2022年32篇最佳AI论文：DALL·E 2、Stable Diffusion、ChatGPT等入选

素颜马尾好姑娘i 2023-09-25 14:56 33阅读 0赞

Mila在读博士[Louis Bouchard][]总结的论文列表，总体比较靠谱。[GitHub][]上还有很多论文的短视频和文字解读、代码链接等。

下面的列表我们添加了论文的主要贡献机构（有些机构虽然有贡献但排名较后有挂名嫌疑的，都被忽略不计了），似乎可以反映出各公司在AI领域的江湖地位：

*  第一档：Google 8篇，Meta 6篇雄踞前二名，OpenAI 3篇但有两篇影响力巨大的（DALL·E 2和ChatGPT），如果按代表作评价，可能不会输给两巨头。
 *  第二档：NVIDIA有2.5篇。
 *  第三档：国内腾讯、百度、微软（出自亚研院）各1篇。国外有三星、迪士尼各1篇。Snap、Adobe都是0.5篇。

高校总共5.5篇，不如两巨头一家，相比之下要逊色很多。其中：

*  特拉维夫有1.5篇位居第一，但慕尼黑的Stable Diffusion影响巨大，应该视为第一档。
 *  CMU、南洋理工各1篇，第二档。
 *  南加大和伯克利各0.5篇，第三档。

从方向来看，**大模型和文生图、跨模态**是今年毫无疑问的热点，此外也有多篇GAN等**视觉**领域的文章。

> \[1\] 三星: Suvorov, R., Logacheva, E., Mashikhin, A., Remizova, A., Ashukha, A., Silvestrov, A., Kong, N., Goka, H., Park, K. and Lempitsky, V., 2022. **Resolution-robust Large Mask Inpainting with Fourier Convolutions.** In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (pp. 2149–2159)., [https://arxiv.org/pdf/2109.07161.pdf][https_arxiv.org_pdf_2109.07161.pdf]
> 
> \[2\] 特拉维夫: Tzaban, R., Mokady, R., Gal, R., Bermano, A.H. and Cohen-Or, D., 2022. **Stitch it in Time: GAN-Based Facial Editing of Real Videos.** [https://arxiv.org/abs/2201.08361][https_arxiv.org_abs_2201.08361]
> 
> \[3\] 南加大&Snap: Kuang, Z., Olszewski, K., Chai, M., Huang, Z., Achlioptas, P. and Tulyakov, S., 2022. **NeROIC: Neural Rendering of Objects from Online Image Collections.** [https://arxiv.org/pdf/2201.02533.pdf][https_arxiv.org_pdf_2201.02533.pdf]
> 
> \[4\] Google: Borsos, Z., Sharifi, M. and Tagliasacchi, M., 2022. **SpeechPainter: Text-conditioned Speech Inpainting.** [https://arxiv.org/pdf/2202.07273.pdf][https_arxiv.org_pdf_2202.07273.pdf]
> 
> \[5\] 腾讯: Wang, X., Li, Y., Zhang, H. and Shan, Y., 2021. **Towards real-world blind face restoration with generative facial prior.** In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 9168–9178), [https://arxiv.org/pdf/2101.04061.pdf][https_arxiv.org_pdf_2101.04061.pdf]
> 
> \[6\] Google: Piergiovanni, A.J., Casser, V., Ryoo, M.S. and Angelova, A., 2021. **4D-Net for learned multi-modal alignment.** In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 15435–15445), [https://openaccess.thecvf.com/content/ICCV2021/papers/Piergiovanni\_4D-Net\_for\_Learned\_Multi-Modal\_Alignment\_ICCV\_2021\_paper.pdf][https_openaccess.thecvf.com_content_ICCV2021_papers_Piergiovanni_4D-Net_for_Learned_Multi-Modal_Alignment_ICCV_2021_paper.pdf].
> 
> \[7\] NVIDIA: Thomas Muller, Alex Evans, Christoph Schied and Alexander Keller, 2022, **Instant Neural Graphics Primitives with a Multiresolution Hash Encoding**, [https://nvlabs.github.io/instant-ngp/assets/mueller2022instant.pdf][https_nvlabs.github.io_instant-ngp_assets_mueller2022instant.pdf]
> 
> \[8\] OpenAI/DALL·E 2: Ramesh et al., 2022, **Hierarchical Text-Conditional Image Generation with CLIP Latents**, [https://cdn.openai.com/papers/dall-e-2.pdf][https_cdn.openai.com_papers_dall-e-2.pdf]
> 
> \[9\] Google: Nitzan, Y., Aberman, K., He, Q., Liba, O., Yarom, M., Gandelsman, Y., Mosseri, I., Pritch, Y. and Cohen-Or, D., 2022. **MyStyle: A Personalized Generative Prior.** arXiv preprint [arXiv:2203.17272][arXiv_2203.17272].
> 
> \[10\] Meta/OPT: Zhang, Susan et al. **OPT: Open Pre-trained Transformer Language Models.** [https://arxiv.org/abs/2205.01068][https_arxiv.org_abs_2205.01068]
> 
> \[11\] 伯克利&Adobe: Epstein, D., Park, T., Zhang, R., Shechtman, E. and Efros, A.A., 2022. **BlobGAN: Spatially Disentangled Scene Representations**. arXiv preprint arXiv:2205.02837.
> 
> \[12\] Google DeepMind: Reed S. et al., 2022. **Gato - A generalist agent**, [A Generalist Agent][]
> 
> \[13\] Google/Imagen: Saharia et al., 2022. **Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding.** [Imagen: Text-to-Image Diffusion Models][Imagen_ Text-to-Image Diffusion Models]
> 
> \[14\] Craiyon: Dayma, et al., 2021, **DALL·E Mini**, doi:10.5281/zenodo.5146400. [GitHub][GitHub 1] （DALL·E的复现，只有一些技术报告，未找到正规论文）
> 
> \[15\] Meta: NLLB Team et al., 2022, **No Language Left Behind: Scaling Human-Centered Machine Translation.** [https://arxiv.org/abs/2207.04672][https_arxiv.org_abs_2207.04672]
> 
> \[16\] CMU: Sheinin, Mark and Chan, Dorian and O’Toole, Matthew and Narasimhan, Srinivasa G., 2022, **Dual-Shutter Optical Vibration Sensing**, Proc. IEEE CVPR. [Dual-Shutter Optical Vibration Sensing][] （CVPR2022最佳论文入围）
> 
> \[17\] Meta: Gafni, O., Polyak, A., Ashual, O., Sheynin, S., Parikh, D. and Taigman, Y., 2022. **Make-a-scene: Scene-based text-to-image generation with human priors.** [https://arxiv.org/pdf/2203.13131.pdf][https_arxiv.org_pdf_2203.13131.pdf]
> 
> \[18\] Meta: Yang, G., Vo, M., Neverova, N., Ramanan, D., Vedaldi, A. and Joo, H., 2022. **Banmo: Building animatable 3d neural models from many casual videos.** In CVPR2022 (pp. 2863-2873). [https://arxiv.org/abs/2112.12761][https_arxiv.org_abs_2112.12761]
> 
> \[19\] 慕尼黑/Stable Diffusion: Rombach, R., Blattmann, A., Lorenz, D., Esser, P. and Ommer, B., 2022. **High-resolution image synthesis with latent diffusion models.** In CVPR2022 (pp. 10684–10695), [https://arxiv.org/pdf/2112.10752.pdf][https_arxiv.org_pdf_2112.10752.pdf]
> 
> \[20\] 南洋理工: Yang, J., Ang, Y.Z., Guo, Z., Zhou, K., Zhang, W. and Liu, Z., 2022. **Panoptic Scene Graph Generation.** arXiv preprint [arXiv:2207.11247][arXiv_2207.11247].
> 
> \[21\] 特拉维夫&NVIDIA: Gal, R., Alaluf, Y., Atzmon, Y., Patashnik, O., Bermano, A.H., Chechik, G. and Cohen-Or, D., 2022. **An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversion.**  [An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversion][An Image is Worth One Word_ Personalizing Text-to-Image Generation using Textual Inversion]
> 
> \[22\] 微软: Ni, B., Peng, H., Chen, M., Zhang, S., Meng, G., Fu, J., Xiang, S. and Ling, H., 2022. **Expanding Language-Image Pretrained Models for General Video Recognition.** arXiv preprint [arXiv:2208.02816][arXiv_2208.02816].
> 
> \[23\] Meta/Make-A-Video: Singer et al., 2022. **Make-A-Video: Text-To-Video Generation without Text-Video Data**, [https://makeavideo.studio/Make-A-Video.pdf][https_makeavideo.studio_Make-A-Video.pdf]
> 
> \[24\] OpenAI/Whisper: Radford, A., Kim, J.W., Xu, T., Brockman, G., McLeavey, C. and Sutskever, I., **Robust Speech Recognition via Large-Scale Weak Supervision.** [GitHub][GitHub 2]
> 
> \[25\] Google: Poole, B., Jain, A., Barron, J.T. and Mildenhall, B., 2022. **DreamFusion: Text-to-3D using 2D Diffusion.** arXiv preprint arXiv:2209.14988. [DreamFusion: Text-to-3D using 2D Diffusion][DreamFusion_ Text-to-3D using 2D Diffusion]
> 
> \[26\] Google/Imagic: Kawar, B., Zada, S., Lang, O., Tov, O., Chang, H., Dekel, T., Mosseri, I. and Irani, M., 2022. **Imagic: Text-Based Real Image Editing with Diffusion Models.** arXiv preprint [arXiv:2210.09276][arXiv_2210.09276].
> 
> \[27\] NVIDIA: Balaji, Y. et al., 2022, **eDiffi: Text-to-Image Diffusion Models with an Ensemble of Expert Denoisers,** [https://arxiv.org/abs/2211.01324][https_arxiv.org_abs_2211.01324]
> 
> \[28\] Google: Li, Z., Wang, Q., Snavely, N. and Kanazawa, A., 2022. **InfiniteNature-Zero: Learning Perpetual View Generation of Natural Scenes from Single Images.** In ECCV (pp. 515–534). Springer, Cham, [https://arxiv.org/abs/2207.11148][https_arxiv.org_abs_2207.11148]
> 
> \[29\] Meta/Galactica: Taylor et al., 2022: **Galactica: A Large Language Model for Science**, [Galactica Demo][]
> 
> \[30\] 百度: Tang, J., Wang, K., Zhou, H., Chen, X., He, D., Hu, T., Liu, J., Zeng, G. and Wang, J., 2022. **Real-time Neural Radiance Talking Portrait Synthesis via Audio-spatial Decomposition.** arXiv preprint [arXiv:2211.12368][arXiv_2211.12368].
> 
> \[31\] OpenAI/ChatGPT: **ChatGPT: Optimizing Language Models for Dialogue**, [ChatGPT: Optimizing Language Models for Dialogue][ChatGPT_ Optimizing Language Models for Dialogue]
> 
> \[32\] 迪士尼/FRAN: Loss et al., DisneyResearch, 2022: **Production-Ready Face Re-Aging for Visual Effects**, [https://studios.disneyresearch.com][https_studios.disneyresearch.com]

这个列表里的论文哪些你认同，哪些你觉得不应该入选最佳的，还有哪些重要论文遗漏？欢迎大家评论。

[吴恩达、Bengio等6位AI大佬2023年度展望 - 智源社区][Bengio_6_AI_2023_ -]

[2022年32篇最佳AI论文：DALL·E 2、Stable Diffusion、ChatGPT等入选 - 智源社区][2022_32_AI_DALL_E 2_Stable Diffusion_ChatGPT_ -]

[2022年最值得关注的十篇论文（威斯康星大学助理教授推荐） - 智源社区][2022_ -]

[CMU博士总结的2022年Top20 AI论文 - 智源社区][CMU_2022_Top20 AI_ -]

[2023-01-28智源社区日报：OpenAI为代码大模型标注、谷歌MusicLM生成复杂音乐、Meta文本生成4D场景、伯克利讲解DL基础到大模型、「画匠」生成图片红包活动 - 智源社区][2023-01-28_OpenAI_MusicLM_Meta_4D_DL_ -]

[Louis Bouchard]: https://www.louisbouchard.ai/about/
[GitHub]: https://github.com/louisfb01/best_AI_papers_2022
[https_arxiv.org_pdf_2109.07161.pdf]: https://arxiv.org/pdf/2109.07161.pdf
[https_arxiv.org_abs_2201.08361]: https://arxiv.org/abs/2201.08361
[https_arxiv.org_pdf_2201.02533.pdf]: https://arxiv.org/pdf/2201.02533.pdf
[https_arxiv.org_pdf_2202.07273.pdf]: https://arxiv.org/pdf/2202.07273.pdf
[https_arxiv.org_pdf_2101.04061.pdf]: https://arxiv.org/pdf/2101.04061.pdf
[https_openaccess.thecvf.com_content_ICCV2021_papers_Piergiovanni_4D-Net_for_Learned_Multi-Modal_Alignment_ICCV_2021_paper.pdf]: https://openaccess.thecvf.com/content/ICCV2021/papers/Piergiovanni_4D-Net_for_Learned_Multi-Modal_Alignment_ICCV_2021_paper.pdf
[https_nvlabs.github.io_instant-ngp_assets_mueller2022instant.pdf]: https://nvlabs.github.io/instant-ngp/assets/mueller2022instant.pdf
[https_cdn.openai.com_papers_dall-e-2.pdf]: https://cdn.openai.com/papers/dall-e-2.pdf
[arXiv_2203.17272]: https://arxiv.org/abs/2203.17272
[https_arxiv.org_abs_2205.01068]: https://arxiv.org/abs/2205.01068
[A Generalist Agent]: https://www.deepmind.com/publications/a-generalist-agent
[Imagen_ Text-to-Image Diffusion Models]: https://imagen.research.google/
[GitHub 1]: https://github.com/borisdayma/dalle-mini
[https_arxiv.org_abs_2207.04672]: https://arxiv.org/abs/2207.04672
[Dual-Shutter Optical Vibration Sensing]: https://imaging.cs.cmu.edu/vibration/
[https_arxiv.org_pdf_2203.13131.pdf]: https://arxiv.org/pdf/2203.13131.pdf
[https_arxiv.org_abs_2112.12761]: https://arxiv.org/abs/2112.12761
[https_arxiv.org_pdf_2112.10752.pdf]: https://arxiv.org/pdf/2112.10752.pdf
[arXiv_2207.11247]: https://arxiv.org/abs/2207.11247
[An Image is Worth One Word_ Personalizing Text-to-Image Generation using Textual Inversion]: https://textual-inversion.github.io/
[arXiv_2208.02816]: https://arxiv.org/abs/2208.02816
[https_makeavideo.studio_Make-A-Video.pdf]: https://makeavideo.studio/Make-A-Video.pdf
[GitHub 2]: https://github.com/openai/whisper
[DreamFusion_ Text-to-3D using 2D Diffusion]: https://dreamfusion3d.github.io/
[arXiv_2210.09276]: https://arxiv.org/abs/2210.09276
[https_arxiv.org_abs_2211.01324]: https://arxiv.org/abs/2211.01324
[https_arxiv.org_abs_2207.11148]: https://arxiv.org/abs/2207.11148
[Galactica Demo]: https://galactica.org/
[arXiv_2211.12368]: https://arxiv.org/abs/2211.12368
[ChatGPT_ Optimizing Language Models for Dialogue]: https://openai.com/blog/chatgpt/
[https_studios.disneyresearch.com]: https://studios.disneyresearch.com/2022/11/30/production-ready-face-re-aging-for-visual-effects/
[Bengio_6_AI_2023_ -]: https://hub.baai.ac.cn/view/22916
[2022_32_AI_DALL_E 2_Stable Diffusion_ChatGPT_ -]: https://hub.baai.ac.cn/view/22798
[2022_ -]: https://hub.baai.ac.cn/view/23205
[CMU_2022_Top20 AI_ -]: https://hub.baai.ac.cn/view/22983
[2023-01-28_OpenAI_MusicLM_Meta_4D_DL_ -]: https://hub.baai.ac.cn/view/23638