[论文解读] Cognitive Model Priors for Predicting Human Decisions
本文通过在认知模型(BEAST变体)生成的合成数据上对神经网络进行预训练并在真实数据上微调,达到对不确定性下人类选择的最前沿预测,并引入一个包含大量选择项的13k数据集。
Human decision-making underlies all economic behavior. For the past four decades, human decision-making under uncertainty has continued to be explained by theoretical models based on prospect theory, a framework that was awarded the Nobel Prize in Economic Sciences. However, theoretical models of this kind have developed slowly, and robust, high-precision predictive models of human decisions remain a challenge. While machine learning is a natural candidate for solving these problems, it is currently unclear to what extent it can improve predictions obtained by current theories. We argue that this is mainly due to data scarcity, since noisy human behavior requires massive sample sizes to be accurately captured by off-the-shelf machine learning methods. To solve this problem, what is needed are machine learning models with appropriate inductive biases for capturing human behavior, and larger datasets. We offer two contributions towards this end: first, we construct "cognitive model priors" by pretraining neural networks with synthetic data generated by cognitive models (i.e., theoretical models developed by cognitive psychologists). We find that fine-tuning these networks on small datasets of real human decisions results in unprecedented state-of-the-art improvements on two benchmark datasets. Second, we present the first large-scale dataset for human decision-making, containing over 240,000 human judgments across over 13,000 decision problems. This dataset reveals the circumstances where cognitive model priors are useful, and provides a new standard for benchmarking prediction of human decisions under uncertainty.
研究动机与目标
- 推动在不确定性下对人类决策预测的改进,超越传统理论。
- 通过将认知理论转化为神经网络的合成训练数据来发展认知模型先验。
- 在 CPC15/CPC18 基准上展示通过对预训练网络在真实数据上微调获得的改进预测。
- 创建大规模公共数据集(synth15、synth18、choices13k)以评测对人类决策的预测。
提出的方法
- 从认知模型 BEAST15 和 BEAST18 生成合成问题集,以将神经网络训练成先验。
- 训练神经网络以在合成数据上逼近 BEAST 的预测,然后在真实的人类决策数据上进行微调。
- 使用带 SET 正则化的稀疏多层感知机(MLP),并以非常小的学习率进行微调以整合先验。
- 在 CPC15 和 CPC18 基准上对比传统机器学习模型和基于理论的基线来评估预测。
- 引入 choices13k,这是一个通过 Mechanical Turk 收集的大规模人类风险决策数据集。
实验结果
研究问题
- RQ1在真实数据稀缺时,认知模型能否提供有效的先验来引导机器学习?
- RQ2对合成认知模型数据进行预训练是否能提升对人类决策数据的预测精度和训练效率?
- RQ3相较于仅理论和纯ML方法,认知模型先验在已建立的 CPC 基准上表现如何?
- RQ4一个大型新的人类决策数据集(choices13k)在基准预测模型方面的价值和泛化能力如何?
- RQ5在训练中使用认知先验是否具有实际好处(如更快收敛、数据效率)?
主要发现
- 对在 synth15 synth18 认知先验上预训练的神经网络进行微调,在 CPC15 和 CPC18 基准上相对基线获得显著的均方误差(MSE)下降。
- CPC15:带认知先验的提升模型的 MSE 为 0.53(×100),而 CPC15 冠军为 0.88,BEAST15 为 0.99。
- CPC18:带认知先验的提升模型的 MSE 为 0.48(×100),而 CPC18 冠军为 0.57,BEAST18 为 0.70。
- 在 CPC18 上,使用 synth18 训练并微调的 MLP 达到 0.48,比 CPC18 冠军高出 0.09(×100 单位)。
- choices13k 数据集(242,879 次判断,覆盖 13,006 个问题)表明认知先验可降低预测误差并加快训练,尤其是在有限的人类数据情况下。
更好的研究,从现在开始
从阅读论文到最终审阅,大幅缩短您的研究时间。
无需绑定信用卡
本解读由 AI 生成,并经人工编辑审核。