Deep Learning for the Soft Cutoff Problem
undefined
Deep Learning for the
Soft Cutoff Problem
Miles Saffran
Introduction
The MATERIAL project
The soft cutoff problem
Metric of evaluation
Figure 3.
French loss in the simple model
Figure 3.
French loss in the simple model
Materials and Methods
Data collection
Input features
Query embedding
Document length
Indri document score
Construction
TensorFlow
Results
F
i
g
u
r
e
1
.
T
r
a
i
n
i
n
g
l
o
s
s
o
v
e
r
e
p
o
c
h
s
Results
F
i
g
u
r
e
2
.
E
n
g
l
i
s
h
l
o
s
s
w
i
t
h
d
i
f
f
e
r
e
n
t
l
e
a
r
n
i
n
g
r
a
t
e
s
Results
Figure 3.
French loss in the simple model
Results
Results
Results
Variance in performance
.1 on English to English (optimal .14)
.15 on Tagalog to Swahili (optimal .35)
Conclusion
Add more features
Use more training data
Include dropout and regularization
Exploring deep learning techniques for solving the soft cutoff problem, this study by Miles Saffran discusses the MATERIAL project, data collection, methods like query embedding and TensorFlow construction, and presents results with training loss trends and performance variances. The conclusion suggests adding more features, using more training data, and incorporating dropout and regularization for improved results.
Download Presentation
Please find below an Image/Link to download the presentation.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. Download presentation by click this link. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.
E N D
Presentation Transcript
Deep Learning for the Soft Cutoff Problem Miles Saffran
Introduction The MATERIAL project The soft cutoff problem Metric of evaluation
Materials and Methods Data collection Input features Query embedding Document length Indri document score Construction TensorFlow
Results Figure 1. Training loss over epochs
Results Figure 2. English loss with different learning rates
Results Variance in performance .1 on English to English (optimal .14) .15 on Tagalog to Swahili (optimal .35)
Conclusion Add more features Use more training data Include dropout and regularization
