Ashish Patel 🇮🇳’s Post

𝗗𝗮𝘆-𝟭𝟲𝟴 Computer Vision Learning Convolutional Neural Networks on Graphs with 𝗙𝗮𝘀𝘁 𝗟𝗼𝗰𝗮𝗹𝗶𝘇𝗲𝗱 𝗦𝗽𝗲𝗰𝘁𝗿𝗮𝗹 𝗙𝗶𝗹𝘁𝗲𝗿𝗶𝗻𝗴 by Ecole polytechnique fédérale de Lausanne, Lausanne, Switzerland Follow me for a similar post:  🇮🇳 Ashish Patel Interesting Facts : 🔸 𝗧𝗵𝗶𝘀 𝗶𝘀 𝗮 𝗽𝗮𝗽𝗲𝗿 𝗶𝗻 𝗡𝗜𝗣𝗦 𝟮𝟬𝟭𝟲 𝘄𝗶𝘁𝗵 𝗼𝘃𝗲𝗿 𝟯𝟰𝟭𝟲 𝗰𝗶𝘁𝗮𝘁𝗶𝗼𝗻𝘀. 🔸 It outperforms Classical CNN etc. ------------------------------------------------------------------- 𝗔𝗺𝗮𝘇𝗶𝗻𝗴 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 : https://lnkd.in/gvC9hvb code : https://lnkd.in/gd7eEyd ------------------------------------------------------------------- 𝗜𝗠𝗣𝗢𝗥𝗧𝗔𝗡𝗖𝗘 🔸 CNN's in the context of spectral graph theory, which provides the necessary mathematical background and efficient numerical schemes to design fast localized convolutional filters on graphs.  🔸 (i) This model whose computational complexity is linear with the dimensionality of the data, (ii) It confirmed that the quality of the input graph is of paramount importance, (iii) It showed that the statistical assumptions of local stationarity and compositionality made by the model are verified for text documents as long as the graph is well constructed. #computervision #artificialintelligence #data