Natural language understanding (NLU) studies often exaggerate or underestimate the capabilities of systems, thereby limiting the reproducibility of their findings. These erroneous evaluations can be attributed to the difficulty of defining and testing NLU adequately. In this position paper, we reconsider this challenge by identifying two types of researcher degrees of freedom. We revisit Turing's original interpretation of the Turing test and indicate that an NLU test does not provide an operational definition; it merely provides inductive evidence that the test subject understands the language sufficiently well to meet stakeholder objectives. In other words, stakeholders are free to arbitrarily define NLU through their objectives. To use the test results as inductive evidence, stakeholders must carefully assess if the interpretation of test scores is valid or not. However, designing and using NLU tests involve other degrees of freedom, such as specifying target skills and defining evaluation metrics. As a result, achieving consensus among stakeholders becomes difficult. To resolve this issue, we propose a validity argument, which is a framework comprising a series of validation criteria across test components. By demonstrating that current practices in NLU studies can be associated with those criteria and organizing them into a comprehensive checklist, we prove that the validity argument can serve as a coherent guideline for designing credible test sets and facilitating scientific communication.
相關內容
Large language models (LLMs) are gaining increasing popularity in both academia and industry, owing to their unprecedented performance in various applications. As LLMs continue to play a vital role in both research and daily use, their evaluation becomes increasingly critical, not only at the task level, but also at the society level for better understanding of their potential risks. Over the past years, significant efforts have been made to examine LLMs from various perspectives. This paper presents a comprehensive review of these evaluation methods for LLMs, focusing on three key dimensions: what to evaluate, where to evaluate, and how to evaluate. Firstly, we provide an overview from the perspective of evaluation tasks, encompassing general natural language processing tasks, reasoning, medical usage, ethics, educations, natural and social sciences, agent applications, and other areas. Secondly, we answer the `where' and `how' questions by diving into the evaluation methods and benchmarks, which serve as crucial components in assessing performance of LLMs. Then, we summarize the success and failure cases of LLMs in different tasks. Finally, we shed light on several future challenges that lie ahead in LLMs evaluation. Our aim is to offer invaluable insights to researchers in the realm of LLMs evaluation, thereby aiding the development of more proficient LLMs. Our key point is that evaluation should be treated as an essential discipline to better assist the development of LLMs. We consistently maintain the related open-source materials at: //github.com/MLGroupJLU/LLM-eval-survey.
The paper explores the concept of the \emph{expectile risk measure} within the framework of the Fundamental Risk Quadrangle (FRQ) theory. According to the FRQ theory, a quadrangle comprises four stochastic functions associated with a random variable: ``error'', ``regret'', ``risk'', and ``deviation''. These functions are interconnected through a stochastic function known as the ``statistic''. Expectile is a risk measure that, similar to VaR (quantile) and CVaR (superquantile), can be employed in risk management. While quadrangles based on VaR and CVaR statistics are well-established and widely used, the paper focuses on the recently proposed quadrangles based on expectile. The aim of this paper is to rigorously examine the properties of these Expectile Quadrangles, with particular emphasis on a quadrangle that encompasses expectile as both a statistic and a measure of risk.
One common trend in recent studies of language models (LMs) is the use of standardized tests for evaluation. However, despite being the fifth most spoken language worldwide, few such evaluations have been conducted in Portuguese. This is mainly due to the lack of high-quality datasets available to the community for carrying out evaluations in Portuguese. To address this gap, we introduce the Brazilian Leading Universities Entrance eXams (BLUEX), a dataset of entrance exams from the two leading universities in Brazil: UNICAMP and USP. The dataset includes annotated metadata for evaluating the performance of NLP models on a variety of subjects. Furthermore, BLUEX includes a collection of recently administered exams that are unlikely to be included in the training data of many popular LMs as of 2023. The dataset is also annotated to indicate the position of images in each question, providing a valuable resource for advancing the state-of-the-art in multimodal language understanding and reasoning. We describe the creation and characteristics of BLUEX and establish a benchmark through experiments with state-of-the-art LMs, demonstrating its potential for advancing the state-of-the-art in natural language understanding and reasoning in Portuguese. The data and relevant code can be found at //github.com/Portuguese-Benchmark-Datasets/BLUEX
Efficient planning of scarce resources in hospitals is a challenging task for which a large variety of Operations Research and Management Science approaches have been developed since the 1950s. While efficient planning of single resources such as operating rooms, beds, or specific types of staff can already lead to enormous efficiency gains, integrated planning of several resources has been shown to hold even greater potential, and a large number of integrated planning approaches have been presented in the literature over the past decades. This paper provides the first literature review that focuses specifically on the Operations Research and Management Science literature related to integrated planning of different resources in hospitals. We collect the relevant literature and analyze it regarding different aspects such as uncertainty modeling and the use of real-life data. Several cross comparisons reveal interesting insights concerning, e.g., relations between the modeling and solution methods used and the practical implementation of the approaches developed. Moreover, we provide a high-level taxonomy for classifying different resource-focused integration approaches and point out gaps in the literature as well as promising directions for future research.
Blockchain is an emerging decentralized data collection, sharing and storage technology, which have provided abundant transparent, secure, tamper-proof, secure and robust ledger services for various real-world use cases. Recent years have witnessed notable developments of blockchain technology itself as well as blockchain-adopting applications. Most existing surveys limit the scopes on several particular issues of blockchain or applications, which are hard to depict the general picture of current giant blockchain ecosystem. In this paper, we investigate recent advances of both blockchain technology and its most active research topics in real-world applications. We first review the recent developments of consensus mechanisms and storage mechanisms in general blockchain systems. Then extensive literature is conducted on blockchain enabled IoT, edge computing, federated learning and several emerging applications including healthcare, COVID-19 pandemic, social network and supply chain, where detailed specific research topics are discussed in each. Finally, we discuss the future directions, challenges and opportunities in both academia and industry.
Automated Driving Systems (ADS) have made great achievements in recent years thanks to the efforts from both academia and industry. A typical ADS is composed of multiple modules, including sensing, perception, planning and control, which brings together the latest advances in multiple domains. Despite these achievements, safety assurance of the systems is still of great significance, since the unsafe behavior of ADS can bring catastrophic consequences and unacceptable economic and social losses. Testing is an important approach to system validation for the deployment in practice; in the context of ADS, it is extremely challenging, due to the system complexity and multidisciplinarity. There has been a great deal of literature that focuses on the testing of ADS, and a number of surveys have also emerged to summarize the technical advances. However, most of these surveys focus on the system-level testing that is performed within software simulators, and thereby ignore the distinct features of individual modules. In this paper, we provide a comprehensive survey on the existing ADS testing literature, which takes into account both module-level and system-level testing. Specifically, we make the following contributions: (1) we build a threat model that reveals the potential safety threats for each module of an ADS; (2) we survey the module-level testing techniques for ADS and highlight the technical differences affected by the properties of the modules; (3) we also survey the system-level testing techniques, but we focus on empirical studies that take a bird's-eye view on the system, the problems due to the collaborations between modules, and the gaps between ADS testing in simulators and real world; (4) we identify the challenges and opportunities in ADS testing, which facilitates the future research in this field.
The generalization mystery in deep learning is the following: Why do over-parameterized neural networks trained with gradient descent (GD) generalize well on real datasets even though they are capable of fitting random datasets of comparable size? Furthermore, from among all solutions that fit the training data, how does GD find one that generalizes well (when such a well-generalizing solution exists)? We argue that the answer to both questions lies in the interaction of the gradients of different examples during training. Intuitively, if the per-example gradients are well-aligned, that is, if they are coherent, then one may expect GD to be (algorithmically) stable, and hence generalize well. We formalize this argument with an easy to compute and interpretable metric for coherence, and show that the metric takes on very different values on real and random datasets for several common vision networks. The theory also explains a number of other phenomena in deep learning, such as why some examples are reliably learned earlier than others, why early stopping works, and why it is possible to learn from noisy labels. Moreover, since the theory provides a causal explanation of how GD finds a well-generalizing solution when one exists, it motivates a class of simple modifications to GD that attenuate memorization and improve generalization. Generalization in deep learning is an extremely broad phenomenon, and therefore, it requires an equally general explanation. We conclude with a survey of alternative lines of attack on this problem, and argue that the proposed approach is the most viable one on this basis.
This paper surveys and organizes research works in a new paradigm in natural language processing, which we dub "prompt-based learning". Unlike traditional supervised learning, which trains a model to take in an input x and predict an output y as P(y|x), prompt-based learning is based on language models that model the probability of text directly. To use these models to perform prediction tasks, the original input x is modified using a template into a textual string prompt x' that has some unfilled slots, and then the language model is used to probabilistically fill the unfilled information to obtain a final string x, from which the final output y can be derived. This framework is powerful and attractive for a number of reasons: it allows the language model to be pre-trained on massive amounts of raw text, and by defining a new prompting function the model is able to perform few-shot or even zero-shot learning, adapting to new scenarios with few or no labeled data. In this paper we introduce the basics of this promising paradigm, describe a unified set of mathematical notations that can cover a wide variety of existing work, and organize existing work along several dimensions, e.g.the choice of pre-trained models, prompts, and tuning strategies. To make the field more accessible to interested beginners, we not only make a systematic review of existing works and a highly structured typology of prompt-based concepts, but also release other resources, e.g., a website //pretrain.nlpedia.ai/ including constantly-updated survey, and paperlist.
The growing energy and performance costs of deep learning have driven the community to reduce the size of neural networks by selectively pruning components. Similarly to their biological counterparts, sparse networks generalize just as well, if not better than, the original dense networks. Sparsity can reduce the memory footprint of regular networks to fit mobile devices, as well as shorten training time for ever growing networks. In this paper, we survey prior work on sparsity in deep learning and provide an extensive tutorial of sparsification for both inference and training. We describe approaches to remove and add elements of neural networks, different training strategies to achieve model sparsity, and mechanisms to exploit sparsity in practice. Our work distills ideas from more than 300 research papers and provides guidance to practitioners who wish to utilize sparsity today, as well as to researchers whose goal is to push the frontier forward. We include the necessary background on mathematical methods in sparsification, describe phenomena such as early structure adaptation, the intricate relations between sparsity and the training process, and show techniques for achieving acceleration on real hardware. We also define a metric of pruned parameter efficiency that could serve as a baseline for comparison of different sparse networks. We close by speculating on how sparsity can improve future workloads and outline major open problems in the field.
Detection and recognition of text in natural images are two main problems in the field of computer vision that have a wide variety of applications in analysis of sports videos, autonomous driving, industrial automation, to name a few. They face common challenging problems that are factors in how text is represented and affected by several environmental conditions. The current state-of-the-art scene text detection and/or recognition methods have exploited the witnessed advancement in deep learning architectures and reported a superior accuracy on benchmark datasets when tackling multi-resolution and multi-oriented text. However, there are still several remaining challenges affecting text in the wild images that cause existing methods to underperform due to there models are not able to generalize to unseen data and the insufficient labeled data. Thus, unlike previous surveys in this field, the objectives of this survey are as follows: first, offering the reader not only a review on the recent advancement in scene text detection and recognition, but also presenting the results of conducting extensive experiments using a unified evaluation framework that assesses pre-trained models of the selected methods on challenging cases, and applies the same evaluation criteria on these techniques. Second, identifying several existing challenges for detecting or recognizing text in the wild images, namely, in-plane-rotation, multi-oriented and multi-resolution text, perspective distortion, illumination reflection, partial occlusion, complex fonts, and special characters. Finally, the paper also presents insight into the potential research directions in this field to address some of the mentioned challenges that are still encountering scene text detection and recognition techniques.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191