序列与生成式

103 种方法属于此方法族。

精选

注意力机制The attention mechanism, introduced by Bahdanau, Cho and Bengio in 2015 and refined by Luong, Pham and Manning the same year, lets a sequence decoder dynamically learn which of the 自编码器An autoencoder is an encoder-decoder neural network, popularised by Hinton and Salakhutdinov in 2006, that compresses data into a low-dimensional latent code and then reconstructs 双向循环神经网络A Bidirectional RNN, introduced by Schuster and Paliwal in 1997, processes a sequence in both forward and backward directions so that every position has access to its full surround Crossformer：用于多元时间序列预测的跨维度依赖TransformerCrossformer is a Transformer-based architecture for multivariate time series forecasting, introduced by Yunhao Zhang and Junchi Yan at ICLR 2023. Unlike earlier Transformer variant CycleGAN：具有循环一致性的非配对图像到图像翻译CycleGAN, introduced by Zhu et al. at ICCV 2017, learns to translate images between two visual domains without requiring paired training examples. It trains two generators and two DeepARDeepAR is Amazon's industrial forecasting model, introduced by Salinas, Flunkert and Gasthaus (2017; published 2020), that uses an autoregressive recurrent neural network to estima

Reading path

This topic's most-referenced foundational methods, in the order they were developed — a place to start if you're new here.

循环神经网络1986–1990by Rumelhart, D. E.; Elman, J. L.
长短期记忆网络（LSTM）1997by Hochreiter, S. & Schmidhuber, J.
门控循环单元 (GRU)2014by Cho, K., van Merrienboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., & Bengio, Y.
生成对抗网络2014by Goodfellow, I. et al.
变分自编码器2014by Kingma, D. P. & Welling, M.
微调Transformer2017–2019by Vaswani et al. (architecture); fine-tuning paradigm popularised by Howard & Ruder, Devlin et al.
多模态Transformer2019–2021by Lu et al. (ViLBERT); Radford et al. (CLIP)
Vision Transformer2021by Dosovitskiy, A. et al.

all methods on this shelf ↓

全部方法 103

注意力机制自编码器双向循环神经网络 Crossformer：用于多元时间序列预测的跨维度依赖Transformer CycleGAN：具有循环一致性的非配对图像到图像翻译 DeepAR 扩散模型域自适应扩散模型域自适应GAN 领域自适应门控循环单元域自适应循环神经网络领域自适应句子嵌入 (Domain-Adaptive Sentence Embeddings)域自适应 Transformer 域自适应变分自编码器领域自适应视觉 Transformer 可解释扩散模型可解释生成对抗网络可解释门控循环单元 (Explainable GRU)可解释长短期记忆网络可解释循环神经网络可解释 Transformer 可解释变分自编码器 FEDformer：频率增强分解Transformer 微调扩散模型微调生成对抗网络微调门控循环单元 (Fine-Tuned GRU)微调长短期记忆网络 (Fine-Tuned LSTM)微调循环神经网络微调文本摘要微调Transformer 微调变分自编码器微调视觉Transformer 门控循环单元 (GRU)生成对抗网络图注意力网络门控循环单元 (GRU)Informer iTransformer：用于多元时间序列预测的倒置Transformer 潜在扩散模型长短期记忆网络（LSTM）长格式Transformer / BigBird 长短期记忆网络掩码自编码器 Moirai：通用时间序列预测Transformer 多语言扩散模型多语言生成对抗网络 (Multilingual GAN)多语言GRU 多语言长短期记忆网络多语言循环神经网络多语言文本摘要多语言变分自编码器多语言视觉Transformer 多模态扩散模型多模态生成对抗网络多模态门控循环单元 (Multimodal GRU)多模态LSTM 多模态循环神经网络多模态Transformer 多模态变分自编码器多模态视觉变换器非平稳Transformer PatchTST Pyraformer：用于长程时间序列预测的金字塔注意力Transformer 循环神经网络 Reformer：长序列的高效Transformer 基于得分的生成模型 SegRNN：用于长期时间序列预测的段循环神经网络多头自注意力机制自监督扩散模型自监督生成对抗网络自监督 GRU 自监督Transformer 自监督变分自编码器自监督视觉Transformer 半监督扩散模型 Semi-supervised GAN 半监督门控循环单元 (Semi-supervised GRU)半监督长短期记忆网络 (Semi-supervised LSTM)半监督式 Transformer 半监督变分自编码器半监督视觉变换器序列到序列模型 Swin Transformer T5（Text-to-Text Transfer Transformer）Temporal Fusion Transformer Time-MoE：面向通用时间序列的混合专家模型 TiRex：基于 xLSTM 的零样本时间序列预测模型迁移学习GAN 迁移学习与变分自编码器扩散模型迁移学习 LSTM 迁移学习循环神经网络迁移学习变分自编码器 Vision Transformer 瓦瑟施泰因生成对抗网络 (WGAN)弱监督扩散模型弱监督生成对抗网络 (Weakly Supervised GAN)弱监督门控循环单元 (Weakly Supervised GRU)弱监督 LSTM 弱监督循环神经网络弱监督 Transformer 弱监督变分自编码器弱监督视觉变换器

深度学习中的更多内容

架构与训练 124