MỘT PHƯƠNG PHÁP TỰ CHƯNG CẤT KIẾN THỨC HIỆU QUẢ ĐỂ PHÂN LOẠI CHẤT THẢI=AN EFFECTIVE SELF-KNOWLEDGE DISTILLATION METHOD FOR WASTE CLASSIFICATION

0 Người đánh giá. Xếp hạng trung bình 0

Tác giả: Thị Hồng Thu Ma

Ngôn ngữ: vie

Ký hiệu phân loại:

Thông tin xuất bản: Tạp chí Khoa học và Công nghệ - Đại học Thái Nguyên, 2024

Mô tả vật lý: tr.85 - 92

Bộ sưu tập: Báo, Tạp chí

ID: 479233

Thêm vào giỏ

Garbage classification is a critical task in overcoming environmental pollution. People classify waste based on what they understand about the waste object rather than on its recyclable status, which leads to inaccurate classification in manual sorting. In addition, the direct processing of toxic waste can be physically dangerous for humans. Several machine learning and deep learning techniques have been proposed using standardized garbage classification datasets. However, these methods still have some disadvantages such as (i) the computational cost and memory of the proposed networks are large
(ii) they are difficult to run in real time
(iii) The training process is complicated because it is pre-trained on large data sets to avoid the phenomenon of "overfitting". In this paper, we propose a new approach based on self-knowledge distillation to overcome the above drawbacks. Experimental results have shown that our model achieves the best performance on the Trash Net dataset without having to pre-train the model. This proves the effectiveness of the self-knowledge distillation method brings to the problem of waste classification.Phân loại rác là một nhiệm vụ trọng tâm trong việc khắc phục ô nhiễm môi trường. Con người phân loại rác dựa trên những gì họ hiểu về đối tượng rác hơn là dựa trên trạng thái có thể tái chế của rác, điều này dẫn đến việc phân loại không chính xác trong cách phân loại thủ công. Ngoài ra, việc tiếp xúc trực tiếp với chất thải độc hại có thể gây nguy hiểm về mặt thể chất cho con người. Một số kỹ thuật học máy và học sâu đã được đề xuất bằng cách sử dụng bộ dữ liệu phân loại rác được chuẩn hóa. Tuy nhiên, các phương pháp này còn tồn tại một số nhược điểm như (i) chi phí tính toán và bộ nhớ của các mạng được đề xuất lớn (ii) khó thể chạy trong thời gian thực
(iii) Quá trình huấn luyện phức tạp do được huấn luyện trước trên các bộ dữ liệu lớn để tránh hiện tượng “overfitting”. Trong bài báo này chúng tôi đề xuất một cách tiếp cận mới dựa trên tự chưng cất kiến thức để khắc phục các nhược điểm trên. Kết quả thử nghiệm cho thấy mô hình được đề xuất đạt hiệu suất tốt nhất trên bộ dữ liệu Trash Net mà không cần phải huấn luyện trước mô hình. Điều này chứng minh sự hiệu quả của tự chưng cất kiến thức cho bài toán phân loại chất thải.

1. Đọc trực tuyến

1. Đọc trực tuyến

Tạo bộ sưu tập với mã QR