Gossip Point HDP model Bar _ gossip

Since I persisted for so long, see understand, put my doctor work to delay actually also a lot of, also do not calculate delay, main is the foundation is poor, this algorithm and trouble, so have seen now also did not have the result. I think if someone at that time can guide me, tell me this is difficult, need a professional knowledge of the theoretical background, I may not go on. And if someone were to study the discussion together, it would probably be much better now. But that's all hypothetical, isn't it.

However, since it has been done, and achieved, we must continue to pay attention to it.

First of all, the HDP model is not omnipotent, but at least for now it seems to me to be a very flexible model. The most famous model extended from HDP is the introduction of the HDP-HMM model with HDP. This shows that HDP not only can be used as a model algorithm, but also as a priori distribution of other models, which is the most attractive place in HDP, I think.

As we know, because the HDP model can generate the number of clustering, so the number of the HMM clustering can be used without human input.

The beginning of the HDP model is used in text processing, but it is now widely used in image video processing.

In the model, there are several features, including, of course, shortcomings. First of all, the calculation is too large, even the decomposition also requires a large amount of computation. However, the HDP model is beautiful, furthermore, it can generate clustering number and complete data modeling, which is more convenient for inference and prediction.