Weekly Activities and Research Updates in Machine Learning - April 18, 2023

Adri Priadana's weekly report details recent activities like courses on machine learning, doctoral thesis research, and face recognition experiments using PyTorch. Updates on FasterNet Block, Split-based Inception Block, ResNet architecture modifications for improved face recognition accuracy are included.

• Machine Learning
• Research
• Face Recognition
• PyTorch
• ResNet

pess Follow

Uploaded on Sep 21, 2024 | 0 Views

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.

Presentation Transcript

1. Weekly Report Adri Priadana April 18, 2023

2. Last Week Activities Doing the course SoC Design Special Topics Introduction to Machine Learning Do the assignment Doctoral Thesis Research Course II 2

3. Last Week Activities TII Journal Face Recognition Learn PyTorch and try to code Split-based Inception Block (SIB) in PyTorch Faced a problem Model Parameters Epoch LFW CFP-FP AgeDB CALFW CPLFW Average R-50 (Keras & Tensorflow) 43.6 M 24 99.67 96.34 97.22 95.88 91.40 96.10 R-50 (PyTorch) 43.6 M 24 99.70 96.59 97.52 95.72 91.48 96.20 R-50 with FasterNetBlock(Scale 2) 25 M 24 99.58 94.73 96.70 95.48 89.92 95.28 R-50 with FasterNetBlock(Scale 4) 28.7 M 24 99.60 95.19 96.78 95.57 90.12 95.45 R-50 with FasterBlock + SIB 20.2 M 24 50.00 50.00 50.00 50.00 50.00 50.00 R-50 with FasterBlock(Scale 2) + SE 25.5 M 24 50.00 50.00 50.00 50.00 50.00 50.00 R-50 with FasterBlock(Scale 2) + SE 25.5 M 12 50.00 50.00 50.00 50.00 50.00 50.00 R-50 with FasterBlock(Scale 2) 25 M 4 50.00 50.00 50.00 50.00 50.00 50.00 R-50 (Reset the Model Code) 43.6 M 4 98.83 85.84 89.60 92.07 83.85 90.04 R-50 with FasterBlock(Scale 2) 25 M 4 50.00 50.00 50.00 50.00 50.00 50.00 Why? 3

4. This Week Activities Doing the course SoC Design Special Topics Prepare for the Min-term Exam (Friday, April 21, 2023) Introduction to Machine Learning Prepare for the Min-term Exam (Tuesday, April 25, 2023) Doctoral Thesis Research Course II Supervise Mid-term Exam Prepare for IECON 2023 Facial Attribute Recognition TII Journal Face Recognition Continue to figure out the model problem in PyTorch 4

5. Thank You 5

6. FasterNet Block 6

7. Split-based Inception Block (SIB) 7

8. ResNet 3, 4, 14, 3 8

9. TII Journal Continue to try to improve or modify the ResNet architecture by developing some mechanism for Face Recognition (Target much more efficient with competitive or outperform accuracy) Try to train on decreased Dataset (10 epoch) Decrease the instance (Ins15) : 1,250,590 (1 epoch 1 hour 10 minutes) Decrease the class (C20000) : 1,530,886 (1 epoch 1 hour 20 minutes) Result Model Parameters LFW CFP-FP AgeDB CALFW CPLFW Average R-50 on C20000 44 M 99.58 93.64 96.05 95.23 88.62 94.62 R-50 on Ins15 44 M 99.58 94.16 95.93 95.22 89.83 94.94 R-50 (SE | GC+CBAM) on Ins15 26 M 99.48 94.24 95.63 95.07 89.73 94.83 | : stage 1,2,3 | 4 of ResNet50 GC : Group Convolution SE : Squeeze and Excitation CBAM : Convolutional Block Attention module 9

10. TII Journal Continue to try to improve or modify the ResNet architecture by developing some mechanism for Face Recognition (Target much more efficient with competitive or outperform accuracy) Try to train on decreased Dataset (24 epoch) Decrease the instance (Ins15) : 1,250,590 (1 epoch 1 hour 10 minutes) Result Model Parameters LFW CFP-FP AgeDB CALFW CPLFW Average R-50 44 M 99.67 96.34 97.22 95.88 91.40 96.10 R-50 (SE | GC+CBAM) 26 M 99.68 95.99 96.65 95.87 90.87 95.81 | : stage 1,2,3 | 4 of ResNet50 GC : Group Convolution SE : Squeeze and Excitation CBAM : Convolutional Block Attention module 10