現今在眾多遞歸神經網路的相關研究中，通常都在探討如何生成出高品質的序列，很少有研究在探討如何生成出多樣化的序列，本篇論文的目的在於提出一種新的模型方法來讓遞歸神經網路可以生成出高品質且多樣化的序列。我們延伸現在最新的生成對抗模型技術，在生成端加入強弱合作的概念，藉由強弱互相合作來生成出兼具品質與多樣的序列。我們更進一步的延伸我們的模型，讓模型自己與自己的缺點合作，達到自我進步的學習，最終，我們的模型有辦法生成出不僅僅是多樣化的序列，還可以生成出屬於自己特色的序列，也就是它可以自己學習出「創造力」。從我們設計的實驗中，可以證明我們提出的強弱合作的概念可以有效提升序列生成的表現。
本篇論文的貢獻包含以下幾點：
1. 我們的研究是第一個能夠讓遞歸神經網路可以生成出高品質且多樣化的序列
2. 我們提出一種新的方法來衡量序列的品質與多樣性
3. 我們在生成對抗網路的領域上開起了一條新的研究方向，讓大家可以去思考如何在對應的任務中，在生成端加入合作的概念來增進模型的表現

Current RNN decoding methods usually focus on how to generate high quality sequences, but they ignore the importance of variety on the collection of outputs. Our work introduces a new model architecture to let RNN generate high quality and variety sequences. We extend the Generative Adversarial Networks and propose Strong-Weak Collaborative GAN. We separate the generator into two part, strong and weak, to cooperatively generate a sequence to cheat discriminator. To further improve our model, we make our model to improve itself, namely Self-improving Collaborative GAN (SIC-GAN). SIC-GAN can generate not only high quality and variety sequences, but also to produce “creative” outputs. Experimental result shows that our model can generate higher quality and more diverse results than all the baseline.

致謝 2
摘要 3
Abstract 4
1. Introduction 6
2. Related Work 8
3. Main Idea: Strong-Weak Collaborative GAN 9
3.1 Generative Adversarial Networks 9
3.2 Strong-Weak Collaborative GAN 9
4. SICGAN: Self-Improving Collaborative GAN 11
4.1 Inspiration: AlphaGo 11
4.2 Modification 11
5. Experiment 13
5.1 Dataset 13
5.2 Training Details 13
5.3 Baseline 14
5.4 Evaluation 14
6. Conclusion & Future Work 18
Reference 19

[1] Mart´ın Arjovsky, Soumith Chintala, and L´eon Bottou. Wasserstein generative adversarial networks. In ICML, volume 70 of Proceedings of Machine Learning Research, pp. 214–223. PMLR, 2017.
[2] Samy Bengio, Oriol Vinyals, Navdeep Jaitly, and Noam Shazeer. Scheduled sampling for sequence prediction with recurrent neural networks. In NIPS, pp. 1171–1179, 2015.
[3] Tong Che, Yanran Li, Ruixiang Zhang, R. Devon Hjelm, Wenjie Li, Yangqiu Song, and Yoshua Bengio. Maximum-likelihood augmented discrete generative adversarial networks. CoRR, abs/1702.07983, 2017.
[4] Kyunghyun Cho. Noisy parallel approximate decoding for conditional recurrent language model. CoRR, abs/1605.03835, 2016.
[5] Ian J. Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron C. Courville, and Yoshua Bengio. Generative adversarial nets. In NIPS, pp. 2672–2680, 2014.
[6] Anirudh Goyal, Alex Lamb, Ying Zhang, Saizheng Zhang, Aaron C. Courville, and Yoshua Bengio. Professor forcing: A new algorithm for training recurrent networks. In NIPS, pp. 4601–4609, 2016.
[7] Jiatao Gu, Kyunghyun Cho, and Victor O. K. Li. Trainable greedy decoding for neural machine translation. In EMNLP, pp. 1958–1968. Association for Computational Linguistics, 2017a. Jiatao Gu, Daniel Jiwoong Im, and Victor O. K. Li. Neural machine translation with gumbel-greedy decoding. CoRR, abs/1706.07518, 2017b.
[8] Ishaan Gulrajani, Faruk Ahmed, Mart´ın Arjovsky, Vincent Dumoulin, and Aaron C. Courville. Improved training of wasserstein gans. CoRR, abs/1704.00028, 2017.
[9] David Ha and Douglas Eck. A neural representation of sketch drawings. CoRR, abs/1704.03477, 2017.
[10] J. Jongejan, H. Rowley, T. Kawashima, J. Kim, and N. Fox-Gieg. The quick, draw! - a.i. experiment. https://quickdraw.withgoogle.com/, 2016.
[11] Yoon Kim. Convolutional neural networks for sentence classification. In EMNLP, pp. 1746–1751. ACL, 2014.
[12] Matt J. Kusner and Jos´e Miguel Hern´andez-Lobato. GANS for sequences of discrete elements with the gumbel-softmax distribution. CoRR, abs/1611.04051, 2016.
[13] Jiwei Li, Will Monroe, and Dan Jurafsky. A simple, fast diverse decoding algorithm for neural generation. CoRR, abs/1611.08562, 2016.
[14] Alec Radford, Luke Metz, and Soumith Chintala. Unsupervised representation learning with deep convolutional generative adversarial networks. CoRR, abs/1511.06434, 2015.
[15] David Silver, Aja Huang, Chris J. Maddison, Arthur Guez, Laurent Sifre, George van den Driessche, Julian Schrittwieser, Ioannis Antonoglou, Vedavyas Panneershelvam, Marc Lanctot, Sander Dieleman, Dominik Grewe, John Nham, Nal Kalchbrenner, Ilya Sutskever, Timothy P. Lillicrap, Madeleine Leach, Koray Kavukcuoglu, Thore Graepel, and Demis Hassabis. Mastering the game of go with deep neural networks and tree search. Nature, 529(7587):484–489, 2016.
[16] Lantao Yu, Weinan Zhang, Jun Wang, and Yong Yu. Seqgan: Sequence generative adversarial nets with policy gradient. In AAAI, pp. 2852–2858. AAAI Press, 2017.
[17] Yizhe Zhang, Zhe Gan, and Lawrence Carin. Generating text via adversarial training. 2016.