grid-forming complex droop control, also known as dispatchable virtual oscillator control. We prove theoretically that complex droop control, as a state-of-the-art grid-forming control, always possesses steady-state equilibria whereas classical droop control does not. We provide quantitative conditions for complex droop control maintaining transient stability (global asymptotic stability) under grid disturbances, which is beyond the well-established local (non-global) stability for classical droop control. For the transient instability of complex droop control, we reveal that the unstable trajectories are bounded, manifesting as limit cycle oscillations. Moreover, we extend our stability results from second-order grid-forming control dynamics to full-order system dynamics that additionally encompass both circuit electromagnetic transients and inner-loop dynamics. Our theoretical results contribute an insightful understanding of the transient stability and instability of complex droop control and offer practical guidelines for parameter tuning and stability guarantees.

eess.SY↗