This is an implementation of different methods for text classification with tensorflow, all the methods and experiments can be found here 文本分类实践
Method | Accuracy | Time(CPU) | Time(GPU) | Summary |
---|---|---|---|---|
lstm | 0.75 | 1h38min13s | 最耗时,误差小,准确率高 | |
cnn | 0.74 | 46min38s | 误差大,准确率低 | |
cnn-3-max | 0.72 | 46min34s | - | |
cnn-6-max | 0.725 | 46min53s | - | |
c-lstm | 0.76 | 59min54s | 误差波动大,准确率波动大 | |
fastText-tf | 0.79(调参后) | 13min28s | 2min42s | 时间短,误差小,准确率相对高 |
fastText-mask | 0.77 | 13min31s | - | |
fastText-fb | 0.77 | 2s | - | 速度极快,准确率不低 |