{"resolved":true,"match_type":"uuid","source_type":"original_book","book_id":"75afffda-b96b-4df5-89ff-2ec037cfed66","title":"AI 3.0（畅销书《复杂》作者、复杂系统前沿科学家梅拉妮·米歇尔，历经10年思考，厘清人工智能与人类智能的全新力作，全景式地展现了人工智能的...","author":null,"edition":null,"publisher":null,"year":null,"isbn":null,"requested_by":"printed_page","book":{"source_type":"original_book","book_id":"75afffda-b96b-4df5-89ff-2ec037cfed66","id":"75afffda-b96b-4df5-89ff-2ec037cfed66","uuid":"75afffda-b96b-4df5-89ff-2ec037cfed66","title":"AI 3.0（畅销书《复杂》作者、复杂系统前沿科学家梅拉妮·米歇尔，历经10年思考，厘清人工智能与人类智能的全新力作，全景式地展现了人工智能的...","normalized_title":"ai30畅销书复杂作者复杂系统前沿科学家梅拉妮米歇尔历经10年思考厘清人工智能与人类智能的全新力作全景式地展现了人工智能的","author":null,"edition":null,"publisher":null,"year":null,"isbn":null,"isbns":[],"parsed":true,"ai_readable":true,"document_type":"book","page_count":491,"navigation_url":"https://workshop.closeclaw.top/ai/book/75afffda-b96b-4df5-89ff-2ec037cfed66","title_url":"https://workshop.closeclaw.top/ai/title/AI%203.0%EF%BC%88%E7%95%85%E9%94%80%E4%B9%A6%E3%80%8A%E5%A4%8D%E6%9D%82%E3%80%8B%E4%BD%9C%E8%80%85%E3%80%81%E5%A4%8D%E6%9D%82%E7%B3%BB%E7%BB%9F%E5%89%8D%E6%B2%BF%E7%A7%91%E5%AD%A6%E5%AE%B6%E6%A2%85%E6%8B%89%E5%A6%AE%C2%B7%E7%B1%B3%E6%AD%87%E5%B0%94%EF%BC%8C%E5%8E%86%E7%BB%8F10%E5%B9%B4%E6%80%9D%E8%80%83%EF%BC%8C%E5%8E%98%E6%B8%85%E4%BA%BA%E5%B7%A5%E6%99%BA%E8%83%BD%E4%B8%8E%E4%BA%BA%E7%B1%BB%E6%99%BA%E8%83%BD%E7%9A%84%E5%85%A8%E6%96%B0%E5%8A%9B%E4%BD%9C%EF%BC%8C%E5%85%A8%E6%99%AF%E5%BC%8F%E5%9C%B0%E5%B1%95%E7%8E%B0%E4%BA%86%E4%BA%BA%E5%B7%A5%E6%99%BA%E8%83%BD%E7%9A%84...","fixed_page_url_template":"https://workshop.closeclaw.top/ai/book/75afffda-b96b-4df5-89ff-2ec037cfed66/page/{printed_page}","path_api_url_template":"https://workshop.closeclaw.top/api/ai/book/75afffda-b96b-4df5-89ff-2ec037cfed66/page/{printed_page}","read_url":"https://workshop.closeclaw.top/api/ai/read/75afffda-b96b-4df5-89ff-2ec037cfed66","page_url":"https://workshop.closeclaw.top/api/ai/books/75afffda-b96b-4df5-89ff-2ec037cfed66/page","search_url":"https://workshop.closeclaw.top/api/ai/books/75afffda-b96b-4df5-89ff-2ec037cfed66/search"},"page":{"printed_page":2,"pdf_page":91,"physical_index":91,"physical_index_base":0,"printed_page_status":"recognized","page_sha256":"79ca44a7c0925b7408c951c6a2f507023a8e25efdee5d7f8432c419b107fea6c"},"chapter":"[美","content":"从神经网络到机器学习<sub>，</sub>谁都不是最后的解药\n\n多层神经网络这一被明斯基和佩珀特所摒弃的、认为很有可能“不育的”带有扩展层的感知机，事实证明反而成了现代人工智能的基础。由于它是我在后续章节中描述的多种方法的基础，所以我在这里花一些篇幅来讲述这些网络是如何工作的。\n\n多层神经网络<sub>，</sub>识别编码中的简单特征\n\n网络是以多种方式相互连接的一组元素的集合。我们都对社交网络很熟悉，社交网络的元素是人，而计算机网络中的元素自然是计算机。在神经网络中，这些元素是模拟神经元，类似于我在前一章中描述的感知机。\n\n在图2-1中，我画了一个简单的多层神经网络的草图，这个多层神经网络用于识别手写数字。网络由两列（层）类似感知机的模拟神经元（图中圆圈）组成。为了简单起见，我将使用“单元”而非“模拟神经元”这个术语来描述网络中的元素。与第01章中检测数字8的感知机一样，该神经网络有\n\n02","previous_page":null,"next_page":null}