{"version":2,"relationshipURLBase":"https://codingai-lec03.pages.dev/course/notebook-blocks/b114c530db554435857d485212f2ecba.html","relationshipResolution":"Resolve relative relationship URLs against relationshipURLBase; complete records and absolute URLs are in blocks.","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/b114c530db554435857d485212f2ecba.html","markdown":"https://codingai-lec03.pages.dev/course/notebook-blocks/b114c530db554435857d485212f2ecba.md","pages":[{"id":"reading-opening","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","unit":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","unitId":"notebook-blocks/b114c530db554435857d485212f2ecba","chapter":"course","number":1}],"aliases":[{"oldId":"b-b40851e0b4cd99daa9c050ef","targetId":"b-b40851e0b4cd99daa9c050ef","page":"reading-opening","operation":"move","beforeSHA256":"b3bb3626e5e17c0995867f42bedbcb72020e0decf77093d796851d5f10c78af6","afterSHA256":"b3bb3626e5e17c0995867f42bedbcb72020e0decf77093d796851d5f10c78af6","reason":"Same source body in authored semantic page"},{"oldId":"b-4f5dc93ca38587759edec42f","targetId":"b-4f5dc93ca38587759edec42f","page":"reading-opening","operation":"move","beforeSHA256":"0254c0745a22d89e14fcea0ddf8ca7243207b15adccd478ff7f67fb8ced0f2f4","afterSHA256":"0254c0745a22d89e14fcea0ddf8ca7243207b15adccd478ff7f67fb8ced0f2f4","reason":"Same source body in authored semantic page"},{"oldId":"b-1d4fab04853ef7a7e59f1613","targetId":"b-1d4fab04853ef7a7e59f1613","page":"reading-opening","operation":"move","beforeSHA256":"e91f24e4e4ca40e2099a0d81d30db6568f4bf2074b499eeee6e3959617bd97ac","afterSHA256":"e91f24e4e4ca40e2099a0d81d30db6568f4bf2074b499eeee6e3959617bd97ac","reason":"Same source body in authored semantic page"},{"oldId":"b-67d8251ea454c91af3b606ec","targetId":"b-67d8251ea454c91af3b606ec","page":"reading-opening","operation":"move","beforeSHA256":"74f0ef9a0e12c18b5c762e1998a58610fd064887211de86861e624d8f17a1e93","afterSHA256":"74f0ef9a0e12c18b5c762e1998a58610fd064887211de86861e624d8f17a1e93","reason":"Same source body in authored semantic page"},{"oldId":"b-a73a83752d66b15864edeb6b","targetId":"b-a73a83752d66b15864edeb6b","page":"reading-opening","operation":"move","beforeSHA256":"03813ef9144d8d18dd0bff567f4b82da7a228ee944b11e4279759dfbbd9d5bde","afterSHA256":"03813ef9144d8d18dd0bff567f4b82da7a228ee944b11e4279759dfbbd9d5bde","reason":"Same source body in authored semantic page"},{"oldId":"b-80f15f89ea5b12c26c54c3bc","targetId":"b-80f15f89ea5b12c26c54c3bc","page":"reading-opening","operation":"move","beforeSHA256":"1ad1c7532a741815e0ddb1fad16351fe19fa25774d57ac128b7c9b592e24436b","afterSHA256":"1ad1c7532a741815e0ddb1fad16351fe19fa25774d57ac128b7c9b592e24436b","reason":"Same source body in authored semantic page"},{"oldId":"b-9dbbdebc838fe6638afa0f5e","targetId":"b-9dbbdebc838fe6638afa0f5e","page":"reading-opening","operation":"move","beforeSHA256":"9cc36d4ff6902af2820b799f48c9096051a537912fc5b75d27a5dbc975c37577","afterSHA256":"9cc36d4ff6902af2820b799f48c9096051a537912fc5b75d27a5dbc975c37577","reason":"Same source body in authored semantic page"}],"blocks":[{"id":"b-b40851e0b4cd99daa9c050ef","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/b114c530db554435857d485212f2ecba.html#b-b40851e0b4cd99daa9c050ef","content":"[LAB 05 · 像素 patch 词典：拟合与重建](/course/notebooks/de39b998077f44cca4b8407085547766.html)","sha256":"b3bb3626e5e17c0995867f42bedbcb72020e0decf77093d796851d5f10c78af6","chapter":{"id":"course","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"unit":{"id":"notebook-blocks/b114c530db554435857d485212f2ecba","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"page":{"id":"reading-opening","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","number":1},"sectionTitle":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","currentQuestion":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":null,"nextBlock":"b-4f5dc93ca38587759edec42f","prerequisites":[],"relatedBlocks":[]},{"id":"b-4f5dc93ca38587759edec42f","type":"h1","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/b114c530db554435857d485212f2ecba.html#b-4f5dc93ca38587759edec42f","content":"# 历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","sha256":"0254c0745a22d89e14fcea0ddf8ca7243207b15adccd478ff7f67fb8ced0f2f4","chapter":{"id":"course","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"unit":{"id":"notebook-blocks/b114c530db554435857d485212f2ecba","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"page":{"id":"reading-opening","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","number":1},"sectionTitle":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","currentQuestion":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-b40851e0b4cd99daa9c050ef","nextBlock":"b-1d4fab04853ef7a7e59f1613","prerequisites":[{"id":"b-b40851e0b4cd99daa9c050ef","url":"#b-b40851e0b4cd99daa9c050ef","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-1d4fab04853ef7a7e59f1613","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/b114c530db554435857d485212f2ecba.html#b-1d4fab04853ef7a7e59f1613","content":"按问题继续","sha256":"e91f24e4e4ca40e2099a0d81d30db6568f4bf2074b499eeee6e3959617bd97ac","chapter":{"id":"course","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"unit":{"id":"notebook-blocks/b114c530db554435857d485212f2ecba","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"page":{"id":"reading-opening","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","number":1},"sectionTitle":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","currentQuestion":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-4f5dc93ca38587759edec42f","nextBlock":"b-67d8251ea454c91af3b606ec","prerequisites":[{"id":"b-4f5dc93ca38587759edec42f","url":"#b-4f5dc93ca38587759edec42f","role":"section-title"},{"id":"b-b40851e0b4cd99daa9c050ef","url":"#b-b40851e0b4cd99daa9c050ef","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-67d8251ea454c91af3b606ec","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/b114c530db554435857d485212f2ecba.html#b-67d8251ea454c91af3b606ec","content":"真实MNIST patch的Lloyd拟合、词典中心、保留误差与历史输出","sha256":"74f0ef9a0e12c18b5c762e1998a58610fd064887211de86861e624d8f17a1e93","chapter":{"id":"course","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"unit":{"id":"notebook-blocks/b114c530db554435857d485212f2ecba","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"page":{"id":"reading-opening","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","number":1},"sectionTitle":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","currentQuestion":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-1d4fab04853ef7a7e59f1613","nextBlock":"b-a73a83752d66b15864edeb6b","prerequisites":[{"id":"b-4f5dc93ca38587759edec42f","url":"#b-4f5dc93ca38587759edec42f","role":"section-title"},{"id":"b-b40851e0b4cd99daa9c050ef","url":"#b-b40851e0b4cd99daa9c050ef","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-a73a83752d66b15864edeb6b","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/b114c530db554435857d485212f2ecba.html#b-a73a83752d66b15864edeb6b","content":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子。本节另建实验身份，改用本册**训练图像的真实 7×7 像素 patch**。每张图切成不重叠的 4×4 个 patch，保留平均亮度大于 0.01 的 patch，学习 16 个中心。训练目标是 patch 到最近中心的平方距离和。","sha256":"03813ef9144d8d18dd0bff567f4b82da7a228ee944b11e4279759dfbbd9d5bde","chapter":{"id":"course","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"unit":{"id":"notebook-blocks/b114c530db554435857d485212f2ecba","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"page":{"id":"reading-opening","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","number":1},"sectionTitle":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","currentQuestion":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-67d8251ea454c91af3b606ec","nextBlock":"b-80f15f89ea5b12c26c54c3bc","prerequisites":[{"id":"b-4f5dc93ca38587759edec42f","url":"#b-4f5dc93ca38587759edec42f","role":"section-title"},{"id":"b-b40851e0b4cd99daa9c050ef","url":"#b-b40851e0b4cd99daa9c050ef","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-80f15f89ea5b12c26c54c3bc","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/b114c530db554435857d485212f2ecba.html#b-80f15f89ea5b12c26c54c3bc","content":"这里没有把数字类别交给词典。它学到的是压缩这些局部像素模式的一组中心；它没有因此证明自己学到了“数字概念”。这也不是 SIFT 描述子拟合，更不是 SimCLR、MAE 或 DINO 的复现。与现代方法的联系在问题层面：学习目标可以来自数据结构，而不一定来自固定类别名称。","sha256":"1ad1c7532a741815e0ddb1fad16351fe19fa25774d57ac128b7c9b592e24436b","chapter":{"id":"course","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"unit":{"id":"notebook-blocks/b114c530db554435857d485212f2ecba","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"page":{"id":"reading-opening","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","number":1},"sectionTitle":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","currentQuestion":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-a73a83752d66b15864edeb6b","nextBlock":"b-9dbbdebc838fe6638afa0f5e","prerequisites":[{"id":"b-4f5dc93ca38587759edec42f","url":"#b-4f5dc93ca38587759edec42f","role":"section-title"},{"id":"b-b40851e0b4cd99daa9c050ef","url":"#b-b40851e0b4cd99daa9c050ef","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-9dbbdebc838fe6638afa0f5e","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/b114c530db554435857d485212f2ecba.html#b-9dbbdebc838fe6638afa0f5e","content":"**预测 → 改一个条件：**先看 16 个中心，再只改 `N_WORDS` 为 8。重跑本节，对照训练距离、保留 patch 重建误差及可辨认笔画。不要同时改 patch 大小与词典大小。","sha256":"9cc36d4ff6902af2820b799f48c9096051a537912fc5b75d27a5dbc975c37577","chapter":{"id":"course","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"unit":{"id":"notebook-blocks/b114c530db554435857d485212f2ecba","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子"},"page":{"id":"reading-opening","title":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","number":1},"sectionTitle":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","currentQuestion":"历史 LAB07 原先的两个中心来自预设分组均值；后来加入的 Lloyd 拟合确实执行了优化，但输入仍是构造的二维描述子","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-80f15f89ea5b12c26c54c3bc","nextBlock":null,"prerequisites":[{"id":"b-4f5dc93ca38587759edec42f","url":"#b-4f5dc93ca38587759edec42f","role":"section-title"},{"id":"b-b40851e0b4cd99daa9c050ef","url":"#b-b40851e0b4cd99daa9c050ef","role":"problem-context"}],"relatedBlocks":[]}]}