用人工智能(AI)办法发掘上市企业财务反常
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="data:image/svg+xml,%3C%3Fxml version=1.0 encoding=UTF-8%3F%3E%3Csvg width=1px height=1px viewBox=0 0 1 1 version=1.1 xmlns=http://www.w3.org/2000/svg xmlns:xlink=http://www.w3.org/1999/xlink%3E%3Ctitle%3E%3C/title%3E%3Cg stroke=none stroke-width=1 fill=none fill-rule=evenodd fill-opacity=0%3E%3Cg transform=translate(-249.000000, -126.000000) fill=%23FFFFFF%3E%3Crect x=249 y=126 width=1 height=1%3E%3C/rect%3E%3C/g%3E%3C/g%3E%3C/svg%3E" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">免责和版权声明：</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">本文的历史价格，事件，时间等未必准确。本文的图表数据仅用于静安笔记个人<span style="color: black;">科研</span>。静安笔记不对您的任何投资<span style="color: black;">行径</span>负责。每篇<span style="color: black;">文案</span>即使大致正确<span style="color: black;">亦</span>只能涵盖一个<span style="color: black;">方向</span>或<span style="color: black;">行业</span>，不<span style="color: black;">表率</span>符合您的投资情景。欢迎完整转载；单独<span style="color: black;">运用</span>本文图表数据请事先征得<span style="color: black;">自己</span>同意，并注明出处。</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">1. 静安笔记的学术背景</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">1996年北京的秋天美轮美奂。我从南方考到北京交通大学信息所读硕士，师从当时的所长袁老师。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">信息所聚焦学术<span style="color: black;">科研</span>：主动视觉，<span style="color: black;">设备</span>人，语音处理，自然语言学习是几个重点<span style="color: black;">科研</span><span style="color: black;">行业</span>。所里博士，博士后<span style="color: black;">亦</span>一大堆。袁老师百忙中和<span style="color: black;">咱们</span>见面，问我线性代数学的怎么样（我说不错），把矩阵论自己补一补，别急着“进实验室上机”。<span style="color: black;">【1】</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="data:image/svg+xml,%3C%3Fxml version=1.0 encoding=UTF-8%3F%3E%3Csvg width=1px height=1px viewBox=0 0 1 1 version=1.1 xmlns=http://www.w3.org/2000/svg xmlns:xlink=http://www.w3.org/1999/xlink%3E%3Ctitle%3E%3C/title%3E%3Cg stroke=none stroke-width=1 fill=none fill-rule=evenodd fill-opacity=0%3E%3Cg transform=translate(-249.000000, -126.000000) fill=%23FFFFFF%3E%3Crect x=249 y=126 width=1 height=1%3E%3C/rect%3E%3C/g%3E%3C/g%3E%3C/svg%3E" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">当时人工智能的学术气氛很浓，1997年LSTM<span style="color: black;">亦</span><span style="color: black;">已然</span>发明了<span style="color: black;">【2】</span>。<span style="color: black;">都数</span>硕士<span style="color: black;">朋友</span>还是热衷于更"HOT"的通信(ATM，以太网)和互联网计算机应用软件热潮。很少有人能用神经网络<span style="color: black;">处理</span>现实问题，并且<span style="color: black;">挣钱</span>。人工智能还不是一个<span style="color: black;">公众</span>词汇，我<span style="color: black;">亦</span>和人工智能擦肩而过。<span style="color: black;">【3】</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">我记得我就曾花费<span style="color: black;">海量</span>时间在学校和首都图书馆查阅三维电磁模型定位癫痫病灶算法，帮紫竹院<span style="color: black;">周边</span>一家民营医疗器械<span style="color: black;">机构</span>做<span style="color: black;">开发</span>。（质量<span style="color: black;">欠好</span>，有时候会把病灶定位在脑外）。</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">当时的高等教育一个是类似于“证书”<span style="color: black;">或</span>某种“信号”功能，<span style="color: black;">便是</span>你在某某大学硕士毕业<span style="color: black;">能够</span>找份每月发薪水的按部就班的工作。像<span style="color: black;">咱们</span>学校毕业想留北京<span style="color: black;">通常</span>都能有北京户口。还有一个<span style="color: black;">便是</span>当时的互联网和网上教育学术资源还<span style="color: black;">无</span><span style="color: black;">此刻</span>这么发达，高校还主导或垄断<span style="color: black;">有些</span>学术和<span style="color: black;">行业</span>知识技能，无论是师资还是硬件环境，<span style="color: black;">例如</span>设备，图书等等。<span style="color: black;">此刻</span>的高等教育<span style="color: black;">已然</span>走向了新时代的分岔口，<span style="color: black;">再也不</span>主导<span style="color: black;">或</span>垄断学习资源。Bryan Caplan<span style="color: black;">这般</span>问道：“Would you rather have a Princeton diploma without a Princeton
education, or a Princeton education without a Princeton diploma? If you
pause to answer, you must think signaling is pretty important.”</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">十年后的大学<span style="color: black;">亦</span>许与今天非常<span style="color: black;">区别</span>。</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="data:image/svg+xml,%3C%3Fxml version=1.0 encoding=UTF-8%3F%3E%3Csvg width=1px height=1px viewBox=0 0 1 1 version=1.1 xmlns=http://www.w3.org/2000/svg xmlns:xlink=http://www.w3.org/1999/xlink%3E%3Ctitle%3E%3C/title%3E%3Cg stroke=none stroke-width=1 fill=none fill-rule=evenodd fill-opacity=0%3E%3Cg transform=translate(-249.000000, -126.000000) fill=%23FFFFFF%3E%3Crect x=249 y=126 width=1 height=1%3E%3C/rect%3E%3C/g%3E%3C/g%3E%3C/svg%3E" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">好吧，回到主题。没想到毕业20年，线性代数，矩阵论穿着人工智能的华贵<span style="color: black;">商场</span>套装，和国家战略契合到<span style="color: black;">一块</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">2. 上市<span style="color: black;">机构</span>业绩暴雷财务一瞥</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">花开两朵，各表一枝。2018年中国A股股市的一大热点<span style="color: black;">便是</span>“业绩暴雷”。上市<span style="color: black;">机构</span>在业务顺利和牛市时粉饰财务报表，在特定情景和时点一次性减记<span style="color: black;">或</span>冲销巨额财产“重新<span style="color: black;">起始</span>”。这些<span style="color: black;">机构</span>有的走在了违法边缘或是<span style="color: black;">已然</span>违法，有的过度<span style="color: black;">运用</span>会计政策的空间迎合市场<span style="color: black;">爱好</span>（<span style="color: black;">例如</span>对成长的追求等等），<span style="color: black;">最后</span>业绩大变脸巨额一次亏个够。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">康得新曾经是一个明星上市<span style="color: black;">机构</span>企业。成长性和股价<span style="color: black;">连续</span><span style="color: black;">报答</span>令人艳羡。我记得当年还曾浏览过这家<span style="color: black;">机构</span>的财报，无奈缺乏起码的行业知识放弃了。2018年康得新<span style="color: black;">已然</span>ST了。当年浓眉大眼的家伙怎么和丑闻和业绩洗澡纠缠在<span style="color: black;">一块</span>了！？</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">媒介</span><span style="color: black;">热榜</span>，康得新账上巨额现金，<span style="color: black;">然则</span>有息负债率<span style="color: black;">亦</span>相当高。<span style="color: black;">倘若</span><span style="color: black;">无</span>其他猫腻，巨额现金和<span style="color: black;">巨额</span>负债只能是一个<span style="color: black;">科研</span>的线索。<span style="color: black;">例如</span>苹果<span style="color: black;">机构</span>现金<span style="color: black;">非常多</span>，<span style="color: black;">然则</span>正在<span style="color: black;">增多</span>有息负债回购股份等等。巴菲特在股东的信里谈到苹果的<span style="color: black;">目的</span>是达到“现金中性”，<span style="color: black;">亦</span><span style="color: black;">便是</span>扣除必要营运资本考量后的净现金为0。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="data:image/svg+xml,%3C%3Fxml version=1.0 encoding=UTF-8%3F%3E%3Csvg width=1px height=1px viewBox=0 0 1 1 version=1.1 xmlns=http://www.w3.org/2000/svg xmlns:xlink=http://www.w3.org/1999/xlink%3E%3Ctitle%3E%3C/title%3E%3Cg stroke=none stroke-width=1 fill=none fill-rule=evenodd fill-opacity=0%3E%3Cg transform=translate(-249.000000, -126.000000) fill=%23FFFFFF%3E%3Crect x=249 y=126 width=1 height=1%3E%3C/rect%3E%3C/g%3E%3C/g%3E%3C/svg%3E" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">康得新<span style="color: black;">到底</span>怎么啦！？静安笔记<span style="color: black;">经过</span>Python编程汇总了康得新2012年~2017年财务数据：<span style="color: black;">【4】</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_png/PETAnHzkYL5dtO4jpGmcHyUicxYhE5wa6YOII2OeawBVCAAgNKUwxFiadxiaPLGWftglfMiaNGYSjQtdvAVqqFzWhQ/640?wx_fmt=png&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">上表<span style="color: black;">能够</span>看出，康得新<span style="color: black;">连续</span><span style="color: black;">数年</span>高成长的<span style="color: black;">背面</span>，<span style="color: black;">应收账款超出比例的增长</span>，<span style="color: black;">ROE的<span style="color: black;">显著</span>波动</span>都是警示信号。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">由此静安笔记想到，能否应用时下伟大的人工智能技术，<span style="color: black;">经过</span>对上市<span style="color: black;">机构</span>的历史财务数据进行<span style="color: black;">设备</span>学习，大数据预警业绩暴雷<span style="color: black;">或</span>个别上市<span style="color: black;">机构</span><span style="color: black;">反常</span>的财务数据呢!?</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_png/PETAnHzkYL4kCRzicGmuy1CJMHx4BHLscmJma1QRQU54kjY1ZZ9V93gVF85SQp0QWjHTb2X9jWF56cAorBA57ibw/640?wx_fmt=png&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">毕业于北京交通大学信息所<span style="color: black;">这般</span>前沿人工智能学术<span style="color: black;">科研</span><span style="color: black;">公司</span>的老程序员静安笔记给出的答案是正面的，鼓舞人心的。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">3. 人工智能之<span style="color: black;">设备</span>学习</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">设备</span>学习<span style="color: black;">便是</span><span style="color: black;">经过</span>构建数据模型来“理解”数据。<span style="color: black;">一般</span>模型有<span style="color: black;">非常多</span>可调参数。学习的过程<span style="color: black;">便是</span>参数适配的过程。狭义看，<span style="color: black;">日前</span>人工智能<span style="color: black;">重点</span><span style="color: black;">指的是</span>深度学习<span style="color: black;">【5】</span>，<span style="color: black;">例如</span>卷积神经网络(CNN)和循环神经网络(RNN)。而<span style="color: black;">设备</span>学习的范畴更宽泛<span style="color: black;">有些</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">静安笔记<span style="color: black;">按照</span>上市<span style="color: black;">机构</span>历史财务数据财务特征，<span style="color: black;">亦</span><span style="color: black;">思虑</span>到数据量比较有限，<span style="color: black;">首要</span>尝试传统的随机森林(random forest)和群集(clustering)<span style="color: black;">设备</span>学习<span style="color: black;">办法</span>检测财务数据<span style="color: black;">反常</span>的上市<span style="color: black;">机构</span>（候选列表）。鉴于上市<span style="color: black;">机构</span>财务数据的时间序列特性，续篇会给出基于循环神经网络的Long Short Term Memory RNN深度学习<span style="color: black;">办法</span>的尝试。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_png/PETAnHzkYL5dtO4jpGmcHyUicxYhE5wa6upvk5YKribvQCI722675rM05ankSajjsP4SwcJibOfW6J3HfEkx8PxZQ/640?wx_fmt=png&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><strong style="color: blue;">随机森林</strong></span>简单的讲<span style="color: black;">便是</span>一组决策树（森林）<span style="color: black;">经过</span>随机因子整合<span style="color: black;">最后</span>决策模型。随机森林<span style="color: black;">设备</span>学习算法属于无监督学习(unsupervised learning)，应用起来简单直接，<span style="color: black;">针对</span>财务信息这些非海量数据训练学习效率很高，<span style="color: black;">况且</span>避免了过度拟合(over fitting)。随机森林应用于<span style="color: black;">反常</span>检测(anomaly detection)<span style="color: black;">已然</span>非常成熟，<span style="color: black;">例如</span><span style="color: black;">运用</span>类似<span style="color: black;">设备</span>学习算法：</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">金融行业，<span style="color: black;">发掘</span>信用卡作<span style="color: black;">坏处</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">医药行业，<span style="color: black;">按照</span><span style="color: black;">病人</span>就诊和用药记录探索<span style="color: black;">病人</span><span style="color: black;">重点</span><span style="color: black;">疾患</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">电子商务，给消费者<span style="color: black;">举荐</span><span style="color: black;">关联</span><span style="color: black;">制品</span>或服务</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">部分静安笔记读者对编程代码很感兴趣，<span style="color: black;">这儿</span>给出<span style="color: black;">重点</span>package，您只要简单搜索就能查看非常完备的sklean package<span style="color: black;">运用</span><span style="color: black;">办法</span>和实例：</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_png/PETAnHzkYL5dtO4jpGmcHyUicxYhE5wa64YmObVpGnuqbgZP6IIkQV1bKSGia2twkcYGPIyzdRPYC4dKMjC2gmRA/640?wx_fmt=png&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">做为</span>输入数据的财务信息整理<span style="color: black;">亦</span>非常简单，<span style="color: black;">经过</span>聚宽<span style="color: black;">得到</span>所有上市<span style="color: black;">机构</span><span style="color: black;">重点</span>财务数据序列（参考本文康得新财务数据图），经过简单预处理(scaler)<span style="color: black;">就可</span>进行训练(training)：</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_png/PETAnHzkYL5dtO4jpGmcHyUicxYhE5wa6iccXQBohqzKuySewEjfYJBA8Ylc7s1c4lKm4pwWfjB5yuc3ToquiaRzg/640?wx_fmt=png&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">最后一步<span style="color: black;">按照</span>给定的阈值给出<span style="color: black;">反常</span>数据，<span style="color: black;">亦</span><span style="color: black;">便是</span><span style="color: black;">关联</span>上市<span style="color: black;">机构</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><strong style="color: blue;">随机森林讨论</strong></span>：值得肯定的是sklearn很容易上手，<span style="color: black;">倘若</span>能够把基本数据整理好，随机森林<span style="color: black;">设备</span>学习算法总是能够给出<span style="color: black;">有些</span>貌似不错的输出。比较耗时的除了整理数据，还要尝试<span style="color: black;">区别</span><span style="color: black;">关联</span>性的输入训练数据（本算法和聚类算法<span style="color: black;">反常</span>数据是其中的一部分）。初步结论还是以行业上市<span style="color: black;">机构</span>为输入，合情<span style="color: black;">恰当</span>，结果似乎<span style="color: black;">亦</span>不错：</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">训练数据：证监会数据“橡胶和塑料制品业”， 2010~2017年报<span style="color: black;">关联</span>几类简单加总计算的财务数据</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">IsolationForest财务<span style="color: black;">反常</span>上市<span style="color: black;">机构</span>输出<span style="color: black;">【6】</span>：珠海中富, 华塑控股, ST康得新, 天铁股份</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">针对</span><span style="color: black;">设备</span>学习尤其是深度学习，<span style="color: black;">怎样</span>判断给出的结果是“正确”的，是一个挺大的<span style="color: black;">行业</span>。<span style="color: black;">针对</span>随机森林IsolationForest，<span style="color: black;">能够</span><span style="color: black;">运用</span><span style="color: black;">sklearn.metrics</span><span style="color: black;">.</span><span style="color: black;">confusion_matrix进行<span style="color: black;">评估</span>。</span>
</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">下篇给出<span style="color: black;">有些</span>学界<span style="color: black;">评估</span>门类参考。本文只是抛砖<span style="color: black;">举荐</span>一个思路。除了参考confusion matrix，<span style="color: black;">亦</span><span style="color: black;">能够</span>用常规<span style="color: black;">办法</span>对<span style="color: black;">反常</span>数据和<span style="color: black;">反常</span>上市<span style="color: black;">机构</span>进行进一步分析，应用不存在致命缺陷。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><strong style="color: blue;">聚类<span style="color: black;">设备</span>学习</strong></span>(Clustering)<span style="color: black;">办法</span>稍微<span style="color: black;">繁杂</span>一点，属于无监督学习(unsupervised learning)，<span style="color: black;">亦</span>可用于<span style="color: black;">发掘</span><span style="color: black;">反常</span>。本例（行业历史财务数据<span style="color: black;">做为</span>训练输入，见上面讨论）中聚类<span style="color: black;">设备</span>学习<span style="color: black;">办法</span>似乎<span style="color: black;">针对</span>输入信息比较挑剔，结果相对输入变化<span style="color: black;">亦</span>比<span style="color: black;">很强</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">下图分别是<span style="color: black;">归类</span>结果和聚类学习<span style="color: black;">反常</span>检出。<span style="color: black;">重视</span>聚类学习结果直观上看并不完美<span style="color: black;">表现</span>“凸性”约束，有些行业<span style="color: black;">做为</span>训练数据“凸性”更差。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_png/PETAnHzkYL4kCRzicGmuy1CJMHx4BHLscwetSYHHkSOx5QztYCTNgqJicY7AOOpDAQbibkKuhntnYxldc8RbEDe7g/640?wx_fmt=png&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_png/PETAnHzkYL4kCRzicGmuy1CJMHx4BHLscyOANOO47xp5iazastMbxOmUUyCJZ48Shwjh5g6N34SqkeTJ1h5IlPSA/640?wx_fmt=png&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">ClusteringForest财务<span style="color: black;">反常</span>上市<span style="color: black;">机构</span>输出<span style="color: black;">【6】</span>
</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">：每次输出<span style="color: black;">区别</span>，珠海中富 或 ‘ST康得新’</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_png/PETAnHzkYL5pibuEN3JEuCUkWYgAhlxibMg5CSYzv2I8wo1HmOwDJ9jSk4ibGNY3Y7qL3oA5ep6Rrp78JasAPFDicg/640?wx_fmt=png&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">下次<span style="color: black;">咱们</span>简单讨论一下<span style="color: black;">运用</span>循环神经网络LSTM做<span style="color: black;">一样</span>的事。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">咱们</span><span style="color: black;">能够</span>提前<span style="color: black;"><strong style="color: blue;">给出总结和<span style="color: black;">意见</span></strong></span>说：</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">基于历史财务信息的<span style="color: black;">设备</span>学习（<span style="color: black;">包含</span>深度学习）用于检测监测上市<span style="color: black;">机构</span>财务数据<span style="color: black;">反常</span>有<span style="color: black;">知道</span>的益处，进一步<span style="color: black;">科研</span><span style="color: black;">恰当</span>的训练数据以及对<span style="color: black;">设备</span>学习<span style="color: black;">办法</span>输出统一<span style="color: black;">评估</span>。</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;"><strong style="color: blue;">基于历史财务信息的</strong><span style="color: black;">设备</span>学习（<span style="color: black;">包含</span>深度学习）<span style="color: black;">能够</span>用于投资者初步否决<span style="color: black;">有些</span>投资决策<span style="color: black;">【7】</span></strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;"><strong style="color: blue;">基于历史财务信息的</strong><span style="color: black;">设备</span>学习（<span style="color: black;">包含</span>深度学习）<span style="color: black;">亦</span><span style="color: black;">能够</span>用于监管当局（请大胆<span style="color: black;">举荐</span>给证监会和易会满主席！带上我）大数据监控分析。</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;"><strong style="color: blue;"><strong style="color: blue;">基于历史财务信息的</strong><span style="color: black;">设备</span>学习（<span style="color: black;">包含</span>深度学习）</strong><span style="color: black;">能够</span>做为对冲基金等初筛做空<span style="color: black;">目的</span>的一个工具。</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">倘若</span>读者<span style="color: black;">伴侣</span>相对自己手中的持仓股票“人工智能”考察一把，不妨留言试试（不<span style="color: black;">保证</span>回复时间，不<span style="color: black;">必定</span>有用，just have fun!）。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;"><span style="color: black;">各位<span style="color: black;">伴侣</span>看完后请点右下角“好看”，谢谢鼓励！</span></strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">【1】我还是要了一台SGI工作站上机时间。一位学俄语的博士<span style="color: black;">亦</span>常用这台<span style="color: black;">设备</span>。博士学了一年英语，竟然考GRE拿了全奖美国高校博士录取。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">【2】https://www.bioinf.jku.at/publications/older/2604.pdf</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">【3】我的毕业论文是《小波变换应用于图像在互联网的渐进网络传输》（大意)。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">【4】数据<span style="color: black;">源自</span>聚宽。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">【5】“Its <span style="color: black;"><strong style="color: blue;">deep</strong></span> if it has more than one stage of non-linear feature transformation" - 2015, LeCun</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">【6】"<span style="color: black;">反常</span>“并不<span style="color: black;">表率</span>财务造假，请放在全文环境下<span style="color: black;">小心</span>看待。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">【7】invert, always invert! 初步否决的<span style="color: black;">道理</span>在于更加<span style="color: black;">警觉</span>关注<span style="color: black;">反常</span>财务<span style="color: black;">机构</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><a style="color: black;">GDP，M2，社融，利率和A股股市（个别<span style="color: black;">照片</span>可能<span style="color: black;">导致</span>轻度不适）</a></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><a style="color: black;">七年之样：上证50，中证500指数价值和收益分析（深度好文）</a></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><a style="color: black;">A股市场14年来行业特征和</a>价值一窥 (深度好文)</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><a style="color: black;">用数<span style="color: black;">据述</span>话：小议A股市场的整体投资价值【长且重要】</a></p>
你的见解独到，让我受益匪浅，期待更多交流。 你的见解真是独到，让我受益匪浅。 谷歌网站排名优化 http://www.fok120.com/ 一看到楼主的气势，我就觉得楼主同在社区里灌水。 感谢您的精彩评论，为我带来了新的思考角度。
页:
[1]