ConvRNN Cell and ConvLSTM Cell

描述

贡献者指南

技术栈

无

领域

machine learning

议题类型

feature

难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。

4

预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。

3-5 days

活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。

stale

清晰度议题是否清楚说明期望改动、验收标准和下一步。

mostly clear

前置要求

Familiarity with Flux.jlUnderstanding of convolutional layersBasic knowledge of recurrent networks

新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。

35

研究方向

Refer to the linked MXNet implementation for reference. Start by examining existing convolutional layers in Flux (e.g., Conv, ConvTranspose) and recurrent cells (e.g., LSTM, GRU). Implement a ConvRNN cell similar to the MXNet version, then combine multiple cells to form ConvLSTM. Ensure compatibility with Flux's automatic differentiation. Open a draft PR early for feedback.