小弟撰寫的類神經pca對圖片的壓縮與解壓縮,對來源圖片training過後,可使用該張圖像的特性(eigenvalue和eigenvetex)來對別張圖解壓縮,非常有趣的方式,再設定threashold時注意時值不要過大,因為這牽涉inverse matrex的計算.-Boy wrote on Neural PCA image compression and decompression of the source image after training, can make use of the characteristics of images (eigenvalue and eigenvetex) to other copies of illustrated compression, a very interesting way, and then set when threashold attention values not too large, because it involves the calculation of inverse matrex.