Table 1 Comparison of computational cost and settings across different models

From: Accurate label free classification of cancerous extracellular vesicles using nanoaperture optical tweezers and deep learning

Model

Parameters

Training time

Epochs to Conv.

Learning rate

Patience

LSTM

48.5M

3h 17m

562

5 × 10−5

200

Bi-LSTM

138.9M

2h 51m

325

5 × 10−5

200

CLSTM

17.4M

9h 58m

520

5 × 10−5

200

CNN

2.1M

2h 22m

760

5 × 10−5

200

Transformer

35k

0h 24m

1070

5 × 10−5

200

TrapNet

2.1M

1h 41m

954

5 × 10−5

200

Back to article page