| Using autoencoders to visualize documents |
| output |
| vector |
| 2000
reconstructed counts |
|
|
|
|
|
|
|
|
|
|
|
|
| • | Instead of using |
||||||||||
| codes to retrieve |
|||||||||||
| documents, we
can |
|||||||||||
| use 2-D codes to |
|||||||||||
| visualize sets of |
|||||||||||
| documents. |
|||||||||||
| – | This
works much |
||||||||||
| better
than 2-D |
|||||||||||
| PCA |
|||||||||||
| 500 neurons |
| 250 neurons |
| 250 neurons |
| 500 neurons |
|
|
| input |
|
| vector |
|
| 2000
word counts |