Recent advances in vision-language learning have achieved notable success on
complete-information question-answering datasets through the integration of
extensive world knowledge. Yet, most models operate passively, responding to
questions based on pre-stored knowledge. In stark contrast, humans possess the
ability to actively explore, accumulate, and reason using both newfound and
existing information to tackle incomplete-information questions. In response to
this gap, we introduce $Conan$, an interactive open-world environment devised
for the assessment of active reasoning. $Conan$ facilitates active exploration
and promotes multi-round abductive inference, reminiscent of rich, open-world
settings like Minecraft. Diverging from previous works that lean primarily on
single-round deduction via instruction following, $Conan$ compels agents to
actively interact with their surroundings, amalgamating new evidence with prior
knowledge to elucidate events from incomplete observations. Our analysis on
$Conan$ underscores the shortcomings of contemporary state-of-the-art models in
active exploration and understanding complex scenarios. Additionally, we
explore Abduction from Deduction, where agents harness Bayesian rules to recast
the challenge of abduction as a deductive process. Through $Conan$, we aim to
galvanize advancements in active reasoning and set the stage for the next
generation of artificial intelligence agents adept at dynamically engaging in
environments.

最近，在视觉 - 语言学习方面取得了显著的成功，通过整合广泛的世界知识在完整信息的问答数据集上。然而，大多数模型是被动操作的，基于预先存储的知识来回答问题。与此相反，人类具有积极探索、积累和推理的能力，可以利用新发现和已有信息来解决不完整信息的问题。为了弥补这一差距，我们介绍了 $Conan$，一个用于主动推理评估的互动式开放世界环境，类似于 Minecraft 的丰富的开放世界设置。与之前主要依靠单一回合的演绎式遵循说明的工作不同，$Conan$ 强迫代理与周围环境积极交互，将新证据与先前知识相结合，阐明从不完整观察中的事件。我们对 $Conan$ 的分析强调了现有先进模型在积极探索和理解复杂场景方面的不足。此外，我们探索了从演绎到演绎的推理，代理利用贝叶斯规则重新构建了推断挑战，将其作为演绎过程。通过 $Conan$，我们旨在推动主动推理的进步，为下一代能够动态参与环境的人工智能代理铺平道路。

开放环境中的主动推理

Active Reasoning in an Open-World Environment

Conversational agents are usually designed for closed-world environments.
Unfortunately, users can behave unexpectedly. Based on the open-world
environment, we often encounter the situation that the training and test data
are sampled from different distributions. Then, data from different
distributions are called out-of-domain (OOD). A robust conversational agent
needs to react to these OOD utterances adequately. Thus, the importance of
robust OOD detection is emphasized. Unfortunately, collecting OOD data is a
challenging task. We have designed an OOD detection algorithm independent of
OOD data that outperforms a wide range of current state-of-the-art algorithms
on publicly available datasets. Our algorithm is based on a simple but
efficient approach of combining metric learning with adaptive decision
boundary. Furthermore, compared to other algorithms, we have found that our
proposed algorithm has significantly improved OOD performance in a scenario
with a lower number of classes while preserving the accuracy for in-domain
(IND) classes.

设计了一种基于度量学习和自适应决策边界的 OOD 检测算法，相较于其他算法，在类别数量较少的情况下显著提高 OOD 性能，保持了 IND 类别的准确性。