Name
Abstract | ||
|---|---|---|
It happens that one you are following in Twitter has several aspects in his personality. Suppose we are so attracted by his bright aspect that we are following his tweets. Then we have to endure dirty tweets from his dark side. In attempt to construct Twitter client software that filters out dirty tweets (from our point of view), we trained neural networks of different depths by the standard back propagation algorithm and tested if they can tell preferable tweets from dirty ones. The result says that the shallow network having only two hidden layers and more than 32 and 64 units per layer achieves over 96% and 97% of accuracy, respectively. |
Year | DOI | Venue |
|---|---|---|
2016 | 10.1109/NBiS.2016.26 | PROCEEDINGS OF 2016 19TH INTERNATIONAL CONFERENCE ON NETWORK-BASED INFORMATION SYSTEMS (NBIS) |
Keywords | Field | DocType |
Neural networks, twitter filtering, back propagation | Data mining,Back propagation algorithm,Client,Computer science,Filter (signal processing),Backpropagation,Artificial neural network | Conference |
ISSN | Citations | PageRank |
2157-0418 | 0 | 0.34 |
References | Authors | |
0 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Shumpei Fujii | 1 | 0 | 0.34 |
| Michitoshi Niibori | 2 | 11 | 10.70 |
| Masaru Kamada | 3 | 115 | 85.04 |