Recently, I saw many students trying their best to find the innovation of their big papers, fidgeting in the teaching and research section every day, which is very painful. Personally, I think that innovation is like an opportunity. If you say it, you will come and leave. Teacher Li Kaifu once said that innovation needs accumulation. Before finding the innovation, we need to read a lot of related papers or books and compare the similarities and differences of related theories or models repeatedly. Only in this way will there be a spark of innovation in our brains! I personally like to record all the papers I have read for future reference. According to incomplete statistics, I read 106 Chinese short articles, 35 English short articles and 27 Chinese big papers (excellent master's or doctoral papers). When writing a small paper or a big paper, I will take out the records of the papers I have read and look at them several times to find the similarities or differences between them? Location. There may be innovative papers, and I will read them again and again until I see them thoroughly. The innovation of some of my small papers and big papers is almost based on the comparison of algorithms in the innovation of related papers. Because of sex. Therefore, association and analogy in scientific thinking are also very useful in the process of paper writing. I joked with my classmates that I must read Dialectics of Nature several times! Combined with my own experience, I think there are several ways to find innovation: the first method is to apply the X theory of A paper to the research of B paper, which can be called "grafting". For example, in the process of writing a big paper, I want to predict the number of hitchhikers in P2P networks. I learned in another paper that the grey system theory can be used to predict the population. I thought, aren't hitchhikers human? Is the predicted population the same? In order to predict the number of hitchhikers, then I can apply the grey system theory to my big thesis, and such an innovation is produced. The rest is to carry out a large number of experimental simulations to prove whether my conclusion is correct. Examples of algorithm interoperability in different fields abound. For example, the classic algorithm Dijkstra algorithm in the network is an example. This algorithm originally belongs to the field of mathematics, and network researchers apply it to routing algorithms. The second way is to improve the related algorithm of B paper by referring to the X algorithm of A paper. This is not an application of the algorithm, but an improvement of the existing algorithm. For example, I used this method when I was writing my paper "Weighted Trust Model Based on Cloud". At that time, I read a short paper "Research on Trust Evaluation Based on Cloud Model", in which there are two algorithms, one of which is very similar to the algorithm I saw in another paper "Subjective Trust Evaluation Based on Cloud Model", except that a weight is missing? . I immediately thought, what if this algorithm is weighted? What will happen? So, an innovation came into being and wrote a paper. The innovation of most papers is the improvement of algorithms, so the second method is used the most. The third way is to combine the X model of A paper and the Y model of B paper into the Z model of C paper. This method does not improve the algorithm, but combines two models into a new model. When writing my big thesis, I have established a model of P2P network (a three-dimensional structure), and I only need an algorithm to traverse all nodes in the network. At this time, I read some books and papers on algorithms, which mentioned that simulated annealing algorithm can traverse the whole network. Therefore, I combine the P2P network model with the simulated annealing algorithm model, and propose a new network traversal algorithm, and the feasibility of the model is proved by simulation experiments. This method can be said to be a comprehensive method. Before application, it is necessary to find out whether the two models can be combined, and it must be demonstrated by experiments. The above three methods are the methods I have used in the process of writing small papers and big papers. I think there is some truth, and I can really find out the innovation. It is very difficult for master students to invent new methods or theories. None of the papers I have read have been done. However, we have the ability to improve, apply or merge this algorithm. Finding innovation and accumulation is the key, that is, reading more books and summarizing more, keeping a curiosity and being good at finding problems. If we regard the search for innovation as "dawn", then now is "the darkest time before dawn".