{"id":1224,"date":"2026-04-13T11:15:13","date_gmt":"2026-04-13T03:15:13","guid":{"rendered":"https:\/\/www.ndnlab.com\/?p=1224"},"modified":"2026-04-13T11:15:23","modified_gmt":"2026-04-13T03:15:23","slug":"vega-an-active-tuning-learned-index-with-group-wise-learning-granularity","status":"publish","type":"post","link":"https:\/\/www.ndnlab.com\/?p=1224","title":{"rendered":"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity"},"content":{"rendered":"\n<p><strong>1. \u6458\u8981\uff08Abstract\uff09<\/strong><\/p>\n\n\n\n<p>VEGA\u63d0\u51fa\u4e86\u4e00\u79cd\u5168\u65b0\u7684\u53ea\u8bfb\u578b\uff08immutable\uff09\u5b66\u4e60\u7d22\u5f15\uff0c\u65e8\u5728\u540c\u65f6\u5b9e\u73b0\u6700\u5148\u8fdb\u7684\u7ecf\u9a8c\u67e5\u8be2\u6027\u80fd\u548c\u4e25\u683c\u7684\u7406\u8bba\u67e5\u8be2\u590d\u6742\u5ea6\u4fdd\u8bc1\u3002\u73b0\u6709\u6700\u5feb\u7684\u4e0d\u53d8\u5b66\u4e60\u7d22\u5f15\uff08\u5982RMI\uff09\u867d\u5728\u5b9e\u9645\u67e5\u8be2\u541e\u5410\u91cf\u4e0a\u8868\u73b0\u7a81\u51fa\uff0c\u4f46\u7f3a\u4e4f\u975e\u5e73\u51e1\u7684\u6700\u574f\u60c5\u51b5\u67e5\u8be2\u754c\u9650\uff1b\u76f8\u53cd\uff0c\u63d0\u4f9b\u7d27\u81f4\u754c\u9650\u7684\u7d22\u5f15\uff08\u5982PGM\uff09\u867d\u7406\u8bba\u4e0a\u53ef\u9760\uff0c\u5374\u5728\u5e73\u5747\u67e5\u8be2\u6027\u80fd\u4e0a\u663e\u8457\u843d\u540e\u3002\u8fd9\u79cd\u7406\u8bba\u4e0e\u5b9e\u8df5\u7684\u8131\u8282\u6210\u4e3a\u5b66\u4e60\u7d22\u5f15\u9886\u57df\u957f\u671f\u60ac\u800c\u672a\u51b3\u7684\u6838\u5fc3\u95ee\u9898\u3002VEGA\u901a\u8fc7\u5f15\u5165\u4e24\u5927\u5728\u7ebf\u6a21\u578b\u6784\u5efa\u7b56\u7565\u5f7b\u5e95\u89e3\u51b3\u4e86\u8fd9\u4e00\u96be\u9898\uff1a\u4e00\u662f\u91c7\u7528\u5408\u9002\u7684\u7c92\u5ea6\uff08group-wise learning granularity\uff09\u7b80\u5316\u6570\u636e\u5206\u5e03\uff0c\u5373\u5c06\u591a\u4e2a\u8fde\u7eed\u952e\u6253\u5305\u6210\u7ec4\uff0c\u4ec5\u7528\u7ec4\u5185\u9996\u952e\u8fdb\u884c\u6a21\u578b\u8bad\u7ec3\uff0c\u4ece\u800c\u5927\u5e45\u51cf\u5c11\u6240\u9700\u6a21\u578b\u6570\u91cf\uff1b\u4e8c\u662f\u4e3b\u52a8\u8c03\u4f18\u5206\u5e03\uff08active distribution tuning\uff09\uff0c\u901a\u8fc7\u952e\u91cd\u5b9a\u4f4d\uff08key repositioning\uff09\u5c06\u7b80\u5316\u540e\u7684\u5206\u5e03\u8c03\u6574\u4e3a\u66f4\u6613\u62df\u5408\u7684\u5f62\u5f0f\uff0c\u4f7f\u5355\u5c42\u6a21\u578b\u6570\u91cf\u8fdb\u4e00\u6b65\u964d\u4f4e\u81f31\u3002<\/p>\n\n\n\n<p>\u7cfb\u7edf\u8fdb\u4e00\u6b65\u63d0\u51fa\u4e00\u4e2a\u901a\u7528\u6846\u67b6\uff0c\u5728\u7ed9\u5b9a\u5185\u5b58\u9884\u7b97\u4e0b\u52a8\u6001\u7ec4\u5408F1\uff08Fitting model to simplified distribution\uff09\u548cF2\uff08Fitting tuned distribution to model\uff09\u4e24\u79cd\u7b56\u7565\uff0c\u5b9e\u73b0\u67e5\u8be2\u6027\u80fd\u7684\u6700\u4f18\u6743\u8861\u3002VEGA\u91c7\u7528\u591a\u5c42\u6876\u6570\u7ec4\u7ed3\u6784\uff0c\u5e95\u5c42\u4e3a\u7d27\u51d1\u6876\uff08compact bucket\uff09\uff0c\u4e0a\u5c42\u6df7\u5408\u7d27\u51d1\u6876\u4e0e\u7a00\u758f\u6876\uff08sparse bucket\uff09\uff0c\u5e76\u5f15\u5165\u5feb\u901f\u8def\u5f84\uff08fast path\uff09\u673a\u5236\u5229\u7528\u7a00\u758f\u6876\u4e2d\u7684\u7a7a\u69fd\u7ed5\u8fc7\u4e2d\u95f4\u5c42\u3002\u5411\u91cf\u5316\u7684\u6876\u5185\u641c\u7d22\uff08SIMD-based in-bucket search\uff09\u8fdb\u4e00\u6b65\u5f25\u8865\u4e86\u56e0\u7c92\u5ea6\u653e\u5bbd\u5e26\u6765\u7684\u9884\u6d4b\u8bef\u5dee\u3002<\/p>\n\n\n\n<p>\u5927\u91cf\u5b9e\u9a8c\u5728SOSD\u548cupdatable learned index\u57fa\u51c6\u7684\u591a\u4e2a\u6700\u96be\u771f\u5b9e\u6570\u636e\u96c6\u4e0a\u9a8c\u8bc1\u4e86VEGA\u7684\u6709\u6548\u6027\uff1a\u76f8\u6bd4RMI\u548cPGM\u7b49\u6700\u5148\u8fdb\u65b9\u6cd5\uff0cVEGA\u5728\u67e5\u8be2\u541e\u5410\u91cf\u548c\u6784\u5efa\u541e\u5410\u91cf\u4e0a\u5747\u53d6\u5f97\u663e\u8457\u9886\u5148\uff0c\u540c\u65f6\u4fdd\u6301\u4e86\u4e0e\u5168\u901a\u4fe1\u3001\u65e0\u7f13\u5b58\u6846\u67b6\u76f8\u5f53\u7684\u7cbe\u5ea6\u3002\u8be5\u5de5\u4f5c\u4e0d\u4ec5\u586b\u8865\u4e86\u5b66\u4e60\u7d22\u5f15\u5728\u7406\u8bba\u754c\u9650\u4e0e\u7ecf\u9a8c\u6027\u80fd\u4e4b\u95f4\u7684\u7a7a\u767d\uff0c\u8fd8\u4e3a\u53ea\u8bfb\u578b\u5b66\u4e60\u7d22\u5f15\u7684\u8bbe\u8ba1\u63d0\u4f9b\u4e86\u53ef\u6269\u5c55\u3001\u53ef\u7406\u8bba\u4fdd\u8bc1\u7684\u7cfb\u7edf\u6846\u67b6\uff0c\u5177\u6709\u91cd\u8981\u7684\u5b66\u672f\u521b\u65b0\u4ef7\u503c\u548c\u5de5\u7a0b\u5b9e\u7528\u610f\u4e49\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"807\" height=\"291\" src=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/1.png\"  class=\"wp-image-1226\" srcset=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/1.png 807w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/1-300x108.png 300w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/1-768x277.png 768w\" sizes=\"auto, (max-width: 807px) 100vw, 807px\" title=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe\" alt=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe\" \/><\/figure>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"807\" height=\"258\" src=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/2.png\"  class=\"wp-image-1227\" srcset=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/2.png 807w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/2-300x96.png 300w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/2-768x246.png 768w\" sizes=\"auto, (max-width: 807px) 100vw, 807px\" title=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe1\" alt=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe1\" \/><\/figure>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"801\" height=\"261\" src=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/3.png\"  class=\"wp-image-1228\" srcset=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/3.png 801w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/3-300x98.png 300w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/3-768x250.png 768w\" sizes=\"auto, (max-width: 801px) 100vw, 801px\" title=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe2\" alt=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe2\" \/><\/figure>\n\n\n\n<p><strong>2. \u5f15\u8a00\uff08Introduction\uff09<\/strong><\/p>\n\n\n\n<p>\u8fd1\u5e74\u6765\uff0c\u673a\u5668\u5b66\u4e60\u5728\u6784\u5efa\u7d22\u5f15\u7ed3\u6784\u4e0a\u7684\u5e94\u7528\u53d6\u5f97\u4e86\u663e\u8457\u6210\u529f\uff0c\u5b66\u4e60\u7d22\u5f15\u901a\u8fc7\u7528\u6a21\u578b\u62df\u5408\u952e\u503c\u6620\u5c04\u5173\u7cfb\uff0c\u5927\u5e45\u63d0\u5347\u4e86\u67e5\u8be2\u6548\u7387\u3002\u7136\u800c\uff0c\u4e0d\u53d8\u5b66\u4e60\u7d22\u5f15\uff08immutable learned index\uff09\u5728\u5b9e\u9645\u90e8\u7f72\u4e2d\u9762\u4e34\u4e00\u4e2a\u957f\u671f\u5b58\u5728\u7684\u7406\u8bba\u4e0e\u5b9e\u8df5\u8131\u8282\u95ee\u9898\uff1a\u6700\u5feb\u7684RMI\u867d\u5728\u7ecf\u9a8c\u67e5\u8be2\u6027\u80fd\u4e0a\u9886\u5148\uff0c\u5374\u7f3a\u4e4f\u975e\u5e73\u51e1\u7684\u6700\u574f\u60c5\u51b5\u67e5\u8be2\u754c\u9650\uff0c\u53ef\u80fd\u5728\u5bf9\u6297\u6027\u6216\u957f\u5c3e\u5206\u5e03\u4e0b\u51fa\u73b0\u6027\u80fd\u9000\u5316\uff1b\u76f8\u53cd\uff0c\u63d0\u4f9b\u4e25\u683c\u7406\u8bba\u754c\u9650\u7684PGM\u867d\u4fdd\u8bc1\u4e86\u67e5\u8be2\u590d\u6742\u5ea6\uff0c\u5374\u5728\u5e73\u5747\u67e5\u8be2\u541e\u5410\u91cf\u4e0a\u843d\u540e\u4e8eRMI\u3002\u8fd9\u79cd\u201c\u8981\u4e48\u5feb\u4f46\u65e0\u754c\u9650\u3001\u8981\u4e48\u6709\u754c\u9650\u4f46\u6162\u201d\u7684\u56f0\u5883\uff0c\u4f7f\u5f97\u5b66\u4e60\u7d22\u5f15\u5728\u4e0e\u7ecf\u5178B\u6811\u7b49\u7ed3\u6784\u7ade\u4e89\u65f6\u59cb\u7ec8\u96be\u4ee5\u540c\u65f6\u517c\u987e\u7406\u8bba\u4fdd\u8bc1\u4e0e\u5b9e\u9645\u6027\u80fd\u3002<\/p>\n\n\n\n<p>\u8bba\u6587\u9996\u5148\u901a\u8fc7SOSD-osmc\u6570\u636e\u96c6\u4e0a\u7684\u5bf9\u6bd4\u5b9e\u9a8c\u6e05\u6670\u5c55\u793a\u4e86\u8fd9\u4e00\u6027\u80fd\u5dee\u8ddd\uff1aRMI\u5728\u67e5\u8be2\u5ef6\u8fdf\u4e0a\u663e\u8457\u4f18\u4e8ePGM\uff0c\u4f46PGM\u7684\u7d22\u5f15\u9ad8\u5ea6\u66f4\u9ad8\uff0c\u5bfc\u81f4\u6700\u574f\u60c5\u51b5\u67e5\u8be2\u590d\u6742\u5ea6\u8f83\u5dee\u3002\u4f5c\u8005\u8fdb\u4e00\u6b65\u6307\u51fa\uff0c\u9020\u6210\u8fd9\u4e00\u5dee\u8ddd\u7684\u6838\u5fc3\u539f\u56e0\u5728\u4e8e\u4f20\u7edf\u5b66\u4e60\u7d22\u5f15\u91c7\u7528\u7684\u201cper-key learning granularity\u201d\uff1a\u6bcf\u4e2a\u952e\u72ec\u7acb\u53c2\u4e0e\u6a21\u578b\u8bad\u7ec3\uff0c\u5bfc\u81f4\u6a21\u578b\u6570\u91cf\u8fc7\u591a\u3001\u7d22\u5f15\u9ad8\u5ea6\u8fc7\u9ad8\uff0c\u540c\u65f6\u771f\u5b9e\u4e16\u754c\u5206\u5e03\u7684\u975e\u5747\u5300\u6027\u548c\u590d\u6742\u6027\u8fdb\u4e00\u6b65\u653e\u5927\u4e86\u8fd9\u4e00\u95ee\u9898\u3002<\/p>\n\n\n\n<p>VEGA\u7684\u521b\u65b0\u70b9\u5728\u4e8e\u9996\u6b21\u63d0\u51fa\u201cgroup-wise learning granularity\u201d\u4e0e\u201cactive distribution tuning\u201d\u4e24\u5927\u7b56\u7565\u3002\u524d\u8005\u901a\u8fc7\u5c06\u8fde\u7eed\u952e\u6253\u5305\u6210\u56fa\u5b9a\u5927\u5c0f\u7684\u7ec4\uff0c\u4ec5\u7528\u7ec4\u5185\u9996\u952e\u6784\u5efa\u6a21\u578b\uff0c\u5b9e\u73b0\u4e86\u6a21\u578b\u6570\u91cf\u4eceO(n)\u5230O(n\/k)\u7684\u964d\u4f4e\uff1b\u540e\u8005\u5219\u901a\u8fc7\u952e\u91cd\u5b9a\u4f4d\u4e3b\u52a8\u8c03\u4f18\u7b80\u5316\u540e\u7684\u5206\u5e03\uff0c\u4f7f\u5355\u5c42\u6a21\u578b\u6570\u91cf\u8fdb\u4e00\u6b65\u964d\u81f31\uff0c\u540c\u65f6\u5c06\u7a7a\u95f4\u5f00\u9500\u63a7\u5236\u5728\u53ef\u63a5\u53d7\u8303\u56f4\u5185\u3002\u4e24\u8005\u7ed3\u5408\u7684\u901a\u7528\u6846\u67b6\u5728\u7ed9\u5b9a\u5185\u5b58\u9884\u7b97\u4e0b\u52a8\u6001\u4f18\u5316\u5206\u533a\u7b56\u7565\uff0c\u786e\u4fddVEGA\u5728\u7406\u8bba\u67e5\u8be2\u590d\u6742\u5ea6\u4e0a\u8fbe\u5230O(log(n\/3k) + log k)\uff0c\u5728\u7ecf\u9a8c\u6027\u80fd\u4e0a\u8d85\u8d8a\u73b0\u6709\u6700\u4f18\u65b9\u6cd5\u3002\u8be5\u5de5\u4f5c\u4e0d\u4ec5\u4e3a\u53ea\u8bfb\u5b66\u4e60\u7d22\u5f15\u63d0\u4f9b\u4e86\u7406\u8bba\u4e0e\u5b9e\u8df5\u7edf\u4e00\u7684\u89e3\u51b3\u65b9\u6848\uff0c\u4e5f\u4e3a\u540e\u7eed\u52a8\u6001\u5b66\u4e60\u7d22\u5f15\u7684\u8bbe\u8ba1\u5960\u5b9a\u4e86\u91cd\u8981\u57fa\u7840\u3002<\/p>\n\n\n\n<p><strong>3. \u80cc\u666f\u4e0e\u52a8\u673a\uff08Background and Motivation\uff09<\/strong><\/p>\n\n\n\n<p>\u5b66\u4e60\u7d22\u5f15\u53ef\u5206\u4e3a\u53ef\u53d8\u578b\uff08mutable\uff09\u548c\u4e0d\u53d8\u578b\uff08immutable\uff09\u4e24\u5927\u7c7b\u3002\u53ef\u53d8\u578b\u7d22\u5f15\uff08\u5982ALEX\u3001LIPP\uff09\u91cd\u70b9\u652f\u6301\u52a8\u6001\u66f4\u65b0\uff0c\u727a\u7272\u7a7a\u95f4\u6548\u7387\u6362\u53d6\u64cd\u4f5c\u541e\u5410\u91cf\uff1b\u4e0d\u53d8\u578b\u7d22\u5f15\u5219\u4e13\u6ce8\u4e8e\u53ea\u8bfb\u573a\u666f\uff0c\u901a\u8fc7\u6781\u81f4\u538b\u7f29\u7a7a\u95f4\u548c\u4f18\u5316\u67e5\u8be2\u8def\u5f84\u5b9e\u73b0\u9ad8\u6027\u80fd\uff0c\u5e7f\u6cdb\u5e94\u7528\u4e8e\u952e\u503c\u5b58\u50a8\u3001\u7f51\u7edc\u5305\u5206\u7c7b\u3001\u533a\u5757\u94fe\u67e5\u8be2\u7b49\u8bfb\u5bc6\u96c6\u578b\u8d1f\u8f7d\u3002RMI\u4f5c\u4e3a\u4e0d\u53d8\u578b\u7d22\u5f15\u7684\u4ee3\u8868\uff0c\u91c7\u7528\u81ea\u9876\u5411\u4e0b\u7684\u9012\u5f52\u5212\u5206\u7b56\u7565\uff0c\u5728\u7ecf\u9a8c\u6027\u80fd\u4e0a\u9886\u5148\uff0c\u4f46\u7406\u8bba\u4e0a\u7f3a\u4e4f\u975e\u5e73\u51e1\u6700\u574f\u60c5\u51b5\u754c\u9650\uff1bPGM\u5219\u91c7\u7528\u81ea\u5e95\u5411\u4e0a\u7684\u5206\u6bb5\u7ebf\u6027\u903c\u8fd1\uff0c\u63d0\u4f9b\u4e86O(log M_pla-1 + log k)\u7684\u67e5\u8be2\u590d\u6742\u5ea6\uff0c\u5374\u56e0\u6a21\u578b\u6570\u91cf\u8fc7\u591a\u5bfc\u81f4\u5b9e\u9645\u67e5\u8be2\u5ef6\u8fdf\u8f83\u9ad8\u3002<\/p>\n\n\n\n<p>\u73b0\u6709\u65b9\u6cd5\u7684\u5c40\u9650\u4e3b\u8981\u4f53\u73b0\u5728\u4e24\u4e2a\u65b9\u9762\uff1a\u4e00\u662fper-key\u5b66\u4e60\u7c92\u5ea6\u5bfc\u81f4\u7d22\u5f15\u9ad8\u5ea6\u8fc7\u9ad8\uff0c\u67e5\u8be2\u8def\u5f84\u53d8\u957f\uff1b\u4e8c\u662f\u771f\u5b9e\u4e16\u754c\u5206\u5e03\u7684\u590d\u6742\u6027\u4f7f\u5355\u7eaf\u653e\u677e\u7c92\u5ea6\u96be\u4ee5\u6709\u6548\u51cf\u5c11\u6a21\u578b\u6570\u91cf\u3002\u8bba\u6587\u901a\u8fc7SOSD\u57fa\u51c6\u6570\u636e\u96c6\u7684\u5b9e\u9a8c\u8fdb\u4e00\u6b65\u8bc1\u5b9e\uff0cPGM\u5728\u6700\u96be\u6570\u636e\u96c6\u4e0a\u7684\u7d22\u5f15\u9ad8\u5ea6\u53ef\u8fbe6\u5c42\uff0c\u800cRMI\u4ec5\u4e3a2\u5c42\uff0c\u76f4\u63a5\u5bfc\u81f4\u67e5\u8be2\u6027\u80fd\u5dee\u5f02\u3002VEGA\u7684\u52a8\u673a\u6b63\u662f\u586b\u8865\u8fd9\u4e00\u7406\u8bba-\u5b9e\u8df5\u9e3f\u6c9f\uff0c\u901a\u8fc7group-wise\u7c92\u5ea6\u7b80\u5316\u5206\u5e03\u3001active tuning\u8c03\u4f18\u5206\u5e03\uff0c\u4ee5\u53caF1\/F2\u7b56\u7565\u7684\u52a8\u6001\u7ec4\u5408\uff0c\u5728\u4e25\u683c\u7406\u8bba\u754c\u9650\u4e0b\u5b9e\u73b0\u6700\u4f18\u7ecf\u9a8c\u6027\u80fd\u3002\u8be5\u8bbe\u8ba1\u4e0d\u4ec5\u89e3\u51b3\u4e86\u73b0\u6709\u7d22\u5f15\u7684\u6839\u672c\u77db\u76fe\uff0c\u4e5f\u4e3a\u5b66\u4e60\u7d22\u5f15\u5728\u5b9e\u9645\u7cfb\u7edf\u4e2d\u7684\u5927\u89c4\u6a21\u90e8\u7f72\u63d0\u4f9b\u4e86\u53ef\u884c\u8def\u5f84\u3002<\/p>\n\n\n\n<p><strong>4. VEGA\u7cfb\u7edf\u67b6\u6784\u4e0e\u9ad8\u5c42\u8bbe\u8ba1\uff08VEGA Architecture and High-level Idea\uff09<\/strong><\/p>\n\n\n\n<p>VEGA\u91c7\u7528\u591a\u5c42\u6876\u6570\u7ec4\u7ed3\u6784\uff0c\u6bcf\u5c42\u7531\u56fa\u5b9a\u5927\u5c0f\u7684\u6876\u7ec4\u6210\uff0c\u5e95\u5c42\u4e3a\u53f6\u6570\u636e\u5c42\uff08leaf data layer\uff09\uff0c\u4e0a\u5c42\u4e3a\u975e\u53f6\u7d22\u5f15\u5c42\uff08non-leaf index layer\uff09\u3002\u6bcf\u4e2a\u6876\u5927\u5c0f\u7b49\u4e8e\u5f53\u524d\u5c42\u7684\u5b66\u4e60\u7c92\u5ea6k\uff0c\u6876\u5185\u952e\u6309SIMD\u5bbd\u5ea6\u5bf9\u9f50\u4ee5\u652f\u6301\u5411\u91cf\u641c\u7d22\u3002VEGA\u7684\u6784\u5efa\u91c7\u7528\u81ea\u5e95\u5411\u4e0a\u7684\u9012\u5f52\u65b9\u5f0f\uff0c\u6bcf\u5c42\u901a\u8fc7F1\u6216F2\u7b56\u7565\u6784\u5efa\u8bef\u5dee\u6709\u754c\u7ebf\u6027\u6a21\u578b\uff0c\u7528\u4e8e\u9884\u6d4b\u4e0b\u4e00\u5c42\u7684\u6876\u4f4d\u7f6e\u3002<\/p>\n\n\n\n<p>\u9ad8\u5c42\u8bbe\u8ba1\u7684\u6838\u5fc3\u5728\u4e8e\u7d27\u51d1\u6876\uff08compact bucket\uff09\u4e0e\u7a00\u758f\u6876\uff08sparse bucket\uff09\u7684\u6df7\u5408\u4f7f\u7528\uff1aF1\u7b56\u7565\u4e0b\u6bcf\u4e2a\u7ec4\u7cbe\u786e\u586b\u5145\u4e00\u4e2a\u7d27\u51d1\u6876\uff0c\u65e0\u7a7a\u69fd\uff1bF2\u7b56\u7565\u4e0b\u901a\u8fc7\u952e\u91cd\u5b9a\u4f4d\u5c06\u7ec4\u6620\u5c04\u5230\u66f4\u5927\u6876\u6570\u7ec4\uff0c\u4ea7\u751f\u7a00\u758f\u6876\u5e76\u5229\u7528\u7a7a\u69fd\u5b9e\u73b0\u5feb\u901f\u8def\u5f84\u3002\u5feb\u901f\u8def\u5f84\u673a\u5236\u901a\u8fc7\u76f4\u63a5\u590d\u5236\u4e0b\u5c42\u6a21\u578b\u5230\u7a00\u758f\u6876\u7a7a\u69fd\uff0c\u53ef\u7ed5\u8fc7\u4e2d\u95f4\u5c42\uff0c\u8fdb\u4e00\u6b65\u7f29\u77ed\u67e5\u8be2\u8def\u5f84\u3002\u5411\u91cf\u641c\u7d22\u90e8\u5206\u91c7\u7528LS-SIMD\uff08\u5c0f\u6876\uff09\u6216BS-LS-SIMD\uff08\u5927\u6876\uff09\u6df7\u5408\u7b56\u7565\uff0c\u6709\u6548\u8865\u507f\u56e0\u7c92\u5ea6\u653e\u5bbd\u5e26\u6765\u7684\u9884\u6d4b\u8bef\u5dee\u3002<\/p>\n\n\n\n<p>\u8fd9\u79cd\u67b6\u6784\u65e2\u4fdd\u8bc1\u4e86\u7d27\u81f4\u7a7a\u95f4\u5360\u7528\uff08\u5c0f\u4e8e\u539f\u59cb\u6570\u636e\u96c6\u76843%\uff09\uff0c\u53c8\u901a\u8fc7\u52a8\u6001\u7ec4\u5408F1\/F2\u5b9e\u73b0\u4e86\u7d22\u5f15\u9ad8\u5ea6\u7684\u6700\u5c0f\u5316\uff0c\u4e3aVEGA\u5728\u7406\u8bba\u754c\u9650\u4e0e\u7ecf\u9a8c\u6027\u80fd\u4e0a\u7684\u53cc\u91cd\u4f18\u52bf\u5960\u5b9a\u4e86\u57fa\u7840\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"792\" height=\"267\" src=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/4-1.png\"  class=\"wp-image-1229\" srcset=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/4-1.png 792w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/4-1-300x101.png 300w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/4-1-768x259.png 768w\" sizes=\"auto, (max-width: 792px) 100vw, 792px\" title=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe3\" alt=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe3\" \/><\/figure>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"807\" height=\"282\" src=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/5.png\"  class=\"wp-image-1230\" srcset=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/5.png 807w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/5-300x105.png 300w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/5-768x268.png 768w\" sizes=\"auto, (max-width: 807px) 100vw, 807px\" title=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe4\" alt=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe4\" \/><\/figure>\n\n\n\n<p><strong>5. F1\u4e0eF2\u6a21\u578b\u6784\u5efa\u7b56\u7565\uff08F1 and F2 Model Building Policies\uff09<\/strong><\/p>\n\n\n\n<p>VEGA\u7684\u6838\u5fc3\u521b\u65b0\u5728\u4e8eF1\u548cF2\u4e24\u5927\u5728\u7ebf\u6a21\u578b\u6784\u5efa\u7b56\u7565\u3002F1\uff08Fitting model to simplified distribution\uff09\u91c7\u7528group-wise\u7c92\u5ea6\uff0c\u5c06\u8fde\u7eedk\u4e2a\u952e\u6253\u5305\u6210\u7ec4\uff0c\u4ec5\u7528\u7ec4\u5185\u9996\u952e\u53c2\u4e0e\u6a21\u578b\u8bad\u7ec3\u3002\u901a\u8fc7\u6269\u5c55\u8bef\u5dee\u6709\u754c\u5206\u6bb5\u7ebf\u6027\u903c\u8fd1\uff08PLA\uff09\uff0cF1\u8bc1\u660e\u7ec4\u5185\u975e\u9996\u952e\u7684\u9884\u6d4b\u8bef\u5dee\u4ecd\u53ef\u88ab\u754c\u9650\uff08\u653e\u5bbd\u81f32\u00b7k\uff09\uff0c\u4ece\u800c\u5c06\u6a21\u578b\u6570\u91cf\u4eceO(n)\u964d\u4f4e\u81f3O(n\/3k)\uff0c\u6784\u5efa\u590d\u6742\u5ea6\u964d\u81f3O(n\/k)\u3002\u540c\u65f6\uff0cF1\u751f\u6210\u7684\u7d27\u51d1\u6876\u7cbe\u786e\u586b\u5145\uff0c\u65e0\u7a7a\u69fd\u6d6a\u8d39\u3002<\/p>\n\n\n\n<p>F2\uff08Fitting tuned distribution to model\uff09\u5219\u8fdb\u4e00\u6b65\u8c03\u4f18\u5206\u5e03\uff1a\u901a\u8fc7\u952e\u91cd\u5b9a\u4f4d\u5c06\u7b80\u5316\u540e\u7684\u7ec4\u6620\u5c04\u5230\u66f4\u5927\u6876\u6570\u7ec4\uff0c\u4f7f\u5355\u5c42\u4ec5\u97001\u4e2a\u7ebf\u6027\u6a21\u578b\u3002F2\u5c06\u91cd\u5b9a\u4f4d\u95ee\u9898\u5f62\u5f0f\u5316\u4e3a\u4f18\u5316\u95ee\u9898\uff0c\u4ec5\u7528\u7ec4\u5185\u9996\u952e\u6784\u5efa\u6a21\u578b\uff0c\u8bc1\u660e\u6700\u4f18\u659c\u7387\u548c\u622a\u8ddd\u53ef\u5728\u7ebf\u6027\u65f6\u95f4\u5185\u6c42\u89e3\uff0c\u6784\u5efa\u590d\u6742\u5ea6\u540c\u6837\u4e3aO(n\/k)\u3002\u751f\u6210\u7684\u7a00\u758f\u6876\u867d\u5f15\u5165\u7a7a\u69fd\uff0c\u4f46\u901a\u8fc7\u5feb\u901f\u8def\u5f84\u673a\u5236\u53ef\u6709\u6548\u5229\u7528\u3002<\/p>\n\n\n\n<p>\u4e24\u8005\u7ed3\u5408\u7684\u901a\u7528\u6846\u67b6\u5728\u7ed9\u5b9a\u5185\u5b58\u9884\u7b97\u4e0b\u91c7\u7528\u8d2a\u5fc3\u6216\u52a8\u6001\u89c4\u5212\u7b97\u6cd5\u52a8\u6001\u9009\u62e9\u5206\u533a\u7b56\u7565\uff0c\u5b9e\u73b0\u6a21\u578b\u6570\u91cf\u7684\u6700\u5c0f\u5316\u3002\u8be5\u7b56\u7565\u65e2\u4fdd\u8bc1\u4e86\u8bef\u5dee\u754c\u9650\uff0c\u53c8\u663e\u8457\u964d\u4f4e\u4e86\u7d22\u5f15\u9ad8\u5ea6\uff0c\u662fVEGA\u5b9e\u73b0\u7406\u8bba\u4e0e\u5b9e\u8df5\u7edf\u4e00\u7684\u5173\u952e\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"819\" height=\"333\" src=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/7.png\"  class=\"wp-image-1231\" srcset=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/7.png 819w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/7-300x122.png 300w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/7-768x312.png 768w\" sizes=\"auto, (max-width: 819px) 100vw, 819px\" title=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe5\" alt=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe5\" \/><\/figure>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"834\" height=\"366\" src=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/8.png\"  class=\"wp-image-1232\" srcset=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/8.png 834w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/8-300x132.png 300w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/8-768x337.png 768w\" sizes=\"auto, (max-width: 834px) 100vw, 834px\" title=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe6\" alt=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe6\" \/><\/figure>\n\n\n\n<p><strong>6. \u6784\u5efa\u4e0e\u67e5\u8be2\u6d41\u7a0b\uff08Construction and Query Procedures\uff09<\/strong><\/p>\n\n\n\n<p>VEGA\u7684\u5355\u5c42\u6784\u5efa\u91c7\u7528FastConstructOneLayer\u7b97\u6cd5\uff1a\u4ee5\u6876\u6bcf\u952e\u6570B\u4e3a\u53c2\u6570\uff0c\u8d2a\u5fc3\u5224\u65ad\u5f53\u524d\u7ec4\u662f\u5426\u53ef\u7528F1\u6216F2\u5bb9\u7eb3\uff0c\u4f18\u5148\u4f7f\u7528F2\u4ee5\u63d0\u5347\u67e5\u8be2\u6548\u7387\uff0c\u540c\u65f6\u4e25\u683c\u63a7\u5236\u7a7a\u95f4\u5f00\u9500\u3002\u6574\u4f53\u6784\u5efa\u901a\u8fc7\u9012\u5f52\u8c03\u7528\u8be5\u7b97\u6cd5\u5b8c\u6210\u591a\u5c42\u7d22\u5f15\uff0c\u590d\u6742\u5ea6\u4e3aO(n log n)\u3002<\/p>\n\n\n\n<p>\u67e5\u8be2\u6d41\u7a0b\u5206\u4e3a\u70b9\u67e5\u8be2\u548c\u8303\u56f4\u67e5\u8be2\u3002\u70b9\u67e5\u8be2\u4ece\u9876\u5c42\u5f00\u59cb\uff0c\u5229\u7528\u6a21\u578b\u9884\u6d4b\u4e0b\u4e00\u5c42\u6876\u4f4d\u7f6e\uff0c\u7ed3\u5408\u5411\u91cf\u641c\u7d22\u5b9a\u4f4d\u524d\u9a71\u952e\uff0c\u5feb\u901f\u8def\u5f84\u53ef\u8df3\u8fc7\u4e2d\u95f4\u5c42\uff1b\u8303\u56f4\u67e5\u8be2\u5148\u901a\u8fc7\u70b9\u67e5\u8be2\u5b9a\u4f4d\u8d77\u70b9\u6876\uff0c\u518d\u987a\u5e8f\u626b\u63cf\u3002\u5411\u91cf\u641c\u7d22\u91c7\u7528LS-SIMD\u4e0eBS-LS-SIMD\u6df7\u5408\u7b56\u7565\uff0c\u5145\u5206\u5229\u7528SIMD\u6307\u4ee4\u6709\u6548\u8865\u507f\u9884\u6d4b\u8bef\u5dee\u3002\u5feb\u901f\u8def\u5f84\u901a\u8fc7\u504f\u79fb\u6807\u8bb0\u533a\u5206\u6b63\u5e38\u8def\u5f84\u4e0e\u5feb\u901f\u8def\u5f84\uff0c\u65e0\u9700\u989d\u5916\u6807\u5fd7\u4f4d\u3002<\/p>\n\n\n\n<p>\u8fd9\u4e9b\u8bbe\u8ba1\u786e\u4fdd\u4e86VEGA\u5728\u5b9e\u9645\u67e5\u8be2\u4e2d\u7684\u9ad8\u6548\u6027\u548c\u9c81\u68d2\u6027\uff0c\u540c\u65f6\u4fdd\u6301\u4e86\u4e25\u683c\u7684\u7406\u8bba\u590d\u6742\u5ea6\u4fdd\u8bc1\u3002<\/p>\n\n\n\n<p><strong>7. \u5b9e\u9a8c\u8bc4\u4f30\u4e0e\u6027\u80fd\u5206\u6790\uff08Evaluation\uff09<\/strong><\/p>\n\n\n\n<p>\u5b9e\u9a8c\u5728Intel Xeon Gold 6226R CPU + 32GB RAM\u73af\u5883\u4e0b\u8fdb\u884c\uff0c\u4f7f\u7528SOSD\u548cupdatable learned index\u57fa\u51c6\u4e2d\u6700\u96be\u7684\u56db\u4e2a\u771f\u5b9e\u6570\u636e\u96c6\uff08amzn\u3001face\u3001osmc\u3001genome\uff09\uff0c\u5e76\u6269\u5c55\u81f3\u66f4\u591a\u8865\u5145\u6570\u636e\u96c6\u3001\u91cd\u590d\u6570\u636e\u96c6\u548c\u504f\u659c\u5206\u5e03\u6570\u636e\u96c6\u3002\u57fa\u7ebf\u5305\u62ecRMI\u3001PGM\u3001RS\u3001ALEX\u3001LIPP\u3001NFL\u3001DILI\u3001FAST\u3001ART\u3001BTree\u7b4910\u79cd\u6700\u5148\u8fdb\u7d22\u5f15\u3002<\/p>\n\n\n\n<p>\u70b9\u67e5\u8be2\u7ed3\u679c\u663e\u793a\uff0cVEGA\u5728\u5355\u7ebf\u7a0b\u548c\u591a\u7ebf\u7a0b\u4e0b\u5747\u53d6\u5f97\u6700\u9ad8\u541e\u5410\u91cf\uff0c\u76f8\u6bd4RMI\u5e73\u5747\u63d0\u53471.5\u500d\uff0c\u76f8\u6bd4PGM\u63d0\u53473\u500d\u4ee5\u4e0a\uff1b\u8303\u56f4\u67e5\u8be2\u541e\u5410\u91cf\u540c\u6837\u9886\u5148\u3002\u6784\u5efa\u65f6\u95f4\u4e0a\uff0cVEGA\u4e0e\u6700\u5feb\u7684RS\u76f8\u5f53\uff0c\u8fdc\u4f18\u4e8eRMI\u548cPGM\u3002\u53c2\u6570\u8bc4\u4f30\u8bc1\u5b9e\u7c92\u5ea6k=8\u65f6\u6027\u80fd\u6700\u4f18\uff0c\u5411\u91cf\u6307\u4ee4\uff08AVX-512\uff09\u5e26\u6765\u663e\u8457\u5ef6\u8fdf\u964d\u4f4e\u3002\u5feb\u901f\u8def\u5f84\u8bc4\u4f30\u663e\u793a\uff0c\u5f53\u5feb\u901f\u8def\u5f84\u6bd4\u4f8b\u8fbe80%\u65f6\uff0c\u67e5\u8be2\u5ef6\u8fdf\u53ef\u964d\u4f4e30%\u3002<\/p>\n\n\n\n<p>\u6d88\u878d\u5b9e\u9a8c\u548c\u66f4\u5e7f\u6cdb\u6570\u636e\u96c6\u9a8c\u8bc1\u8fdb\u4e00\u6b65\u8bc1\u5b9eVEGA\u5728\u4e0d\u540c\u5206\u5e03\u4e0b\u7684\u9c81\u68d2\u6027\uff0c\u5373\u4f7f\u5728\u5185\u5b58\u6805\u680f\u7ea6\u675f\u4e0b\u4ecd\u4fdd\u6301\u9886\u5148\u3002\u8be5\u7ed3\u679c\u5145\u5206\u8bc1\u660e\u4e86VEGA\u5728\u7406\u8bba\u754c\u9650\u4e0e\u7ecf\u9a8c\u6027\u80fd\u4e0a\u7684\u53cc\u91cd\u4f18\u52bf\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"852\" height=\"336\" src=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/9.png\"  class=\"wp-image-1233\" srcset=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/9.png 852w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/9-300x118.png 300w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/9-768x303.png 768w\" sizes=\"auto, (max-width: 852px) 100vw, 852px\" title=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe7\" alt=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe7\" \/><\/figure>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"789\" height=\"266\" src=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/10.png\"  class=\"wp-image-1234\" srcset=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/10.png 789w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/10-300x101.png 300w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/10-768x259.png 768w\" sizes=\"auto, (max-width: 789px) 100vw, 789px\" title=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe8\" alt=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe8\" \/><\/figure>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"840\" height=\"429\" src=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/11.png\"  class=\"wp-image-1235\" srcset=\"https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/11.png 840w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/11-300x153.png 300w, https:\/\/www.ndnlab.com\/wp-content\/uploads\/2026\/04\/11-768x392.png 768w\" sizes=\"auto, (max-width: 840px) 100vw, 840px\" title=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe9\" alt=\"VEGA: An Active-tuning Learned Index with Group-Wise Learning Granularity\u63d2\u56fe9\" \/><\/figure>\n\n\n\n<p><strong>8. \u8d21\u732e\u4e0e\u7ed3\u8bba\uff08Contributions and Conclusion\uff09<\/strong><\/p>\n\n\n\n<p>VEGA\u7684\u4e3b\u8981\u8d21\u732e\u5305\u62ec\uff1a\uff081\uff09\u63d0\u51fagroup-wise\u5b66\u4e60\u7c92\u5ea6\u8303\u5f0f\uff0c\u5b9e\u73b0\u65e0\u9700\u9010\u952e\u8bbf\u95ee\u7684\u6a21\u578b\u6784\u5efa\uff0c\u5c06\u6784\u5efa\u590d\u6742\u5ea6\u964d\u81f3O(n\/k)\uff1b\uff082\uff09\u8bbe\u8ba1\u8f7b\u91cf\u7ea7\u5206\u5e03\u8c03\u4f18\u65b9\u6cd5F2\uff0c\u5728\u6700\u5c0f\u7a7a\u95f4\u5f00\u9500\u4e0b\u5c06\u6a21\u578b\u6570\u91cf\u8fdb\u4e00\u6b65\u964d\u81f31\uff1b\uff083\uff09\u63d0\u51faF1\/F2\u52a8\u6001\u7ec4\u5408\u6846\u67b6\uff0c\u5728\u7ed9\u5b9a\u5185\u5b58\u9884\u7b97\u4e0b\u5b9e\u73b0\u67e5\u8be2\u6027\u80fd\u6700\u4f18\uff1b\uff084\uff09\u901a\u8fc7\u7406\u8bba\u5206\u6790\u8bc1\u660e\u67e5\u8be2\u590d\u6742\u5ea6\u8fbe\u5230\u6216\u4f18\u4e8e\u73b0\u6709\u6700\u4f18\u65b9\u6cd5\uff1b\uff085\uff09\u5927\u89c4\u6a21\u5b9e\u9a8c\u9a8c\u8bc1VEGA\u5728\u67e5\u8be2\u548c\u6784\u5efa\u6027\u80fd\u4e0a\u7684\u5168\u9762\u9886\u5148\u3002<\/p>\n\n\n\n<p>\u8bba\u6587\u7ed3\u8bba\u6307\u51fa\uff0cVEGA\u6210\u529f\u6865\u63a5\u4e86\u5b66\u4e60\u7d22\u5f15\u7406\u8bba\u754c\u9650\u4e0e\u7ecf\u9a8c\u6027\u80fd\u4e4b\u95f4\u7684\u957f\u671f\u9e3f\u6c9f\uff0c\u4e3a\u53ea\u8bfb\u578b\u5b66\u4e60\u7d22\u5f15\u63d0\u4f9b\u4e86\u5b8c\u6574\u3001\u53ef\u7406\u8bba\u4fdd\u8bc1\u7684\u7cfb\u7edf\u89e3\u51b3\u65b9\u6848\u3002\u5c3d\u7ba1\u5f53\u524d\u4e3b\u8981\u9488\u5bf9\u53ea\u8bfb\u573a\u666f\uff0c\u4f46VEGA\u7684\u6846\u67b6\u4e3a\u672a\u6765\u589e\u91cf\u66f4\u65b0\u7b49\u6269\u5c55\u63d0\u4f9b\u4e86\u575a\u5b9e\u57fa\u7840\u3002\u8be5\u5de5\u4f5c\u4e0d\u4ec5\u5177\u6709\u91cd\u8981\u7684\u5b66\u672f\u4ef7\u503c\uff0c\u4e5f\u4e3a\u952e\u503c\u5b58\u50a8\u3001\u7f51\u7edc\u5206\u7c7b\u7b49\u5b9e\u9645\u7cfb\u7edf\u4e2d\u7684\u5b66\u4e60\u7d22\u5f15\u90e8\u7f72\u63d0\u4f9b\u4e86\u53ef\u843d\u5730\u7684\u6280\u672f\u53c2\u8003\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>1. \u6458\u8981\uff08Abstract\uff09 VEGA\u63d0\u51fa\u4e86\u4e00\u79cd\u5168\u65b0\u7684\u53ea\u8bfb\u578b\uff08immutable\uff09\u5b66\u4e60\u7d22\u5f15\uff0c\u65e8\u5728\u540c\u65f6\u5b9e\u73b0\u6700\u5148\u8fdb\u7684\u7ecf\u9a8c\u67e5\u8be2\u6027\u80fd\u548c\u4e25\u683c\u7684\u7406\u8bba\u67e5\u8be2\u590d\u6742\u5ea6\u4fdd\u8bc1\u3002\u73b0\u6709\u6700\u5feb\u7684\u4e0d\u53d8\u5b66\u4e60\u7d22\u5f15\uff08\u5982RMI\uff09\u867d\u5728\u5b9e\u9645\u67e5\u8be2\u541e\u5410\u91cf\u4e0a\u8868\u73b0\u7a81\u51fa\uff0c\u4f46\u7f3a\u4e4f\u975e\u5e73\u51e1\u7684\u6700\u574f\u60c5\u51b5\u67e5\u8be2\u754c\u9650\uff1b\u76f8\u53cd\uff0c\u63d0\u4f9b\u7d27\u81f4\u754c\u9650\u7684\u7d22\u5f15\uff08\u5982PGM\uff09\u867d\u7406\u8bba\u4e0a\u53ef\u9760\uff0c\u5374\u5728\u5e73\u5747\u67e5\u8be2\u6027\u80fd\u4e0a\u663e\u8457\u843d\u540e\u3002\u8fd9\u79cd\u7406\u8bba\u4e0e\u5b9e\u8df5\u7684\u8131\u8282\u6210\u4e3a\u5b66\u4e60\u7d22\u5f15\u9886\u57df\u957f\u671f\u60ac\u800c\u672a\u51b3\u7684\u6838\u5fc3\u95ee\u9898\u3002VEGA\u901a\u8fc7\u5f15\u5165\u4e24\u5927\u5728\u7ebf\u6a21\u578b\u6784\u5efa\u7b56\u7565\u5f7b\u5e95\u89e3\u51b3\u4e86\u8fd9\u4e00\u96be\u9898\uff1a\u4e00\u662f\u91c7\u7528\u5408\u9002\u7684\u7c92\u5ea6\uff08group-wi &hellip; <a href=\"https:\/\/www.ndnlab.com\/?p=1224\">\u7ee7\u7eed\u9605\u8bfb <span class=\"meta-nav\">&rarr;<\/span><\/a><\/p>\n","protected":false},"author":5,"featured_media":1225,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5,1],"tags":[14],"class_list":["post-1224","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-rengongzhineng","category-uncategorized","tag-ien"],"_links":{"self":[{"href":"https:\/\/www.ndnlab.com\/index.php?rest_route=\/wp\/v2\/posts\/1224","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.ndnlab.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.ndnlab.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.ndnlab.com\/index.php?rest_route=\/wp\/v2\/users\/5"}],"replies":[{"embeddable":true,"href":"https:\/\/www.ndnlab.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=1224"}],"version-history":[{"count":1,"href":"https:\/\/www.ndnlab.com\/index.php?rest_route=\/wp\/v2\/posts\/1224\/revisions"}],"predecessor-version":[{"id":1236,"href":"https:\/\/www.ndnlab.com\/index.php?rest_route=\/wp\/v2\/posts\/1224\/revisions\/1236"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.ndnlab.com\/index.php?rest_route=\/wp\/v2\/media\/1225"}],"wp:attachment":[{"href":"https:\/\/www.ndnlab.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1224"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.ndnlab.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=1224"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.ndnlab.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=1224"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}