Most automated program verifiers for separation logic use either symbolic execution or verification condition generation to extract proof obligations, which are then handed over to an SMT solver. Existing verification algorithms are designed to be sound, but differ in performance and completeness. These characteristics may also depend on the programs and properties to be verified. Consequently, developers and users of program verifiers have to select a verification algorithm carefully for their application domain. Taking an informed decision requires a systematic comparison of the performance and completeness characteristics of the verification algorithms used by modern separation logic verifiers, but such a comparison does not exist. This paper describes five verification algorithms for separation logic, three that are used in existing tools and two novel algorithms that combine characteristics of existing symbolic execution and verification condition generation algorithms. A detailed evaluation of implementations of these five algorithms in the Viper infrastructure assesses their performance and completeness for different classes of input programs. Based on the experimental results, we identify candidate portfolios of algorithms that maximize completeness and performance.
相關內容
The prevalent approaches of unsupervised 3D object detection follow cluster-based pseudo-label generation and iterative self-training processes. However, the challenge arises due to the sparsity of LiDAR scans, which leads to pseudo-labels with erroneous size and position, resulting in subpar detection performance. To tackle this problem, this paper introduces a Commonsense Prototype-based Detector, termed CPD, for unsupervised 3D object detection. CPD first constructs Commonsense Prototype (CProto) characterized by high-quality bounding box and dense points, based on commonsense intuition. Subsequently, CPD refines the low-quality pseudo-labels by leveraging the size prior from CProto. Furthermore, CPD enhances the detection accuracy of sparsely scanned objects by the geometric knowledge from CProto. CPD outperforms state-of-the-art unsupervised 3D detectors on Waymo Open Dataset (WOD), PandaSet, and KITTI datasets by a large margin. Besides, by training CPD on WOD and testing on KITTI, CPD attains 90.85% and 81.01% 3D Average Precision on easy and moderate car classes, respectively. These achievements position CPD in close proximity to fully supervised detectors, highlighting the significance of our method. The code will be available at //github.com/hailanyi/CPD.
In task-oriented conversational AI evaluation, unsupervised methods poorly correlate with human judgments, and supervised approaches lack generalization. Recent advances in large language models (LLMs) show robust zeroshot and few-shot capabilities across NLP tasks. This paper explores using LLMs for automated dialogue quality evaluation, experimenting with various configurations on public and proprietary datasets. Manipulating factors such as model size, in-context examples, and selection techniques, we examine "chain-of-thought" (CoT) reasoning and label extraction procedures. Our results show that (1) larger models yield more accurate dialogue labels; (2) algorithmic selection of in-context examples outperforms random selection; (3) CoT reasoning where an LLM is asked to provide justifications before outputting final labels improves performance; and (4) fine-tuned LLMs outperform out-of-the-box ones. Our results indicate that LLMs that are suitably fine-tuned and have sufficient reasoning capabilities can be leveraged for automated dialogue evaluation.
Database systems are often confronted with queries that join many tables but ultimately only output comparatively small aggregate information. Despite all advances in query optimisation, the explosion of intermediate results as opposed to a much smaller final result challenges modern relational database management systems (DBMSs). In this work, we propose the integration of optimisation techniques into relational DBMSs that aim at minimising, and often entirely eliminating, the need for materialising join results for aggregate queries, provided that they satisfy certain conditions. Apart from novel logical optimisations aimed at practicability, we also provide new, natural, physical operators for combining joins and counting with the aim of reducing the size of intermediate results. We experimentally validate the efficacy of our optimisations through their implementation in Spark SQL, but we note that they are naturally applicable in any RDBMS. Our experiments show consistent significant speed-ups -- often by factor 2 and higher -- for analytical and graph queries. At the same time, we observe no performance degradation, even on queries which, from a theoretical point of view, are least amenable to the proposed optimisations.
As an alternative to Kripke models, simplicial complexes are a versatile semantic primitive on which to interpret epistemic logic. Given a set of vertices, a simplicial complex is a downward closed set of subsets, called simplexes, of the vertex set. A maximal simplex is called a facet. Impure simplicial complexes represent that some agents (processes) are dead. It is known that impure simplicial complexes categorically correspond to so-called partial epistemic (Kripke) models. In this contribution, we define a notion of bisimulation to compare impure simplicial complexes and show that it has the Hennessy-Milner property. These results are for a logical language including atoms that express whether agents are alive or dead. Without these atoms no reasonable standard notion of bisimulation exists, as we amply justify by counterexamples, because such a restricted language is insufficiently expressive.
The interest in linear complexity models for large language models is on the rise, although their scaling capacity remains uncertain. In this study, we present the scaling laws for linear complexity language models to establish a foundation for their scalability. Specifically, we examine the scaling behaviors of three efficient linear architectures. These include TNL, a linear attention model with data-independent decay; HGRN2, a linear RNN with data-dependent decay; and cosFormer2, a linear attention model without decay. We also include LLaMA as a baseline architecture for softmax attention for comparison. These models were trained with six variants, ranging from 70M to 7B parameters on a 300B-token corpus, and evaluated with a total of 1,376 intermediate checkpoints on various downstream tasks. These tasks include validation loss, commonsense reasoning, and information retrieval and generation. The study reveals that existing linear complexity language models exhibit similar scaling capabilities as conventional transformer-based models while also demonstrating superior linguistic proficiency and knowledge retention.
This work proposes a framework of benchmark functions designed to facilitate the creation of test cases for numerical optimisation techniques. The framework, written in Python 3, is designed to be easy to install, use, and expand. The collection includes some of the most used multi-modal continuous functions present in literature, which can be instantiated using an arbitrary number of dimensions. Meta-information of each benchmark function, like search boundaries and position of known optima, are included and made easily accessible through class methods. Built-in interactive visualisation capabilities, baseline techniques, and rigorous testing protocols complement the features of the framework. The framework can be found here: \url{//gitlab.com/luca.baronti/python_benchmark_functions
The problem-solving performance of many evolutionary algorithms, including genetic programming systems used for program synthesis, depends on the values of hyperparameters including mutation rates. The mutation method used to produce some of the best results to date on software synthesis benchmark problems, Uniform Mutation by Addition and Deletion (UMAD), adds new genes into a genome at a predetermined rate and then deletes genes at a rate that balances the addition rate, producing no size change on average. While UMAD with a predetermined addition rate outperforms many other mutation and crossover schemes, we do not expect a single rate to be optimal across all problems or all generations within one run of an evolutionary system. However, many current adaptive mutation schemes such as self-adaptive mutation rates suffer from pathologies like the vanishing mutation rate problem, in which the mutation rate quickly decays to zero. We propose an adaptive bandit-based scheme that addresses this problem and essentially removes the need to specify a mutation rate. Although the proposed scheme itself introduces hyperparameters, we either set these to good values or ensemble them in a reasonable range. Results on software synthesis and symbolic regression problems validate the effectiveness of our approach.
Causal discovery aims to uncover cause-and-effect relationships encoded in causal graphs by leveraging observational, interventional data, or their combination. The majority of existing causal discovery methods are developed assuming infinite interventional data. We focus on data interventional efficiency and formalize causal discovery from the perspective of online learning, inspired by pure exploration in bandit problems. A graph separating system, consisting of interventions that cut every edge of the graph at least once, is sufficient for learning causal graphs when infinite interventional data is available, even in the worst case. We propose a track-and-stop causal discovery algorithm that adaptively selects interventions from the graph separating system via allocation matching and learns the causal graph based on sampling history. Given any desired confidence value, the algorithm determines a termination condition and runs until it is met. We analyze the algorithm to establish a problem-dependent upper bound on the expected number of required interventional samples. Our proposed algorithm outperforms existing methods in simulations across various randomly generated causal graphs. It achieves higher accuracy, measured by the structural hamming distance (SHD) between the learned causal graph and the ground truth, with significantly fewer samples.
Recent contrastive representation learning methods rely on estimating mutual information (MI) between multiple views of an underlying context. E.g., we can derive multiple views of a given image by applying data augmentation, or we can split a sequence into views comprising the past and future of some step in the sequence. Contrastive lower bounds on MI are easy to optimize, but have a strong underestimation bias when estimating large amounts of MI. We propose decomposing the full MI estimation problem into a sum of smaller estimation problems by splitting one of the views into progressively more informed subviews and by applying the chain rule on MI between the decomposed views. This expression contains a sum of unconditional and conditional MI terms, each measuring modest chunks of the total MI, which facilitates approximation via contrastive bounds. To maximize the sum, we formulate a contrastive lower bound on the conditional MI which can be approximated efficiently. We refer to our general approach as Decomposed Estimation of Mutual Information (DEMI). We show that DEMI can capture a larger amount of MI than standard non-decomposed contrastive bounds in a synthetic setting, and learns better representations in a vision domain and for dialogue generation.
Benefit from the quick development of deep learning techniques, salient object detection has achieved remarkable progresses recently. However, there still exists following two major challenges that hinder its application in embedded devices, low resolution output and heavy model weight. To this end, this paper presents an accurate yet compact deep network for efficient salient object detection. More specifically, given a coarse saliency prediction in the deepest layer, we first employ residual learning to learn side-output residual features for saliency refinement, which can be achieved with very limited convolutional parameters while keep accuracy. Secondly, we further propose reverse attention to guide such side-output residual learning in a top-down manner. By erasing the current predicted salient regions from side-output features, the network can eventually explore the missing object parts and details which results in high resolution and accuracy. Experiments on six benchmark datasets demonstrate that the proposed approach compares favorably against state-of-the-art methods, and with advantages in terms of simplicity, efficiency (45 FPS) and model size (81 MB).
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