A numerical search approach is used to design high-order diagonally implicit Runge-Kutta (DIRK) schemes equipped with embedded error estimators, some of which have identical diagonal elements (SDIRK) and explicit first stage (ESDIRK). In each of these classes, we present new A-stable schemes of order six (the highest order of previously known A-stable DIRK-type schemes) up to order eight. For each order, we include one scheme that is only A-stable as well as schemes that are L-stable, stiffly accurate, and/or have stage order two. The latter types require more stages, but give better convergence rates for differential-algebraic equations (DAEs), and those which have stage order two give better accuracy for moderately stiff problems. The development of the eighth-order schemes requires, in addition to imposing A-stability, finding highly accurate numerical solutions for a system of 200 equations in over 100 variables, which is accomplished via a combination of global and local optimization strategies. The accuracy, stability, and adaptive stepsize control of the schemes are demonstrated on diverse problems.
相關內容
Dynamic multi-objective optimisation (DMO) handles optimisation problems with multiple (often conflicting) objectives in varying environments. Such problems pose various challenges to evolutionary algorithms, which have popularly been used to solve complex optimisation problems, due to their dynamic nature and resource restrictions in changing environments. This paper proposes vector autoregressive evolution (VARE) consisting of vector autoregression (VAR) and environment-aware hypermutation to address environmental changes in DMO. VARE builds a VAR model that considers mutual relationship between decision variables to effectively predict the moving solutions in dynamic environments. Additionally, VARE introduces EAH to address the blindness of existing hypermutation strategies in increasing population diversity in dynamic scenarios where predictive approaches are unsuitable. A seamless integration of VAR and EAH in an environment-adaptive manner makes VARE effective to handle a wide range of dynamic environments and competitive with several popular DMO algorithms, as demonstrated in extensive experimental studies. Specially, the proposed algorithm is computationally 50 times faster than two widely-used algorithms (i.e., TrDMOEA and MOEA/D-SVR) while producing significantly better results.
A simple yet effective way of modeling survival data with cure fraction is by considering Box-Cox transformation cure model (BCTM) that unifies mixture and promotion time cure models. In this article, we numerically study the statistical properties of the BCTM when applied to interval censored data. Time-to-events associated with susceptible subjects are modeled through proportional hazards structure that allows for non-homogeneity across subjects, where the baseline hazard function is estimated by distribution-free piecewise linear function with varied degrees of non-parametricity. Due to missing cured statuses for right censored subjects, maximum likelihood estimates of model parameters are obtained by developing an expectation-maximization (EM) algorithm. Under the EM framework, the conditional expectation of the complete data log-likelihood function is maximized by considering all parameters (including the Box-Cox transformation parameter $\alpha$) simultaneously, in contrast to conventional profile-likelihood technique of estimating $\alpha$. The robustness and accuracy of the model and estimation method are established through a detailed simulation study under various parameter settings, and an analysis of real-life data obtained from a smoking cessation study.
Multi-label text classification is a critical task in the industry. It helps to extract structured information from large amount of textual data. We propose Text to Topic (Text2Topic), which achieves high multi-label classification performance by employing a Bi-Encoder Transformer architecture that utilizes concatenation, subtraction, and multiplication of embeddings on both text and topic. Text2Topic also supports zero-shot predictions, produces domain-specific text embeddings, and enables production-scale batch-inference with high throughput. The final model achieves accurate and comprehensive results compared to state-of-the-art baselines, including large language models (LLMs). In this study, a total of 239 topics are defined, and around 1.6 million text-topic pairs annotations (in which 200K are positive) are collected on approximately 120K texts from 3 main data sources on Booking.com. The data is collected with optimized smart sampling and partial labeling. The final Text2Topic model is deployed on a real-world stream processing platform, and it outperforms other models with 92.9% micro mAP, as well as a 75.8% macro mAP score. We summarize the modeling choices which are extensively tested through ablation studies, and share detailed in-production decision-making steps.
An automated resource analysis technique is introduced, targeting a Call-By-Push-Value abstract machine, with memory prediction as a practical goal. The machine has a polymorphic and linear type system enhanced with a first-order logical fragment, which encodes both low-level operational semantics of resource manipulations and high-level synthesis of algorithmic complexity. Resource analysis must involve a diversity of static analysis, for escape, aliasing, algorithmic invariants, and more. Knowing this, we implement the Automated Amortized Resource Analysis framework (AARA) from scratch in our generic system. In this setting, access to resources is a state-passing effect which produces a compile-time approximation of runtime resource usage. We implemented type inference constraint generation for our calculus, accompanied with an elaboration of bounds for iterators on algebraic datatypes, for minimal ML-style programming languages with Call-by-Value and Call-By-Push-Value semantics. The closed-formed bounds are derived as multivariate polynomials over the integers. This now serves as a base for the development of an experimental toolkit for automated memory analysis of functional languages.
This paper shows a novel machine learning model for realized volatility (RV) prediction using a normalizing flow, an invertible neural network. Since RV is known to be skewed and have a fat tail, previous methods transform RV into values that follow a latent distribution with an explicit shape and then apply a prediction model. However, knowing that shape is non-trivial, and the transformation result influences the prediction model. This paper proposes to jointly train the transformation and the prediction model. The training process follows a maximum-likelihood objective function that is derived from the assumption that the prediction residuals on the transformed RV time series are homogeneously Gaussian. The objective function is further approximated using an expectation-maximum algorithm. On a dataset of 100 stocks, our method significantly outperforms other methods using analytical or naive neural-network transformations.
We present a data structure to randomly sample rows from the Khatri-Rao product of several matrices according to the exact distribution of its leverage scores. Our proposed sampler draws each row in time logarithmic in the height of the Khatri-Rao product and quadratic in its column count, with persistent space overhead at most the size of the input matrices. As a result, it tractably draws samples even when the matrices forming the Khatri-Rao product have tens of millions of rows each. When used to sketch the linear least squares problems arising in CANDECOMP / PARAFAC tensor decomposition, our method achieves lower asymptotic complexity per solve than recent state-of-the-art methods. Experiments on billion-scale sparse tensors validate our claims, with our algorithm achieving higher accuracy than competing methods as the decomposition rank grows.
Training or finetuning large-scale language models (LLMs) such as GPT-3 requires substantial computation resources, motivating recent efforts to explore parameter-efficient adaptation to downstream tasks. One practical area of research is to treat these models as black boxes and interact with them through their inference APIs. In this paper, we investigate how to optimize few-shot text classification without accessing the gradients of the LLMs. To achieve this, we treat the black-box model as a feature extractor and train a classifier with the augmented text data. Data augmentation is performed using prompt-based finetuning on an auxiliary language model with a much smaller parameter size than the black-box model. Through extensive experiments on eight text classification datasets, we show that our approach, dubbed BT-Classifier, significantly outperforms state-of-the-art black-box few-shot learners and performs on par with methods that rely on full-model tuning.
Cross-Domain Recommendation (CDR) stands as a pivotal technology addressing issues of data sparsity and cold start by transferring general knowledge from the source to the target domain. However, existing CDR models suffer limitations in adaptability across various scenarios due to their inherent complexity. To tackle this challenge, recent advancements introduce universal CDR models that leverage shared embeddings to capture general knowledge across domains and transfer it through "Multi-task Learning" or "Pre-train, Fine-tune" paradigms. However, these models often overlook the broader structural topology that spans domains and fail to align training objectives, potentially leading to negative transfer. To address these issues, we propose a motif-based prompt learning framework, MOP, which introduces motif-based shared embeddings to encapsulate generalized domain knowledge, catering to both intra-domain and inter-domain CDR tasks. Specifically, we devise three typical motifs: butterfly, triangle, and random walk, and encode them through a Motif-based Encoder to obtain motif-based shared embeddings. Moreover, we train MOP under the "Pre-training \& Prompt Tuning" paradigm. By unifying pre-training and recommendation tasks as a common motif-based similarity learning task and integrating adaptable prompt parameters to guide the model in downstream recommendation tasks, MOP excels in transferring domain knowledge effectively. Experimental results on four distinct CDR tasks demonstrate the effectiveness of MOP than the state-of-the-art models.
Multi-Agent Path Finding (MAPF) is a fundamental problem in robotics and AI, with numerous applications in real-world scenarios. One such scenario is filming scenes with multiple actors, where the goal is to capture the scene from multiple angles simultaneously. Here, we present a formation-based filming directive of task assignment followed by a Conflict-Based MAPF algorithm for efficient path planning of multiple agents to achieve filming objectives while avoiding collisions. We propose an extension to the standard MAPF formulation to accommodate actor-specific requirements and constraints. Our approach incorporates Conflict-Based Search, a widely used heuristic search technique for solving MAPF problems. We demonstrate the effectiveness of our approach through experiments on various MAPF scenarios in a simulated environment. The proposed algorithm enables the efficient online task assignment of formation-based filming to capture dynamic scenes, making it suitable for various filming and coverage applications.
Pre-trained Language Models (PLMs) which are trained on large text corpus via self-supervised learning method, have yielded promising performance on various tasks in Natural Language Processing (NLP). However, though PLMs with huge parameters can effectively possess rich knowledge learned from massive training text and benefit downstream tasks at the fine-tuning stage, they still have some limitations such as poor reasoning ability due to the lack of external knowledge. Research has been dedicated to incorporating knowledge into PLMs to tackle these issues. In this paper, we present a comprehensive review of Knowledge-Enhanced Pre-trained Language Models (KE-PLMs) to provide a clear insight into this thriving field. We introduce appropriate taxonomies respectively for Natural Language Understanding (NLU) and Natural Language Generation (NLG) to highlight these two main tasks of NLP. For NLU, we divide the types of knowledge into four categories: linguistic knowledge, text knowledge, knowledge graph (KG), and rule knowledge. The KE-PLMs for NLG are categorized into KG-based and retrieval-based methods. Finally, we point out some promising future directions of KE-PLMs.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191