Bayesian optimization (BO) is widely used to optimize expensive-to-evaluate black-box functions.BO first builds a surrogate model to represent the objective function and assesses its uncertainty. It then decides where to sample by maximizing an acquisition function (AF) based on the surrogate model. However, when dealing with high-dimensional problems, finding the global maximum of the AF becomes increasingly challenging. In such cases, the initialization of the AF maximizer plays a pivotal role, as an inadequate setup can severely hinder the effectiveness of the AF. This paper investigates a largely understudied problem concerning the impact of AF maximizer initialization on exploiting AFs' capability. Our large-scale empirical study shows that the widely used random initialization strategy often fails to harness the potential of an AF. In light of this, we propose a better initialization approach by employing multiple heuristic optimizers to leverage the historical data of black-box optimization to generate initial points for the AF maximize. We evaluate our approach with a range of heavily studied synthetic functions and real-world applications. Experimental results show that our techniques, while simple, can significantly enhance the standard BO and outperform state-of-the-art methods by a large margin in most test cases.
相關內容
Large Language Models (LLMs) have emerged as a powerful tool in advancing the Text-to-SQL task, significantly outperforming traditional methods. Nevertheless, as a nascent research field, there is still no consensus on the optimal prompt templates and design frameworks. Additionally, existing benchmarks inadequately explore the performance of LLMs across the various sub-tasks of the Text-to-SQL process, which hinders the assessment of LLMs' cognitive capabilities and the optimization of LLM-based solutions. To address the aforementioned issues, we firstly construct a new dataset designed to mitigate the risk of overfitting in LLMs. Then we formulate five evaluation tasks to comprehensively assess the performance of diverse methods across various LLMs throughout the Text-to-SQL process.Our study highlights the performance disparities among LLMs and proposes optimal in-context learning solutions tailored to each task. These findings offer valuable insights for enhancing the development of LLM-based Text-to-SQL systems.
The semiparametric factor model serves as a vital tool to describe the dependence patterns in the data. It recognizes that the common features observed in the data are actually explained by functions of specific exogenous variables.Unlike traditional factor models, where the focus is on selecting the number of factors, our objective here is to identify the appropriate number of common functions, a crucial parameter in this model. In this paper, we develop a novel data-driven method to determine the number of functional factors using cross validation (CV). Our proposed method employs a two-step CV process that ensures the orthogonality of functional factors, which we refer to as Functional Twice Cross-Validation (FTCV). Extensive simulations demonstrate that FTCV accurately selects the number of common functions and outperforms existing methods in most cases.Furthermore, by specifying market volatility as the exogenous force, we provide real data examples that illustrate the interpretability of selected common functions in characterizing the influence on U.S. Treasury Yields and the cross correlations between Dow30 returns.
Information retrieval models have witnessed a paradigm shift from unsupervised statistical approaches to feature-based supervised approaches to completely data-driven ones that make use of the pre-training of large language models. While the increasing complexity of the search models have been able to demonstrate improvements in effectiveness (measured in terms of relevance of top-retrieved results), a question worthy of a thorough inspection is - "how explainable are these models?", which is what this paper aims to evaluate. In particular, we propose a common evaluation platform to systematically evaluate the explainability of any ranking model (the explanation algorithm being identical for all the models that are to be evaluated). In our proposed framework, each model, in addition to returning a ranked list of documents, also requires to return a list of explanation units or rationales for each document. This meta-information from each document is then used to measure how locally consistent these rationales are as an intrinsic measure of interpretability - one that does not require manual relevance assessments. Additionally, as an extrinsic measure, we compute how relevant these rationales are by leveraging sub-document level relevance assessments. Our findings show a number of interesting observations, such as sentence-level rationales are more consistent, an increase in complexity mostly leads to less consistent explanations, and that interpretability measures offer a complementary dimension of evaluation of IR systems because consistency is not well-correlated with nDCG at top ranks.
It is a longstanding conjecture that every simple drawing of a complete graph on $n \geq 3$ vertices contains a crossing-free Hamiltonian cycle. We strengthen this conjecture to "there exists a crossing-free Hamiltonian path between each pair of vertices" and show that this stronger conjecture holds for several classes of simple drawings, including strongly c-monotone drawings and cylindrical drawings. As a second main contribution, we give an overview on different classes of simple drawings and investigate inclusion relations between them up to weak isomorphism.
Artificial Intelligence (AI) is increasingly employed in various decision-making tasks, typically as a Recommender, providing recommendations that the AI deems correct. However, recent studies suggest this may diminish human analytical thinking and lead to humans' inappropriate reliance on AI, impairing the synergy in human-AI teams. In contrast, human advisors in group decision-making perform various roles, such as analyzing alternative options or criticizing decision-makers to encourage their critical thinking. This diversity of roles has not yet been empirically explored in AI assistance. In this paper, we examine three AI roles: Recommender, Analyzer, and Devil's Advocate, and evaluate their effects across two AI performance levels. Our results show each role's distinct strengths and limitations in task performance, reliance appropriateness, and user experience. Notably, the Recommender role is not always the most effective, especially if the AI performance level is low, the Analyzer role may be preferable. These insights offer valuable implications for designing AI assistants with adaptive functional roles according to different situations.
Recently, large language models (LLMs) have notably positioned them as capable tools for addressing complex optimization challenges. Despite this recognition, a predominant limitation of existing LLM-based optimization methods is their struggle to capture the relationships among decision variables when relying exclusively on numerical text prompts, especially in high-dimensional problems. Keeping this in mind, we first propose to enhance the optimization performance using multimodal LLM capable of processing both textual and visual prompts for deeper insights of the processed optimization problem. This integration allows for a more comprehensive understanding of optimization problems, akin to human cognitive processes. We have developed a multimodal LLM-based optimization framework that simulates human problem-solving workflows, thereby offering a more nuanced and effective analysis. The efficacy of this method is evaluated through extensive empirical studies focused on a well-known combinatorial optimization problem, i.e., capacitated vehicle routing problem. The results are compared against those obtained from the LLM-based optimization algorithms that rely solely on textual prompts, demonstrating the significant advantages of our multimodal approach.
The curse-of-dimensionality taxes computational resources heavily with exponentially increasing computational cost as the dimension increases. This poses great challenges in solving high-dimensional PDEs, as Richard E. Bellman first pointed out over 60 years ago. While there has been some recent success in solving numerically partial differential equations (PDEs) in high dimensions, such computations are prohibitively expensive, and true scaling of general nonlinear PDEs to high dimensions has never been achieved. We develop a new method of scaling up physics-informed neural networks (PINNs) to solve arbitrary high-dimensional PDEs. The new method, called Stochastic Dimension Gradient Descent (SDGD), decomposes a gradient of PDEs into pieces corresponding to different dimensions and randomly samples a subset of these dimensional pieces in each iteration of training PINNs. We prove theoretically the convergence and other desired properties of the proposed method. We demonstrate in various diverse tests that the proposed method can solve many notoriously hard high-dimensional PDEs, including the Hamilton-Jacobi-Bellman (HJB) and the Schr\"{o}dinger equations in tens of thousands of dimensions very fast on a single GPU using the PINNs mesh-free approach. Notably, we solve nonlinear PDEs with nontrivial, anisotropic, and inseparable solutions in 100,000 effective dimensions in 12 hours on a single GPU using SDGD with PINNs. Since SDGD is a general training methodology of PINNs, it can be applied to any current and future variants of PINNs to scale them up for arbitrary high-dimensional PDEs.
Likelihood-based deep generative models such as score-based diffusion models and variational autoencoders are state-of-the-art machine learning models approximating high-dimensional distributions of data such as images, text, or audio. One of many downstream tasks they can be naturally applied to is out-of-distribution (OOD) detection. However, seminal work by Nalisnick et al. which we reproduce showed that deep generative models consistently infer higher log-likelihoods for OOD data than data they were trained on, marking an open problem. In this work, we analyse using the gradient of a data point with respect to the parameters of the deep generative model for OOD detection, based on the simple intuition that OOD data should have larger gradient norms than training data. We formalise measuring the size of the gradient as approximating the Fisher information metric. We show that the Fisher information matrix (FIM) has large absolute diagonal values, motivating the use of chi-square distributed, layer-wise gradient norms as features. We combine these features to make a simple, model-agnostic and hyperparameter-free method for OOD detection which estimates the joint density of the layer-wise gradient norms for a given data point. We find that these layer-wise gradient norms are weakly correlated, rendering their combined usage informative, and prove that the layer-wise gradient norms satisfy the principle of (data representation) invariance. Our empirical results indicate that this method outperforms the Typicality test for most deep generative models and image dataset pairings.
While subword tokenizers such as BPE and WordPiece are typically used to build vocabularies for NLP models, the method of decoding text into a sequence of tokens from these vocabularies is often left unspecified, or ill-suited to the method in which they were constructed. We provide a controlled analysis of seven tokenizer inference methods across four different algorithms and three vocabulary sizes, performed on a novel intrinsic evaluation suite we curated for English, combining measures rooted in morphology, cognition, and information theory. We show that for the most commonly used tokenizers, greedy inference performs surprisingly well; and that SaGe, a recently-introduced contextually-informed tokenizer, outperforms all others on morphological alignment.
The design space of current quantum computers is expansive with no obvious winning solution. This leaves practitioners with a clear question: "What is the optimal system configuration to run an algorithm?". This paper explores hardware design trade-offs across NISQ systems to guide algorithm and hardware design choices. The evaluation is driven by algorithmic workloads and algorithm fidelity models which capture architectural features such as gate expressivity, fidelity, and crosstalk. We also argue that the criteria for gate design and selection should be extended from maximizing average fidelity to a more comprehensive approach that takes into account the gate expressivity with respect to algorithmic structures. We consider native entangling gates (CNOT, ECR, CZ, ZZ, XX, Sycamore, $\sqrt{\text{iSWAP}}$), proposed gates (B Gate, $\sqrt[4]{\text{CNOT}}$, $\sqrt[8]{\text{CNOT}}$), as well as parameterized gates (FSim, XY). Our methodology is driven by a custom synthesis driven circuit compilation workflow, which is able to produce minimal circuit representations for a given system configuration. By providing a method to evaluate the suitability of algorithms for hardware platforms, this work emphasizes the importance of hardware-software co-design for quantum computing.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191