arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1904.09533cs.LGcs.SDeess.ASstat.ML
收藏

基于GAN的神经网络模型生成与自动选择

GAN-based Generation and Automatic Selection of Explanations for Neural Networks

Queen Mary University of London(伦敦玛丽女王大学);The Alan Turing Institute(艾伦·图灵研究所);KTH Royal Institute of Technology(KTH皇家理工学院)

Saumitra Mishra, Daniel Stoller, Emmanouil Benetos, Bob L. Sturm, Simon Dixon

摘要:

解释训练的深度神经网络(DNN)的一种方式是通过检查模型中的神经元响应的特性,诸如通过迭代地优化模型输入(例如,图像)以最大限度地激活特定的神经元。然而,这需要仔细选择超参数来为每个感兴趣的神经元生成可解释的示例,并且目前的方法依赖于对每个设置的手动定性评估,这非常缓慢。我们引入了一个新的度量标准,使用Fréchet Inception Distance(FID)来鼓励真实数据和生成数据的模型激活之间的相似性。这提供了一种有效的方式来评估针对每个超参数设置的一组生成的示例。我们还提出了一种新的基于GAN的方法来生成解释,该方法可以在输入空间中进行有效的搜索,并施加一个有利于现实输出的强先验。我们将我们的方法应用于训练的分类模型,以预测音乐音频记录是否包含歌声。我们的研究结果表明,该指标成功地选择了超参数,导致可解释的例子,避免了手动评估的需要。此外,我们看到,合成的例子,以最大限度地提高或最小化的预测概率唱歌的声音存在表现出声乐或非声乐的特点,分别,这表明我们的方法是能够产生合适的解释,理解概念学习的神经网络。

英文摘要:

One way to interpret trained deep neural networks (DNNs) is by inspecting characteristics that neurons in the model respond to, such as by iteratively optimising the model input (e.g., an image) to maximally activate specific neurons. However, this requires a careful selection of hyper-parameters to generate interpretable examples for each neuron of interest, and current methods rely on a manual, qualitative evaluation of each setting, which is prohibitively slow. We introduce a new metric that uses Fréchet Inception Distance (FID) to encourage similarity between model activations for real and generated data. This provides an efficient way to evaluate a set of generated examples for each setting of hyper-parameters. We also propose a novel GAN-based method for generating explanations that enables an efficient search through the input space and imposes a strong prior favouring realistic outputs. We apply our approach to a classification model trained to predict whether a music audio recording contains singing voice. Our results suggest that this proposed metric successfully selects hyper-parameters leading to interpretable examples, avoiding the need for manual evaluation. Moreover, we see that examples synthesised to maximise or minimise the predicted probability of singing voice presence exhibit vocal or non-vocal characteristics, respectively, suggesting that our approach is able to generate suitable explanations for understanding concepts learned by a neural network.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑