Chat with friends and sentiment

　　sentiment: has been on their own in the group of research direction is particularly confused, on the one hand I want to let themselves in the group research Spark, but the group of senior and mentor to spark is not too familiar with, there is no specific task driven to complete this matter; On the other hand, I think the senior Learning is more familiar with, then perhaps can learn something. But today and classmate Kam-kun exchange and feel for the present of their own deep learning can not make anything particularly useful achievements, this direction is more suitable for research rather than do engineering, but I do research information is not large, so deep Learning in this direction may be of little help to future employment. However, spark words, for the group, Wang and senior are strongly support, and last semester I basically half play half to come over, almost did not do what the group and to their useful facts, this need to deeply self-criticism. Aside from this, spark is promising for both group and future jobs, and I'm a big data-oriented platform, well-learned data processing platforms, and a good way to build on it, and learn some of the ways to analyze and work with it, combined with deep learning and other technologies, This will certainly be beneficial for future development.

which way to go in the future Floss (encounter every new thing to try to use a)? 1. Use all builds of the spark biosphere: including the development package mllib,sparksql,sparkstreamming, which requires a summary of the use of each tool, performance analysis of the document; 2. Spark the corresponding paper and spark The video and PPT of the development application presented on the submit is well researched, and this needs to summarize the work of spark in theory, application and some trick documents; 3. Record every pit that has been trampled in spark practice and follow the spark Community's discussions and analysis; 4.    Back to Spark's source code, log analysis of the unclear process in spark, preferably this build process needs to be distributed test platform. 5. After the spark has been cooked up, it's going to start digging into the entire framework of the spark ecosystem, such as the in-depth study of yarn and HDFs, and then the analysis of some NoSQL databases.

Chat with friends and sentiment