Innovation has always been a tangled topic. Most of the graduate design of graduate students requires algorithm innovation, and doctoral graduates require a lot of innovation. Here, we will talk about how to carry out reasonable algorithm innovation based on the work experience of the team over the past few years.
1. Innovative perspective
Usually, we use an algorithm. Here is a simple particle, the PSO particle swarm optimization algorithm. Through simulation, we will get the convergence speed of the algorithm, simulation accuracy and other parameters. If we need to innovate the algorithm, we generally need to consider the performance indicators of the original algorithm, such as the improvement of the convergence speed and accuracy. For some low requirements, we can improve the accuracy while ensuring the convergence speed. In the case of the same accuracy, speed up the convergence speed. If the requirement is high, consider how to improve the convergence speed and accuracy at the same time. Under normal circumstances, for this situation, we need to consider other similar algorithms, and these algorithms must have certain characteristics in fast convergence and high precision, so that we can integrate the advantages of multiple algorithms to achieve the algorithm Innovate and improve the performance of the original algorithm.
For low-level innovation, this is generally the way to go.
2. Basis for innovation
What is the basis for innovation? The so-called innovation basis is the innovative algorithm we use. The angle considered has a theoretical basis. We cannot modify the original algorithm out of thin air, so that even if a better solution of the layout is obtained, it cannot prove that the overall situation is better. Therefore, as I said, we need to choose some other off-the-shelf algorithms with better performance than the original algorithm to integrate the algorithms.
3. Verification of innovation
For the improved algorithm, we need to use a large number of test samples for comparative analysis to verify whether the various performance indicators of the algorithm are improved compared to the original algorithm.
Four, complete innovation
Complete innovation, this is a high-demand innovation, generally has two levels, one is interdisciplinary innovation, and the other is theoretical development.
Looking at the situation in the past few years, generally, a slightly more demanding master’s degree will involve interdisciplinary algorithm innovation. For example, we apply some formulas in physics to formulas in biochemistry and combine them to innovate, or Use a set of mathematical formulas to study some liberal arts problems.
The development of theory, this is the innovation of the highest requirements, that is, there is no existing theoretical basis, we need to research and summarize the relevant theories. The general student’s topic will not involve this requirement, and it is usually some more complicated. For engineering projects or doctoral topics, some new formulas need to be calculated based on some known conditions to characterize a process or phenomenon. This is very difficult.
Welcome to add this blog’s assistant WeChat account: c840893
☆☆☆☆☆☆☆☆☆☆☆☆☆☆☆☆☆☆☆☆☆☆
