youtube 推荐器如何训练 video_id 嵌入?

How does youtube recommender train video_id embedding?

Youtube 结合了几个视频并使用平均嵌入,但训练嵌入。

来自 Deep Neural Networks for YouTube Recommendations

的 3.2

The network requires fixed-sized dense inputs and simply averaging the embeddings performed best among several strategies (sum, component-wise max, etc.). Importantly, the embeddings are learned jointly with all other model parameters through normal gradient descent backpropagation updates.

当您输入嵌入的平均值时,如何训练嵌入?

这就像先有鸡还是先有蛋的问题。如果你想为视频平均嵌入,你需要一个嵌入,但是在你有训练数据(这是平均嵌入)之前你怎么能有一个嵌入

为了社区的利益,在回答部分提到了澄清(即使它出现在 eugene 的评论部分)。

Youtube Recommender may train video_id embedding by just Randomly Initializing Embeddings and start from there.. average the Random Embeddings, and Back propagate, and Iterate.