Interactive systems that facilitate exposure to examples can augment problem solving performance. However designers of such systems are often faced with many practical design decisions about how users will interact with examples, with little clear theoretical guidance. To understand how example interaction design choices affect whether/how people benefit from examples, we conducted an experiment where 182 participants worked on a controlled analog to an exploratory creativity task, with access to examples of varying diversity and presentation interfaces. Task performance was worse when examples were presented in a list, compared to contextualized in the exploration space or shown in a dropdown list. Example lists were associated with more fixation, whereas contextualized examples were associated with using examples to formulate a model of the problem space to guide exploration. We discuss implications of these results for a theoretical framework that maps design choices to fundamental psychological mechanisms of creative inspiration from examples.
相關內容
We present AlloyInEcore, a tool for specifying metamodels with their static semantics to facilitate automated, formal reasoning on models. Software development projects require that software systems be specified in various models (e.g., requirements models, architecture models, test models, and source code). It is crucial to reason about those models to ensure the correct and complete system specifications. AlloyInEcore allows the user to specify metamodels with their static semantics, while, using the semantics, it automatically detects inconsistent models, and completes partial models. It has been evaluated on three industrial case studies in the automotive domain (//modelwriter.github.io/AlloyInEcore/).
We present SeaEval, a benchmark for multilingual foundation models. In addition to characterizing how these models understand and reason with natural language, we also investigate how well they comprehend cultural practices, nuances, and values. Alongside standard accuracy metrics, we investigate the brittleness of foundation models in the dimensions of semantics and multilinguality. Our analyses span both open-sourced and closed models, leading to empirical results across classic NLP tasks, reasoning, and cultural comprehension. Key findings indicate (1) Most models exhibit varied behavior when given paraphrased instructions. (2) Many models still suffer from exposure bias (e.g., positional bias, majority label bias). (3) For questions rooted in factual, scientific, and commonsense knowledge, consistent responses are expected across multilingual queries that are semantically equivalent. Yet, most models surprisingly demonstrate inconsistent performance on these queries. (4) Multilingually-trained models have not attained "balanced multilingual" capabilities. Our endeavors underscore the need for more generalizable semantic representations and enhanced multilingual contextualization. SeaEval can serve as a launchpad for more thorough investigations and evaluations for multilingual and multicultural scenarios.
We prove that training neural networks on 1-D data is equivalent to solving a convex Lasso problem with a fixed, explicitly defined dictionary matrix of features. The specific dictionary depends on the activation and depth. We consider 2-layer networks with piecewise linear activations, deep narrow ReLU networks with up to 4 layers, and rectangular and tree networks with sign activation and arbitrary depth. Interestingly in ReLU networks, a fourth layer creates features that represent reflections of training data about themselves. The Lasso representation sheds insight to globally optimal networks and the solution landscape.
Deployed multimodal systems can fail in ways that evaluators did not anticipate. In order to find these failures before deployment, we introduce MultiMon, a system that automatically identifies systematic failures -- generalizable, natural-language descriptions of patterns of model failures. To uncover systematic failures, MultiMon scrapes a corpus for examples of erroneous agreement: inputs that produce the same output, but should not. It then prompts a language model (e.g., GPT-4) to find systematic patterns of failure and describe them in natural language. We use MultiMon to find 14 systematic failures (e.g., "ignores quantifiers") of the CLIP text-encoder, each comprising hundreds of distinct inputs (e.g., "a shelf with a few/many books"). Because CLIP is the backbone for most state-of-the-art multimodal systems, these inputs produce failures in Midjourney 5.1, DALL-E, VideoFusion, and others. MultiMon can also steer towards failures relevant to specific use cases, such as self-driving cars. We see MultiMon as a step towards evaluation that autonomously explores the long tail of potential system failures. Code for MULTIMON is available at //github.com/tsb0601/MultiMon.
Despite the importance of trust in human-AI interactions, researchers must adopt questionnaires from other disciplines that lack validation in the AI context. Motivated by the need for reliable and valid measures, we investigated the psychometric quality of two trust questionnaires, the Trust between People and Automation scale (TPA) by Jian et al. (2000) and the Trust Scale for the AI Context (TAI) by Hoffman et al. (2023). In a pre-registered online experiment (N = 1485), participants observed interactions with trustworthy and untrustworthy AI (autonomous vehicle and chatbot). Results support the psychometric quality of the TAI while revealing opportunities to improve the TPA, which we outline in our recommendations for using the two questionnaires. Furthermore, our findings provide additional empirical evidence of trust and distrust as two distinct constructs that may coexist independently. Building on our findings, we highlight the opportunities and added value of measuring both trust and distrust in human-AI research and advocate for further work on both constructs.
Probing the memorization of large language models holds significant importance. Previous works have established metrics for quantifying memorization, explored various influencing factors, such as data duplication, model size, and prompt length, and evaluated memorization by comparing model outputs with training corpora. However, the training corpora are of enormous scale and its pre-processing is time-consuming. To explore memorization without accessing training data, we propose a novel approach, named ROME, wherein memorization is explored by comparing disparities across memorized and non-memorized. Specifically, models firstly categorize the selected samples into memorized and non-memorized groups, and then comparing the demonstrations in the two groups from the insights of text, probability, and hidden state. Experimental findings show the disparities in factors including word length, part-of-speech, word frequency, mean and variance, just to name a few.
Prompt engineering in LLMs has shown potential for improving translation quality. However, the potential of incorporating translation concepts in prompt design remains largely underexplored. Against this backdrop, this paper discusses the effectiveness of incorporating the conceptual tool of translation brief and the personas of translator and author into prompt design for translation tasks in ChatGPT. Findings suggest that, although certain elements are constructive in facilitating human to human communication for translation tasks, their effectiveness is limited for improving translation quality in ChatGPT. This accentuates the need for more explorative research on how translation theorists and practitioners can develop the current set of conceptual tools rooted in the human to human communication paradigm for translation purposes in this emerging workflow involving human machine interaction.
Skill mastery is a priority for success in all fields. We present a parallel between the development of skill mastery and the process of solving jigsaw puzzles. We show that iterative random sampling solves jigsaw puzzles in two phases: a lag phase that is characterized by little change and occupies the majority of the time, and a growth phase that marks rapid and imminent puzzle completion. Changes in the proportions of the number of single pieces and larger pieces can be overlaid on the timeline and progression of skill mastery. An emphasis is placed on the development of connections between pieces, which serves as an indicator of increasing puzzle completion and increasing skill mastery. Our manuscript provides a straightforward visual of skill mastery in the context of a common recreational activity.
Current deep learning research is dominated by benchmark evaluation. A method is regarded as favorable if it empirically performs well on the dedicated test set. This mentality is seamlessly reflected in the resurfacing area of continual learning, where consecutively arriving sets of benchmark data are investigated. The core challenge is framed as protecting previously acquired representations from being catastrophically forgotten due to the iterative parameter updates. However, comparison of individual methods is nevertheless treated in isolation from real world application and typically judged by monitoring accumulated test set performance. The closed world assumption remains predominant. It is assumed that during deployment a model is guaranteed to encounter data that stems from the same distribution as used for training. This poses a massive challenge as neural networks are well known to provide overconfident false predictions on unknown instances and break down in the face of corrupted data. In this work we argue that notable lessons from open set recognition, the identification of statistically deviating data outside of the observed dataset, and the adjacent field of active learning, where data is incrementally queried such that the expected performance gain is maximized, are frequently overlooked in the deep learning era. Based on these forgotten lessons, we propose a consolidated view to bridge continual learning, active learning and open set recognition in deep neural networks. Our results show that this not only benefits each individual paradigm, but highlights the natural synergies in a common framework. We empirically demonstrate improvements when alleviating catastrophic forgetting, querying data in active learning, selecting task orders, while exhibiting robust open world application where previously proposed methods fail.
This work considers the question of how convenient access to copious data impacts our ability to learn causal effects and relations. In what ways is learning causality in the era of big data different from -- or the same as -- the traditional one? To answer this question, this survey provides a comprehensive and structured review of both traditional and frontier methods in learning causality and relations along with the connections between causality and machine learning. This work points out on a case-by-case basis how big data facilitates, complicates, or motivates each approach.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191