arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.01867cs.CLcs.AIcs.CV
收藏

基于双重强化的图像去渲染规范生成

Dual Reinforcement-Based Specification Generation for Image De-Rendering

UNC Chapel Hill(北卡罗来纳大学教堂山分校);Bloomberg LP(彭博有限合伙企业)

Ramakanth Pasunuru, David Rosenberg, Gideon Mann, Mohit Bansal

摘要:

深度学习的进步已经在通过去渲染计算机生成的图像来推断图形程序方面取得了有希望的进展。然而,目前的方法并没有探索哪些解码方法导致更好的归纳偏差,用于推断图形程序。在我们的工作中,我们首先探索了LSTM-RNN与Transformer网络作为顺序无关图形程序解码器的有效性。由于这些是序列模型,我们必须在图形程序中选择对象的顺序进行似然训练。我们发现,LSTM的性能对序列排序(随机顺序与基于模式的顺序)高度敏感,而Transformer的性能大致与序列排序无关。此外,我们提出了一种基于策略梯度的强化学习方法,通过基于图形程序规范和渲染图像的多个不同奖励,在解码器中获得更好的归纳偏差。我们还探索了这些互补奖励的组合。我们在两个图形程序生成数据集上实现了最先进的结果。

英文摘要:

Advances in deep learning have led to promising progress in inferring graphics programs by de-rendering computer-generated images. However, current methods do not explore which decoding methods lead to better inductive bias for inferring graphics programs. In our work, we first explore the effectiveness of LSTM-RNN versus Transformer networks as decoders for order-independent graphics programs. Since these are sequence models, we must choose an ordering of the objects in the graphics programs for likelihood training. We found that the LSTM performance was highly sensitive to the sequence ordering (random order vs. pattern-based order), while Transformer performance was roughly independent of the sequence ordering. Further, we present a policy gradient based reinforcement learning approach for better inductive bias in the decoder via multiple diverse rewards based both on the graphics program specification and the rendered image. We also explore the combination of these complementary rewards. We achieve state-of-the-art results on two graphics program generation datasets.

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。

↑