PyTorchKR Papers

최신 논문

총 516편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2504.08535

Secondary Safety Control for Systems with Sector Bounded Nonlinearities [Extended Version]

We consider the problem of safety verification and safety-aware controller synthesis for systems with sector bounded nonlinearities. We aim to keep the states of the system within a given safe set under potential actuator and sensor attacks. Specifically, we adopt the setup that a controller has already been designed to stabilize the plant. Using invariant sets and barrier certificate theory, we first give sufficient conditions to verify the safety of the closed-loop system under attacks. Furthermore, by using a subset of sensors that are assumed to be free of attacks, we provide a synthesis method for a secondary controller that enhances the safety of the system. The sufficient conditions to verify safety are derived using Lyapunov-based tools and the S-procedure. Using the projection lemma, the conditions are then formulated as linear matrix inequality (LMI) problems which can be solved efficiently. Lastly, our theoretical results are illustrated through numerical simulations.

arXiv:2504.06403

Leveraging Non-Steady-State Frequency-Domain Data in Willems' Fundamental Lemma

Willems' fundamental lemma enables data-driven analysis and control by characterizing an unknown system's behavior directly in terms of measured data. In this work, we extend a recent frequency-domain variant of this result--previously limited to steady-state data--to incorporate non-steady-state data including transient phenomena. This approach eliminates the need to wait for transients to decay during data collection, significantly reducing the experiment duration. Unlike existing frequency-domain system identification methods, our approach integrates transient data without preprocessing, making it well-suited for direct data-driven analysis and control. We demonstrate its effectiveness by isolating transients in the collected data and performing FRF evaluation at arbitrary frequencies in a numerical case study including noise.

arXiv:2504.04942

Lemmanaid: Neuro-Symbolic Lemma Conjecturing

Automatically conjecturing useful, interesting and novel lemmas would greatly improve automated reasoning tools and lower the bar for formalizing mathematics in proof assistants. It is however a very challenging task for both neural and symbolic approaches. We present the first steps towards a practical neuro-symbolic lemma conjecturing tool, Lemmanaid, that combines Large Language Models (LLMs) and symbolic methods, and evaluate it on proof libraries for the Isabelle proof assistant. We train an LLM to generate lemma templates that describe the shape of a lemma, and use symbolic methods to fill in the details. We compare Lemmanaid against an LLM trained to generate complete lemma statements as well as previous fully symbolic conjecturing methods. Lemmanaid outperforms both neural and symbolic methods on test sets from Isabelle's HOL library and from its Archive of Formal Proofs, discovering between 29-39.5% of the gold standard human written lemmas. This is 8-15% more lemmas than the neural-only method. By leveraging the best of both symbolic and neural methods we can generate useful lemmas for a wide range of input domains, facilitating computer-assisted theory development and formalization.

arXiv:2503.21608

Nonlinear Multiple Response Regression and Learning of Latent Spaces

Identifying low-dimensional latent structures within high-dimensional data has long been a central topic in the machine learning community, driven by the need for data compression, storage, transmission, and deeper data understanding. Traditional methods, such as principal component analysis (PCA) and autoencoders (AE), operate in an unsupervised manner, ignoring label information even when it is available. In this work, we introduce a unified method capable of learning latent spaces in both unsupervised and supervised settings. We formulate the problem as a nonlinear multiple-response regression within an index model context. By applying the generalized Stein's lemma, the latent space can be estimated without knowing the nonlinear link functions. Our method can be viewed as a nonlinear generalization of PCA. Moreover, unlike AE and other neural network methods that operate as "black boxes", our approach not only offers better interpretability but also reduces computational complexity while providing strong theoretical guarantees. Comprehensive numerical experiments and real data analyses demonstrate the superior performance of our method.

arXiv:2503.21040

Local Stability and Stabilization of Quadratic-Bilinear Systems using Petersen's Lemma

Quadratic-bilinear (QB) systems arise in many areas of science and engineering. In this paper, we present a scalable approach for designing locally stabilizing state-feedback control laws and certifying the local stability of QB systems. Sufficient conditions are established for local stability and stabilization based on quadratic Lyapunov functions, which also provide ellipsoidal inner-estimates for the region of attraction and region of stabilizability of an equilibrium point. Our formulation exploits Petersen's Lemma to convert the problem of certifying the sign-definiteness of the Lyapunov condition into a line search over a single scalar parameter. The resulting linear matrix inequality (LMI) conditions scale quadratically with the state dimension for both stability analysis and control synthesis, thus enabling analysis and control of QB systems with hundreds of state variables without resorting to specialized implementations. We demonstrate the approach on three benchmark problems from the existing literature. In all cases, we find our formulation yields comparable approximations of stability domains as determined by other established tools that are otherwise restricted to systems with up to tens of state variables.

arXiv:2503.19605공식 코드 · Papers with Code

Lean Formalization of Generalization Error Bound by Rademacher Complexity

We formalize the generalization error bound using Rademacher complexity in the Lean 4 theorem prover. Generalization error quantifies the gap between a learning machine's performance on given training data versus unseen test data, and Rademacher complexity serves as an estimate of this error based on the complexity of learning machines, or hypothesis class. Unlike traditional methods such as PAC learning and VC dimension, Rademacher complexity is applicable across diverse machine learning scenarios including deep learning and kernel methods. We formalize key concepts and theorems, including the empirical and population Rademacher complexities, and establish generalization error bounds through formal proofs of McDiarmid's inequality, Hoeffding's lemma, and symmetrization arguments.

arXiv:2503.19867

Geometric Meta-Learning via Coupled Ricci Flow: Unifying Knowledge Representation and Quantum Entanglement

This paper establishes a unified framework integrating geometric flows with deep learning through three fundamental innovations. First, we propose a thermodynamically coupled Ricci flow that dynamically adapts parameter space geometry to loss landscape topology, formally proved to preserve isometric knowledge embedding (Theorem~\ref{thm:isometric}). Second, we derive explicit phase transition thresholds and critical learning rates (Theorem~\ref{thm:critical}) through curvature blowup analysis, enabling automated singularity resolution via geometric surgery (Lemma~\ref{lem:surgery}). Third, we establish an AdS/CFT-type holographic duality (Theorem~\ref{thm:ads}) between neural networks and conformal field theories, providing entanglement entropy bounds for regularization design. Experiments demonstrate 2.1$\times$ convergence acceleration and 63\% topological simplification while maintaining $\mathcal{O}(N\log N)$ complexity, outperforming Riemannian baselines by 15.2\% in few-shot accuracy. Theoretically, we prove exponential stability (Theorem~\ref{thm:converge}) through a new Lyapunov function combining Perelman entropy with Wasserstein gradient flows, fundamentally advancing geometric deep learning.

arXiv:2503.17439공식 코드 · Papers with Code

LEMMA: Learning from Errors for MatheMatical Advancement in LLMs

Large language models (LLMs) have demonstrated remarkable reasoning capability in solving mathematical problems. However, existing approaches primarily focus on improving the quality of correct training data, e.g., distilling high-quality correct solutions from advanced models, neglecting the value contained in error data, potentially hindering the model's reflective ability. Though some studies attempt to leverage error data, they often involve complex mechanisms, such as Monte Carlo Tree Search (MCTS) to explore error nodes. In this work, we propose to enhance LLMs' reasoning ability by Learning from Errors for Mathematical Advancement (LEMMA). LEMMA constructs data consisting of an incorrect solution with an erroneous step and a reflection connection to a correct solution for fine-tuning. Specifically, we systematically analyze the model-generated error types and introduce an error-type grounded mistake augmentation method to collect diverse and representative errors. Correct solutions are either from fixing the errors or generating a fresh start. Through a model-aware smooth reflection connection, the erroneous solution is transferred to the correct one. By fine-tuning on the constructed dataset, the model is able to self-correct errors autonomously within the generation process without relying on external critique models. Experimental results demonstrate that LEMMA achieves significant performance improvements over other strong baselines.

arXiv:2503.17718공식 코드 · Papers with Code

Flexible WMMSE Beamforming for MU-MIMO Movable Antenna Communications

Movable antennas offer new potential for wireless communication by introducing degrees of freedom in antenna positioning, which has recently been explored for improving sum rates. In this paper, we aim to fully leverage the capabilities of movable antennas (MAs) by assuming that both the transmitter and receiver can optimize their antenna positions in multi-user multiple-input multiple-output (MU-MIMO) communications. Recognizing that WMMSE beamforming is a highly effective method for maximizing the MU-MIMO sum rate, we modify it to integrate antenna position optimization for MA systems, which we refer to as flexible WMMSE (F-WMMSE) beamforming. Importantly, we reformulate the subproblems within WMMSE to develop regularized sparse optimization frameworks to achieve joint beamforming (antenna coefficient optimization) and element movement (antenna position optimization). We then propose a regularized least squares-based simultaneous orthogonal matching pursuit (RLS-SOMP) algorithm to address the resulting sparse optimization problem. To enhance practical applications, the low-complexity implementation of the proposed framework is developed based on the pre-calculations and matrix inverse lemma. The overall F-WMMSE algorithm converges similarly to WMMSE, and our findings indicate that F-WMMSE achieves a significant sum rate improvement compared to traditional WMMSE, exceeding 20% under appropriate simulation conditions

arXiv:2503.11075

Inverter Control with Time-Varying and Nonconvex State and Input Constraints

The growing integration of inverter-based resources (IBRs) into modern power systems poses significant challenges for maintaining reliable operation under dynamic and constrained conditions. This paper focuses on the power tracking problem for grid-connected IBRs, addressing the complexities introduced by voltage and power factor constraints. Voltage constraints, being time-varying and nonlinear input constraints, often conflict with power factor constraints, which are state constraints. These conflicts, coupled with stability requirements, add substantial complexity to control design. To overcome these challenges, we propose a computationally efficient static state-feedback controller that guarantees stability and satisfies operational constraints. The concept of achievability is introduced to evaluate whether power setpoints can be accurately tracked while adhering to all constraints. Using a parameterization framework and the S-lemma, we develop criteria to assess and maximize the continuous achievable region for IBR operation. This framework allows system operators to ensure safety and stability by precomputing a finite set of control gains, significantly reducing online computational requirements. The proposed approach is validated through simulations, demonstrating its effectiveness in handling time-varying grid disturbances and achieving reliable control performance.

arXiv:2503.10580

On the Injective Norm of Sums of Random Tensors and the Moments of Gaussian Chaoses

We prove an upper bound on the expected $\ell_p$ injective norm of sums of subgaussian random tensors. Our proof is simple and does not rely on any explicit geometric or chaining arguments. Instead, it follows from a simple application of the PAC-Bayesian lemma, a tool that has proven effective at controlling the suprema of certain ``smooth'' empirical processes in recent years. Our bound strictly improves a very recent result of Bandeira, Gopi, Jiang, Lucca, and Rothvoss. In the Euclidean case ($p=2$), our bound sharpens a result of Lata{\l}a that was central to proving his estimates on the moments of Gaussian chaoses. As a consequence, we obtain an elementary proof of this fundamental result.

arXiv:2503.08081공식 코드 · Papers with Code

TRUST: Stability and Safety Controller Synthesis for Unknown Dynamical Models Using a Single Trajectory

TRUST is an open-source software tool developed for data-driven controller synthesis of dynamical systems with unknown mathematical models, ensuring either stability or safety properties. By collecting only a single input-state trajectory from the unknown system and satisfying a rank condition that ensures the system is persistently excited according to the Willems et al.'s fundamental lemma, TRUST aims to design either control Lyapunov functions (CLF) or control barrier certificates (CBC), along with their corresponding stability or safety controllers. The tool implements sum-of-squares (SOS) optimization programs solely based on data to enforce stability or safety properties across four system classes: (i) continuous-time nonlinear polynomial systems, (ii) continuous-time linear systems, (iii) discrete-time nonlinear polynomial systems, and (iv) discrete-time linear systems. TRUST is a Python-based web application featuring an intuitive, reactive graphic user interface (GUI) built with web technologies. It can be accessed at https://trust.tgo.dev or installed locally, and supports both manual data entry and data file uploads. Leveraging the power of the Python backend and a JavaScript frontend, TRUST is designed to be highly user-friendly and accessible across desktop, laptop, tablet, and mobile devices. We apply TRUST to a set of physical benchmarks with unknown dynamics, ensuring either stability or safety properties across the four supported classes of models.

arXiv:2503.03908

On the Convergence of Adam-Type Algorithm for Bilevel Optimization under Unbounded Smoothness

Adam has become one of the most popular optimizers for training modern deep neural networks, such as transformers. However, its applicability is largely restricted to single-level optimization problems. In this paper, we aim to extend vanilla Adam to tackle bilevel optimization problems, which have important applications in machine learning, such as meta-learning. In particular, we study stochastic bilevel optimization problems where the lower-level function is strongly convex and the upper-level objective is nonconvex with potentially unbounded smoothness. This unbounded smooth objective function covers a broad class of neural networks, including transformers, which may exhibit non-Lipschitz gradients. In this work, we introduce AdamBO, a single-loop Adam-type method that achieves $\widetilde{O}(\epsilon^{-4})$ oracle complexity to find $\epsilon$-stationary points, where the oracle calls involve stochastic gradient or Hessian/Jacobian-vector product evaluations. The key to our analysis is a novel randomness decoupling lemma that provides refined control over the lower-level variable. We conduct extensive experiments on various machine learning tasks involving bilevel formulations with recurrent neural networks (RNNs) and transformers, demonstrating the effectiveness of our proposed Adam-type algorithm.

arXiv:2503.04068

Quantitative Flow Approximation Properties of Narrow Neural ODEs

In this note, we revisit the problem of flow approximation properties of neural ordinary differential equations (NODEs). The approximation properties have been considered as a flow controllability problem in recent literature. The neural ODE is considered {\it narrow} when the parameters have dimension equal to the input of the neural network, and hence have limited width. We derive the relation of narrow NODEs in approximating flows of shallow but wide NODEs. Due to existing results on approximation properties of shallow neural networks, this facilitates understanding which kind of flows of dynamical systems can be approximated using narrow neural ODEs. While approximation properties of narrow NODEs have been established in literature, the proofs often involve extensive constructions or require invoking deep controllability theorems from control theory. In this paper, we provide a simpler proof technique that involves only ideas from ODEs and Gr{\"o}nwall's lemma. Moreover, we provide an estimate on the number of switches needed for the time dependent weights of the narrow NODE to mimic the behavior of a NODE with a single layer wide neural network as the velocity field.

arXiv:2503.01720공식 코드 · Papers with Code

Quality Measures for Dynamic Graph Generative Models

Deep generative models have recently achieved significant success in modeling graph data, including dynamic graphs, where topology and features evolve over time. However, unlike in vision and natural language domains, evaluating generative models for dynamic graphs is challenging due to the difficulty of visualizing their output, making quantitative metrics essential. In this work, we develop a new quality metric for evaluating generative models of dynamic graphs. Current metrics for dynamic graphs typically involve discretizing the continuous-evolution of graphs into static snapshots and then applying conventional graph similarity measures. This approach has several limitations: (a) it models temporally related events as i.i.d. samples, failing to capture the non-uniform evolution of dynamic graphs; (b) it lacks a unified measure that is sensitive to both features and topology; (c) it fails to provide a scalar metric, requiring multiple metrics without clear superiority; and (d) it requires explicitly instantiating each static snapshot, leading to impractical runtime demands that hinder evaluation at scale. We propose a novel metric based on the \textit{Johnson-Lindenstrauss} lemma, applying random projections directly to dynamic graph data. This results in an expressive, scalar, and application-agnostic measure of dynamic graph similarity that overcomes the limitations of traditional methods. We also provide a comprehensive empirical evaluation of metrics for continuous-time dynamic graphs, demonstrating the effectiveness of our approach compared to existing methods. Our implementation is available at https://github.com/ryienh/jl-metric.

arXiv:2502.17384

On the Dichotomy Between Privacy and Traceability in $\ell_p$ Stochastic Convex Optimization

In this paper, we investigate the necessity of memorization in stochastic convex optimization (SCO) under $\ell_p$ geometries. Informally, we say a learning algorithm memorizes $m$ samples (or is $m$-traceable) if, by analyzing its output, it is possible to identify at least $m$ of its training samples. Our main results uncover a fundamental tradeoff between traceability and excess risk in SCO. For every $p\in [1,\infty)$, we establish the existence of a risk threshold below which any sample-efficient learner must memorize a \em{constant fraction} of its sample. For $p\in [1,2]$, this threshold coincides with best risk of differentially private (DP) algorithms, i.e., above this threshold, there are algorithms that do not memorize even a single sample. This establishes a sharp dichotomy between privacy and traceability for $p \in [1,2]$. For $p \in (2,\infty)$, this threshold instead gives novel lower bounds for DP learning, partially closing an open problem in this setup. En route of proving these results, we introduce a complexity notion we term \em{trace value} of a problem, which unifies privacy lower bounds and traceability results, and prove a sparse variant of the fingerprinting lemma.

arXiv:2502.15955

Compression Barriers for Autoregressive Transformers

A key limitation of autoregressive Transformers is the large memory needed at inference-time to cache all previous key-value (KV) embeddings. Prior works address this by compressing the KV cache, but often assume specific structural properties of the embeddings. This raises the following natural question: Can truly sublinear space utilization be achieved without such assumptions? In this work, we answer this question in the negative. Any algorithm for attention-based token generation must use $\Theta(nd)$ space, where $n$ is the number of tokens generated so far and $d = \Omega(\log n)$ is the dimension of the KV embeddings. Our proof involves a reduction from a classic communication complexity problem and uses a randomized construction that leverages properties of projections in the spirit of the Johnson-Linderstrauss lemma. For the low-dimensional regime $d = o(\log n)$, we show that any algorithm requires $\Omega(d\cdot e^d)$ space and prove, using tight bounds on covering numbers, that SubGen, proposed by Zandieh, Han, Mirrokni and Karbasi, matches this bound. Further, we investigate how sparsity assumptions enable token generation in truly sublinear space, presenting impossibility results and proposing a new KV cache compression algorithm for sliding window attention when the value cache outside the window is unmasked. Finally, we analyze token generation's time complexity, using an indistinguishability argument to prove that no non-adaptive algorithm can compute attention online in sublinear time for all tokens.

arXiv:2502.10301

Residualised Treatment Intensity and the Estimation of Average Partial Effects

This paper introduces R-OLS, an estimator for the average partial effect (APE) of a continuous treatment variable on an outcome variable in the presence of non-linear and non-additively separable confounding of unknown form. Identification of the APE is achieved by generalising Stein's Lemma (Stein, 1981), leveraging an exogenous error component in the treatment along with a flexible functional relationship between the treatment and the confounders. The identification results for R-OLS are used to characterize the properties of Double/Debiased Machine Learning (Chernozhukov et al., 2018), specifying the conditions under which the APE is estimated consistently. A novel decomposition of the ordinary least squares estimand provides intuition for these results. Monte Carlo simulations demonstrate that the proposed estimator outperforms existing methods, delivering accurate estimates of the true APE and exhibiting robustness to moderate violations of its underlying assumptions. The methodology is further illustrated through an empirical application to Fetzer (2019).

arXiv:2502.10581

Do We Need to Verify Step by Step? Rethinking Process Supervision from a Theoretical Perspective

As large language models have evolved, it has become crucial to distinguish between process supervision and outcome supervision -- two key reinforcement learning approaches to complex reasoning tasks. While process supervision offers intuitive advantages for long-term credit assignment, the precise relationship between these paradigms has remained an open question. Conventional wisdom suggests that outcome supervision is fundamentally more challenging due to the trajectory-level coverage problem, leading to significant investment in collecting fine-grained process supervision data. In this paper, we take steps towards resolving this debate. Our main theorem shows that, under standard data coverage assumptions, reinforcement learning through outcome supervision is no more statistically difficult than through process supervision, up to polynomial factors in horizon. At the core of this result lies the novel Change of Trajectory Measure Lemma -- a technical tool that bridges return-based trajectory measure and step-level distribution shift. Furthermore, for settings with access to a verifier or a rollout capability, we prove that any policy's advantage function can serve as an optimal process reward model, providing a direct connection between outcome and process supervision. These findings suggest that the empirically observed performance gap -- if any -- between outcome and process supervision likely stems from algorithmic limitations rather than inherent statistical difficulties, potentially transforming how we approach data collection and algorithm design for reinforcement learning.

arXiv:2502.09531

Data-Enabled Predictive Control for Flexible Spacecraft

Spacecraft are vital to space exploration and are often equipped with lightweight, flexible appendages to meet strict weight constraints. These appendages pose significant challenges for modeling and control due to their inherent nonlinearity. Data-driven control methods have gained traction to address such challenges. This paper introduces, to the best of the authors' knowledge, the first application of the data-enabled predictive control (DeePC) framework to boundary control for flexible spacecraft. Leveraging the fundamental lemma, DeePC constructs a non-parametric model by utilizing recorded past trajectories, eliminating the need for explicit model development. The developed method also incorporates dimension reduction techniques to enhance computational efficiency. Through comprehensive numerical simulations, this study compares the proposed method with Lyapunov-based control, demonstrating superior performance and offering a thorough evaluation of data-driven control for flexible spacecraft.