3月7日音訊,中國科學(xué)院自動(dòng)化研討所曾毅研討員課題組研討發(fā)現(xiàn),人工智能與人類尚有顯著認(rèn)知差距:深度神經(jīng)網(wǎng)絡(luò)對(duì)幻覺輪廓“視而不見”。
該課題組近日在(Cell Press)細(xì)胞出版社旗下期刊Patterns上發(fā)表了一篇題為“Challenging Deep Learning Models with Image Distortion based on the Abutting Grating Illusion”的新研討。