arXiv subjects

Yingzhao Lian

Publications and source records attributed to Yingzhao Lian.

At least 19 recordsLinked to original sources

arXiv2026

Boogu-Image-0.1: Boosting Open Agentic Multimodal Generation via Understanding under a Minimal Budget

We introduce Boogu-Image-0.1, an open-source unified multimodal understanding and generation model family, comprising Base, Turbo, Edit, and Edit-Turbo variants. It delivers competitive performance in high-quality text-to-image generation, fast inference, instruction-based editing, and bilingual (Chinese-English) text rendering. Closed-source multimodal systems like Nano-Banana-Pro and GPT-Image-2 achieve strong performance through system-level integration rather than a single model, yet their internal practices remain largely undisclosed. In this work, we demonstrate that strengthening the understanding capability of the system, through a stronger multimodal encoder, agentic prompt rewriting, and related techniques, together with improvements in data quality, training pipelines, and agentic inference-time scaling, can substantially enhance generation and editing performance even under highly constrained compute budgets. Comprehensive evaluations show that Boogu-Image-0.1 consistently matches or surpasses other open-source models across standard benchmarks, and achieves results approaching leading closed-source systems. Notably, this is accomplished with only 208.62 million unique images. The base model's theoretical training cost is only approximately \$400K. We share practical discussions that we believe are valuable to the broader research community, and release weights, code, and recipes under Apache 2.0 to advance the open ecosystem for unified multimodal understanding and generation. Our code is available here: https://github.com/Boogu-Project/Boogu-Image.

cs.CV↗

arXiv2025

DiLA: Enhancing LLM Tool Learning with Differential Logic Layer

Considering the challenges faced by large language models (LLMs) in logical reasoning and planning, prior efforts have sought to augment LLMs with access to external solvers. While progress has been made on simple reasoning problems, solving classical constraint satisfaction problems, such as the Boolean Satisfiability Problem (SAT) and Graph Coloring Problem (GCP), remains difficult for off-the-shelf solvers due to their intricate expressions and exponential search spaces. In this paper, we propose a novel differential logic layer-aided language modeling (DiLA) approach, where logical constraints are integrated into the forward and backward passes of a network layer, to provide another option for LLM tool learning. In DiLA, LLM aims to transform the language description to logic constraints and identify initial solutions of the highest quality, while the differential logic layer focuses on iteratively refining the LLM-prompted solution. Leveraging the logic layer as a bridge, DiLA enhances the logical reasoning ability of LLMs on a range of reasoning problems encoded by Boolean variables, guaranteeing the efficiency and correctness of the solution process. We evaluate the performance of DiLA on two classic reasoning problems and empirically demonstrate its consistent outperformance against existing prompt-based and solver-aided approaches.

cs.CL↗

arXiv2025

Scaling Up, Speeding Up: A Benchmark of Speculative Decoding for Efficient LLM Test-Time Scaling

Test-time scaling has emerged as a powerful paradigm for enhancing the reasoning capabilities of large language models (LLMs) by allocating additional computational resources during inference. However, this paradigm is inherently inefficient due to the generation of redundant and repetitive reasoning traces, leading to significant computational overhead. Speculative decoding offers a promising avenue for mitigating this inefficiency, yet its efficacy in the structured, repetition-rich context of test-time scaling remains largely unexplored. To bridge this gap, we introduce the first comprehensive benchmark designed to evaluate speculative decoding methods for accelerating LLM test-time scaling. Our benchmark provides consistent experimental protocols across representative test-time scaling paradigms (e.g., Best-of-N sampling and multi-round thinking), enabling a fair comparison of three major categories of speculative decoding: model-based, training-based, and n-gram-based methods. Extensive experiments reveal that simple n-gram-based methods effectively capture repetitive patterns, demonstrating unique potential in accelerating test-time scaling. This phenomenon demonstrates the value of integrating n-gram-based methods with model-based or training-based approaches to balance acceleration for both repetitive and diverse reasoning in test-time scaling. We hope this benchmark spurs further research on speculative decoding for test-time scaling, enabling faster and more practical reasoning in LLMs through better handling of repetitive and diverse reasoning paths.

cs.CL↗

arXiv2024

Adaptive Data-Driven Prediction in a Building Control Hierarchy: A Case Study of Demand Response in Switzerland

By providing various services, such as Demand Response (DR), buildings can play a crucial role in the energy market due to their significant energy consumption. However, effectively commissioning buildings for such desired functionalities requires significant expert knowledge and design effort, considering the variations in building dynamics and intended use. In this study, we introduce an adaptive data-driven prediction scheme based on Willems' Fundamental Lemma within the building control hierarchy. This scheme offers a versatile, flexible, and user-friendly interface for diverse prediction and control objectives. We provide an easy-to-use tuning process and an adaptive update pipeline for the scheme, both validated through extensive prediction tests. We evaluate the proposed scheme by coordinating a building and an energy storage system to provide Secondary Frequency Control (SFC) in a Swiss DR program. Specifically, we integrate the scheme into a three-layer hierarchical SFC control framework, and each layer of this hierarchy is designed to achieve distinct operational goals. Apart from its flexibility, our approach significantly improves cost efficiency, resulting in a 28.74% reduction in operating costs compared to a conventional control scheme, as demonstrated by a 52-day experiment in an actual building. Our findings emphasize the potential of the proposed scheme to reduce the commissioning costs of advanced building control strategies and to facilitate the adoption of new techniques in building control.

eess.SY↗

arXiv2024

SeaDAG: Semi-autoregressive Diffusion for Conditional Directed Acyclic Graph Generation

We introduce SeaDAG, a semi-autoregressive diffusion model for conditional generation of Directed Acyclic Graphs (DAGs). Considering their inherent layer-wise structure, we simulate layer-wise autoregressive generation by designing different denoising speed for different layers. Unlike conventional autoregressive generation that lacks a global graph structure view, our method maintains a complete graph structure at each diffusion step, enabling operations such as property control that require the full graph structure. Leveraging this capability, we evaluate the DAG properties during training by employing a graph property decoder. We explicitly train the model to learn graph conditioning with a condition loss, which enhances the diffusion model's capacity to generate graphs that are both realistic and aligned with specified properties. We evaluate our method on two representative conditional DAG generation tasks: (1) circuit generation from truth tables, where precise DAG structures are crucial for realizing circuit functionality, and (2) molecule generation based on quantum properties. Our approach demonstrates promising results, generating high-quality and realistic DAGs that closely align with given conditions.

cs.LG↗

arXiv2024

Robustly Learning Regions of Attraction from Fixed Data

While stability analysis is a mainstay for control science, especially computing regions of attraction of equilibrium points, until recently most stability analysis tools always required explicit knowledge of the model or a high-fidelity simulator representing the system at hand. In this work, a new data-driven Lyapunov analysis framework is proposed. Without using the model or its simulator, the proposed approach can learn a piece-wise affine Lyapunov function with a finite and fixed off-line dataset. The learnt Lyapunov function is robust to any dynamics that are consistent with the off-line dataset, and its computation is based on second order cone programming. Along with the development of the proposed scheme, a slight generalization of classical Lyapunov stability criteria is derived, enabling an iterative inference algorithm to augment the region of attraction.

math.OC↗

arXiv2024

RTLRewriter: Methodologies for Large Models aided RTL Code Optimization

Register Transfer Level (RTL) code optimization is crucial for enhancing the efficiency and performance of digital circuits during early synthesis stages. Currently, optimization relies heavily on manual efforts by skilled engineers, often requiring multiple iterations based on synthesis feedback. In contrast, existing compiler-based methods fall short in addressing complex designs. This paper introduces RTLRewriter, an innovative framework that leverages large models to optimize RTL code. A circuit partition pipeline is utilized for fast synthesis and efficient rewriting. A multi-modal program analysis is proposed to incorporate vital visual diagram information as optimization cues. A specialized search engine is designed to identify useful optimization guides, algorithms, and code snippets that enhance the model ability to generate optimized RTL. Additionally, we introduce a Cost-aware Monte Carlo Tree Search (C-MCTS) algorithm for efficient rewriting, managing diverse retrieved contents and steering the rewriting results. Furthermore, a fast verification pipeline is proposed to reduce verification cost. To cater to the needs of both industry and academia, we propose two benchmarking suites: the Large Rewriter Benchmark, targeting complex scenarios with extensive circuit partitioning, optimization trade-offs, and verification challenges, and the Small Rewriter Benchmark, designed for a wider range of scenarios and patterns. Our comparative analysis with established compilers such as Yosys and E-graph demonstrates significant improvements, highlighting the benefits of integrating large models into the early stages of circuit design. We provide our benchmarks at https://github.com/yaoxufeng/RTLRewriter-Bench.

cs.AR↗

arXiv2024

Efficient Recursive Data-enabled Predictive Control (Extended Version)

In the field of model predictive control, Data-enabled Predictive Control (DeePC) offers direct predictive control, bypassing traditional modeling. However, challenges emerge with increased computational demand due to recursive data updates. This paper introduces a novel recursive updating algorithm for DeePC. It emphasizes the use of Singular Value Decomposition (SVD) for efficient low-dimensional transformations of DeePC in its general form, as well as a fast SVD update scheme. Importantly, our proposed algorithm is highly flexible due to its reliance on the general form of DeePC, which is demonstrated to encompass various data-driven methods that utilize Pseudoinverse and Hankel matrices. This is exemplified through a comparison to Subspace Predictive Control, where the algorithm achieves asymptotically consistent prediction for stochastic linear time-invariant systems. Our proposed methodologies' efficacy is validated through simulation studies.

eess.SY↗

arXiv2023

Hypergraph-Based Fast Distributed AC Power Flow Optimization

This paper presents a novel distributed approach for solving AC power flow (PF) problems. The optimization problem is reformulated into a distributed form using a communication structure corresponding to a hypergraph, by which complex relationships between subgrids can be expressed as hyperedges. Then, a hypergraph-based distributed sequential quadratic programming (HDQ) approach is proposed to handle the reformulated problems, and the hypergraph-based distributed sequential quadratic programming (HDSQP) is used as the inner algorithm to solve the corresponding QP subproblems, which are respectively condensed using Schur complements with respect to coupling variables defined by hyperedges. Furthermore, we rigorously establish the convergence guarantee of the proposed algorithm with a locally quadratic rate and the one-step convergence of the inner algorithm when using the Levenberg-Marquardt regularization. Our analysis also demonstrates that the computational complexity of the proposed algorithm is much lower than the state-of-art distributed algorithm. We implement the proposed algorithm in an open-source toolbox, i.e., rapidPF, and conduct numerical tests that validate the proof and demonstrate the great potential of the proposed distributed algorithm in terms of communication effort and computational speed.

math.OC↗

arXiv2023

A Proximal-Point Lagrangian Based Parallelizable Nonconvex Solver for Bilinear Model Predictive Control

Nonlinear model predictive control has been widely adopted to manipulate bilinear systems with dynamics that include products of the inputs and the states. These systems are ubiquitous in chemical processes, mechanical systems, and quantum physics, to name a few. Running a bilinear MPC controller in real time requires solving a non-convex optimization problem within a limited sampling time. This paper proposes a novel parallel proximal-point Lagrangian based bilinear MPC solver via an interlacing horizon-splitting scheme. The resulting algorithm converts the non-convex MPC control problem into a set of parallelizable small-scale multi-parametric quadratic programs (mpQPs) and an equality-constrained linear-quadratic regulator problem. As a result, the solutions of mpQPs can be pre-computed offline to enable efficient online computation. The proposed algorithm is validated on a simulation of an HVAC system control. It is deployed on a TI LaunchPad XL F28379D microcontroller to execute speed control on a field-controlled DC motor, where the MPC updates at 10 ms and solves the problem in 1.764 ms on average and at most 2.088 ms.

eess.SY↗

arXiv2023

Fault Detection via Occupation Kernel Principal Component Analysis

The reliable operation of automatic systems is heavily dependent on the ability to detect faults in the underlying dynamical system. While traditional model-based methods have been widely used for fault detection, data-driven approaches have garnered increasing attention due to their ease of deployment and minimal need for expert knowledge. In this paper, we present a novel principal component analysis (PCA) method that uses occupation kernels. Occupation kernels result in feature maps that are tailored to the measured data, have inherent noise-robustness due to the use of integration, and can utilize irregularly sampled system trajectories of variable lengths for PCA. The occupation kernel PCA method is used to develop a reconstruction error approach to fault detection and its efficacy is validated using numerical simulations.

stat.ML↗

arXiv2023

Physically Consistent Multiple-Step Data-Driven Predictions Using Physics-based Filters

(Extended Version) Data-driven control can facilitate the rapid development of controllers, offering an alternative to conventional approaches. In order to maintain consistency between any known underlying physical laws and a data-driven decision-making process, preprocessing of raw data is necessary to account for measurement noise and any inconsistencies it may introduce. In this paper, we present a physics-based filter to achieve this and demonstrate its effectiveness through practical applications, using real-world datasets collected in a building on the Ecole Polytechnique Federale de Lausanne (EPFL) campus. Two distinct use cases are explored: indoor temperature control and demand response bidding.

eess.SY↗

arXiv2023

Distributed data-driven predictive control for cooperatively smoothing mixed traffic flow

Cooperative control of connected and automated vehicles (CAVs) promises smoother traffic flow. In mixed traffic, where human-driven vehicles with unknown dynamics coexist, data-driven predictive control techniques allow for CAV safe and optimal control with measurable traffic data. However, the centralized control setting in most existing strategies limits their scalability for large-scale mixed traffic flow. To address this problem, this paper proposes a cooperative DeeP-LCC (Data-EnablEd Predictive Leading Cruise Control) formulation and its distributed implementation algorithm. In cooperative DeeP-LCC, the traffic system is naturally partitioned into multiple subsystems with one single CAV, which collects local trajectory data for subsystem behavior predictions based on the Willems' fundamental lemma. Meanwhile, the cross-subsystem interaction is formulated as a coupling constraint. Then, we employ the Alternating Direction Method of Multipliers (ADMM) to design the distributed DeeP-LCC algorithm. This algorithm achieves both computation and communication efficiency, as well as trajectory data privacy, through parallel calculation. Our simulations on different traffic scales verify the real-time wave-dampening potential of distributed DeeP-LCC, which can reduce fuel consumption by over 31.84% in a large-scale traffic system of 100 vehicles with only 5%-20% CAVs.

eess.SY↗

arXiv2023

Adaptive Robust Data-driven Building Control via Bi-level Reformulation: an Experimental Result

Data-driven control approaches for the minimization of energy consumption of buildings have the potential to significantly reduce deployment costs and increase uptake of advanced control in this sector. A number of recent approaches based on the application of Willems' fundamental lemma for data-driven controller design from input/output measurements are very promising for deterministic LTI systems. This paper \change{proposes a systematic way to handle unknown measurement noise and measurable process noise}, and extends these data-driven control schemes to adaptive building control via a robust bi-level formulation, whose upper level ensures robustness and whose lower level guarantees prediction quality. Corresponding numerical improvements and an active excitation mechanism are proposed to enable a computationally efficient reliable operation. The efficacy of the proposed scheme is validated by \change{a multi-zone building simulation} and a real-world experiment on a single-zone conference building on the EPFL campus. The real-world experiment includes a 20-day non-stop test, where, without extra modeling effort, our proposed controller improves 18.4\% energy efficiency against an industry-standard controller, while also robustly ensuring occupant comfort.

eess.SY↗

arXiv2022

Scheduling Delays and Curtailment for Household Appliances with Deterministic Load Profiles using MPC

Smart home appliances can time-shift and curtail their power demand to assist demand side management or allow operation with limited power, as in an off-grid application. This paper proposes a scheduling process to start appliances with time-varying deterministic load profiles. Self-triggered model predictive control is used to limit the household net power demand below a given threshold. Meanwhile, deterministic load profiles are more difficult to schedule compared to variable charging or thermal loads because system failure will occur once power demand is not satisfied. The proposed scheme formulates the decision of the load shifting time as a continuous optimization problem, and an inhomogeneous time grid system is introduced to handle the optimization of different appliances and their consensus at this resolution. The efficacy of the proposed scheme is studied by numerical comparison with a mixed-integer MPC controller and by a case study of three home appliances and an interruptible washing machine.

eess.SY↗

arXiv2022

Lessons Learned from Data-Driven Building Control Experiments: Contrasting Gaussian Process-based MPC, Bilevel DeePC, and Deep Reinforcement Learning

This manuscript offers the perspective of experimentalists on a number of modern data-driven techniques: model predictive control relying on Gaussian processes, adaptive data-driven control based on behavioral theory, and deep reinforcement learning. These techniques are compared in terms of data requirements, ease of use, computational burden, and robustness in the context of real-world applications. Our remarks and observations stem from a number of experimental investigations carried out in the field of building control in diverse environments, from lecture halls and apartment spaces to a hospital surgery center. The final goal is to support others in identifying what technique is best suited to tackle their own problems.

eess.SY↗

arXiv2022

Data-driven input reconstruction and experimental validation

This paper addresses a data-driven input reconstruction problem based on Willems' Fundamental Lemma in which unknown input estimators (UIEs) are constructed directly from historical I/O data. Given only output measurements, the inputs are estimated by the UIE, which is shown to asymptotically converge to the true input without knowing the initial conditions. Both open-loop and closed-loop UIEs are developed based on Lyapunov conditions and the Luenberger-observer-type feedback, whose convergence properties are studied. An experimental study is presented demonstrating the efficacy of the closed-loop UIE for estimating the occupancy of a building on the EPFL campus via measured carbon dioxide levels.

eess.SY↗

arXiv2022

On the Optimality and Convergence Properties of the Iterative Learning Model Predictive Controller

In this technical note we analyse the performance improvement and optimality properties of the Learning Model Predictive Control (LMPC) strategy for linear deterministic systems. The LMPC framework is a policy iteration scheme where closed-loop trajectories are used to update the control policy for the next execution of the control task. We show that, when a Linear Independence Constraint Qualification (LICQ) condition holds, the LMPC scheme guarantees strict iterative performance improvement and optimality, meaning that the closed-loop cost evaluated over the entire task converges asymptotically to the optimal cost of the infinite-horizon control problem. Compared to previous works this sufficient LICQ condition can be easily checked, it holds for a larger class of systems and it can be used to adaptively select the prediction horizon of the controller, as demonstrated by a numerical example.

math.OC↗