深層驗證者網絡:深層判別模型與深層生成模型的驗證（multimedia）

• 2020 年 2 月 15 日
• 筆記

人工智能安全是許多深度學習應用程序(如自動駕駛)的主要關注點。在一個經過訓練的深度學習模型中，一個重要的自然問題是如何可靠地驗證模型的預測。在本文中，我們提出了一個新的框架-深度驗證網絡(DVN)來檢測不可靠的輸入或預測的深度判別模型，使用單獨訓練的深度生成模型。該模型基於帶解糾纏約束的條件變分自動編碼器，將標籤信息從潛在表象中分離出來。我們對該模型給出了直觀和理論上的證明。利用預測模型對驗證者網絡進行獨立訓練，消除了對驗證者網絡重新訓練的需要。我們測試了驗證器網絡工作，包括非分佈檢測和反例檢測問題，以及結構化的預處理任務中的異常檢測問題，如圖像字幕生成。我們在所有這些問題上取得了最先進的成果。

原文題目：Deep Verifier Networks: Verification of Deep Discriminative Models with Deep Generative Models

原文：AI Safety is a major concern in many deep learn- ing applications such as autonomous driving. Given a trained deep learning model, an important natural problem is how to reliably verify the model』s prediction. In this paper, we propose a novel framework — deep verifier networks (DVN) to detect unreliable inputs or predictions of deep discriminative models, using separately trained deep generative models. Our proposed model is based on conditional variational auto-encoders with disentanglement constraints to separate the label information from the latent representation. We give both intuitive and theoretical justifications for the model. Our verifier network is trained independently with the prediction model, which eliminates the need of retraining the verifier net- work for a new model. We test the verifier net- work on both out-of-distribution detection and adversarial example detection problems, as well as anomaly detection problems in structured pre- diction tasks such as image caption generation. We achieve state-of-the-art results in all of these problems.

原文作者：Tong Che, Xiaofeng Liu, Site Li, Yubin Ge, Ruixiang Zhang, Caiming Xiong, Yoshua Bengio

原文鏈接：https://arxiv.org/abs/1911.07421