First, coding standards - do a fine programmer
This semester I used GNU Coding Standards (GNU coding standard) at the following link:
http://www.sourceformat.com/pdf/cpp-coding-standard-gnu.pdf
Through the network I know some coding standard C ++ language, for different programming languages, a specific coding standards can be very different, but the purpose is the same, it is to ensure completion of the code requirements along with high quality good readability, maintainability. For example, we can specify a project of C language program to follow such a provision: naming variables, writing and #include header files and so on.
Here are some widely used coding standard:
• GNU Coding Standards
• Guidelines for the Use of the C Language in Vehicle Based Software
• C++ Coding Guidelines
• SUN Code Conventions for Java
Here are some introduction coding, coding standard books:
• C ++ coding standards, Chen Shizhong, Posts & Telecom Press, 2002
• High-quality programming guide: C ++ / C language, Lin Rui, etc. Electronic Industry Press, 2003
Note: The following summary is only given based on existing experience research group, is not applicable for all scenarios.
For high quality engineering, usually do:
1. The code is simple refining, appearance, readability, high efficiency, high multiplexing, portability, high cohesion, low coupling, there is no redundancy, does not conform to these principles, must be specified.
2. normative code there are rules. Special typesetting, a special syntax, special instructions must be specified.
Second, the beauty of mathematics - infinite magical algorithm
These days the shallow reading "mathematical beauty" of this book, I think it benefits greatly, at least this is a good book!
After reading the directory, I read two chapters I am interested in the election: "The beauty of mathematical series SIXTEEN - Do not put all your eggs in one basket - Maximum Entropy" (upper and lower two).
Start rolling dice to begin with, the probability on each face up to one-sixth of all, this is a probability "entropy maximization" - can be considered to maximize the entropy model.
When we need to predict the probability distribution of a random event, we should forecast all known conditions are met, and the unknown circumstances do not make any subjective assumptions. In this case, the probability distribution of most uniform, minimum risk prediction. Because then the probability distribution of maximum entropy, so people call this model is called "maximum entropy model." "Do not put all your eggs in one basket, in fact, a simple statement of the principle of maximum entropy, because when we encounter uncertainty, it is necessary to retain the possibilities.
The maximum entropy model exists and is unique, in the form of a simple exponential function, the algorithm from GIS development to IIS, the iteration time and stability have been improved. This algorithm is simple and complex set in one simple form, implementation complexity, but widely used in many Google products, such as machine translation, directly or indirectly, use the maximum entropy model.
After reading a brief two chapters, I'm also quite interested in the algorithm, a section of code similar to the "template" or models on a number of very complex difficult cumbersome steps can be achieved. Before my memory of some of the algorithms is disdain, thinking I do not need these algorithms can achieve the goal I want. Now that I think it is naive, to stand on the shoulders of giants overlooking, why not do it?
Third, the term target - strong motivation for learning
Algorithm design and analysis of these semesters, I want to be able to make a simple procedure to check the weight of the (first set small goals), if enough time, capacity, and I want to be a voice recognition function more or mapping programs.