他的UTD科研证明,AI算法能够给数字营销带来“巨变”……| 学术对话No.1
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">倘若</span>一个商场<span style="color: black;">连续</span>促销,你会有立即购物的冲动吗?可能不会!<span style="color: black;">无</span>间隔的促销或许会降低你的消费激情。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">倘若</span>一个线上购物平台<span style="color: black;">连续</span>的向你推出同一品类的东西,你会买账吗?可能不会!过于<span style="color: black;">类似</span>的商品会让你产生审美疲劳,并失去下单的欲望。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">然而,你相信吗,<span style="color: black;">针对</span>这些问题的改进,不需要商家或品牌方,仅仅<span style="color: black;">经过</span>人工智能就<span style="color: black;">能够</span>实现!从语音识别到图像处理、从自动驾驶汽车到医疗<span style="color: black;">设备</span>人,如今再到营销<span style="color: black;">行业</span>,人工智能的<span style="color: black;">“深度强化学习”</span>为数字<span style="color: black;">行业</span>带来<span style="color: black;">非常多</span>前所未有的机会。</p><img src="https://mmbiz.qpic.cn/mmbiz_jpg/vk4jak4unGonVd9GCgmA8CUJ5dhFbWrOFiaKucZj2d3VT1k8Vaicbj62bAF5jsWhJKneSfeficmwjp07ib7VDmZKOg/640?wx_fmt=jpeg&wxfrom=5&wx_lazy=1&wx_co=1&tp=webp" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">照片</span><span style="color: black;">源自</span>:©千库网</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">浙江大学管理学院市场营销学系教授、品质管理硕士学位项目论文导师王小毅和团队正是<span style="color: black;">发掘</span>了“深度强化学习”为数字营销<span style="color: black;">行业</span>革命带来的巨大启示和变化,对其进行了一系列深入<span style="color: black;">科研</span>,并设计了一种基于DRL的<strong style="color: blue;">个性化<span style="color: black;">目的</span>定位策略</strong>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">日前,王小毅与团队在管理科学<span style="color: black;">行业</span>国际顶刊<strong style="color: blue;">《MANAGEMENT SCIENCE》</strong>上<span style="color: black;">发布</span>了<span style="color: black;">相关</span>数智营销的<span style="color: black;">研究</span>成果<strong style="color: blue;">"<span style="color: black;">Deep Reinforcement Learning for Sequential Targeting</span></strong>"(运用深度强化学习实现动态连续定位营销)。</p><img src="https://mmbiz.qpic.cn/mmbiz_png/vk4jak4unGonVd9GCgmA8CUJ5dhFbWrOuhMYdRnBZWmnmwvEm6urN2S5TibG3GRsmXyz0ricH8LicvHaGZxEhxNIA/640?wx_fmt=png&wxfrom=5&wx_lazy=1&wx_co=1&tp=webp" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><a style="color: black;">Deep Reinforcement Learning for Sequential Targeting.pdf</a></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">成果在线<span style="color: black;">发布</span>,点击链接查看详情</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">到底</span>什么是“深度强化学习”?</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">人工智能算法是<span style="color: black;">怎样</span>助力数字营销发展的?</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">研究</span>团队都做了<span style="color: black;">那些</span>尝试?</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">本期【<strong style="color: blue;">“<span style="color: black;">研究</span>”</strong>】,让<span style="color: black;">咱们</span><span style="color: black;">一块</span>走进王小毅团队的“AI营销世界”。</p><img src="https://mmbiz.qpic.cn/mmbiz_png/vk4jak4unGonVd9GCgmA8CUJ5dhFbWrOMCYdf7qr1A834rX094UT3fcxbYacv9nuicuguSHJXdxEIUEelaXqSzA/640?wx_fmt=png&wxfrom=5&wx_lazy=1&wx_co=1&tp=webp" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">学者简介</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">王小毅:浙江大学管理学院市场营销学教授、博导,<strong style="color: blue;">品质管理硕士学位项目论文导师,</strong>数字化转型与脑机智能营销交叉<span style="color: black;">科研</span>专家,<span style="color: black;">得到</span>中国<span style="color: black;">商场</span>联合会科技进步一等奖等省部级奖励多项。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">在《管理世界》、Management Science、MarketingScience、JournalofMarketingResearch、InformationSystemsResearch等期刊<span style="color: black;">发布</span>论文,谷歌学术H指数为20。</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;">她们</span><span style="color: black;">为么</span>聚焦<span style="color: black;">设备</span>学习来</strong></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;">科研</span>营销定位策略?</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">目的</span>市场定位营销(Target Marketing)<span style="color: black;">一般</span><span style="color: black;">指的是</span>企业识别各个<span style="color: black;">区别</span>的购买者群体,<span style="color: black;">选取</span>其中一个或几个<span style="color: black;">做为</span><span style="color: black;">目的</span>市场,运用适当的市场营销组合,集中力量为<span style="color: black;">目的</span>市场服务,满足<span style="color: black;">目的</span>市场的需要。然而,在数字化时代,企业的营销策略越来越依赖于与消费者的<strong style="color: blue;">高频次互动</strong>和<span style="color: black;">针对</span><strong style="color: blue;">营销策略的快速<span style="color: black;">调节</span></strong>。此时,经典营销理论的传统思路和手段<span style="color: black;">办法</span>面临着越来越多的挑战。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">传统的定位营销策略<span style="color: black;">常常</span>只<span style="color: black;">思虑</span>与顾客进行一次性“买卖”,较为依赖前期策划和庞大的<span style="color: black;">宣传</span>投放,忽略了时间<span style="color: black;">原因</span>对消费者<span style="color: black;">行径</span>的影响和促销活动的连续性,成本巨大且结果不确定。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">实质</span><span style="color: black;">状况</span>中,企业需要在多个周期内依次决定向谁发放优惠券、<span style="color: black;">选取</span>哪个促销活动以及决定两个活动之间的等待时长。与此<span style="color: black;">同期</span>,消费者的真实偏好<span style="color: black;">亦</span>会随着时间的推移而快速改变。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">因此呢</span>,不论是学术界和产业界都迫切需要一种随着<span style="color: black;">持续</span>变化的顾客<span style="color: black;">行径</span>而进行<span style="color: black;">调节</span>的<strong style="color: blue;">适应性定位营销策略</strong>。</p><img src="https://mmbiz.qpic.cn/mmbiz_png/vk4jak4unGonVd9GCgmA8CUJ5dhFbWrO3jRurHzjiaxPpRvVpXichgHafK7fFMJq0M6Yjulboun55HhxxMLkeM8w/640?wx_fmt=png&wxfrom=5&wx_lazy=1&wx_co=1&tp=webp" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">企业<span style="color: black;">平常</span>的动态连续促销策略</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">做为</span>一种新兴的人工智能算法,“深度强化学习”(<span style="color: black;">Deep Reinforcement Learning,DRL</span>
</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">)算法<span style="color: black;">能够</span>在<span style="color: black;">无</span>人类监督的<span style="color: black;">状况</span>下<span style="color: black;">连续</span>“学习”<span style="color: black;">加强</span>性能,对<span style="color: black;">处理</span><span style="color: black;">以上</span>问题<span style="color: black;">拥有</span>巨大的<span style="color: black;">潜能</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">“深度强化学习(DRL)”是一种基于奖励的学习<span style="color: black;">办法</span>,它<span style="color: black;">能够</span><span style="color: black;">帮忙</span>定位营销策略实现个性化、强适应性,但它<span style="color: black;">亦</span>面临着<span style="color: black;">许多</span>挑战,例如<span style="color: black;">针对</span>时间影响的处理;<span style="color: black;">针对</span><span style="color: black;">繁杂</span>消费者<span style="color: black;">行径</span>维度的处理;<span style="color: black;">针对</span>策略效果的<span style="color: black;">评定</span>等。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">正是基于此背景,王小毅教授与团队提出了这种<span style="color: black;">创立</span>在DRL算法<span style="color: black;">基本</span>上的<strong style="color: blue;">个性化<span style="color: black;">目的</span>定位策略</strong>
</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">,并<span style="color: black;">运用</span>一种基于量化的不确定性学习启发式<span style="color: black;">办法</span>来使DRL适应<span style="color: black;">繁杂</span>的消费者<span style="color: black;">行径</span>维度。</p>
<img src="https://mmbiz.qpic.cn/mmbiz_png/vk4jak4unGonVd9GCgmA8CUJ5dhFbWrO368D8aBVibMX2ia5Fh0FCnetuibXuVuKBcov9nRYEF5uVfQSIRDIgORWw/640?wx_fmt=png&wxfrom=5&wx_lazy=1&wx_co=1&tp=webp" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">采用双流法( Two-Stream Computations)的竞争网络架构(Double-Dueling Network Architecture)</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;">研究</span>团队都做出了<span style="color: black;">那些</span>尝试?</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">人工智能的“自主学习”和<span style="color: black;">持续</span>升级,是<span style="color: black;">经过</span><strong style="color: blue;">“深度学习”</strong>和<strong style="color: blue;">“强化学习”</strong>等技术来实现的。深度学习<span style="color: black;">拥有</span>较强的感知能力,<span style="color: black;">然则</span>缺乏<span style="color: black;">必定</span>的决策能力;而强化学习<span style="color: black;">拥有</span>决策能力,对感知问题束手无策。<span style="color: black;">因此呢</span>,将两者结合起来,<span style="color: black;">优良</span>互补,就能为<span style="color: black;">繁杂</span>系统的感知决策问题<span style="color: black;">供给</span>新的<span style="color: black;">处理</span>思路,这<span style="color: black;">便是</span><span style="color: black;">咱们</span>所说的<strong style="color: blue;">“深度强化学习(DRL)”,</strong><span style="color: black;">亦</span>是一种更接近人类思维方式的人工智能<span style="color: black;">办法</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">近年来,“深度强化学习(DRL)”取得了巨大突破,这种人工算法<span style="color: black;">经过</span>在给定奖励或<span style="color: black;">处罚</span>反馈的<span style="color: black;">状况</span>下,让系统以<span style="color: black;">持续</span>试错的方式与营销环境进行连续交互,从而寻找到“最佳策略”,以完成学习的过程。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">王小毅教授团队正是基于DRL人工算法,在序贯性定位营销设置的场景下,设计了这项<strong style="color: blue;">个性化定位营销策略。</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">这项策略<span style="color: black;">首要</span><span style="color: black;">运用</span>连续价格促销来吸引顾客的即时<span style="color: black;">重视</span>力并锁定<span style="color: black;">她们</span>,并在每两次价格促销之间<span style="color: black;">供给</span>一个非促销期(即冷却期),并且随着时间的推移<span style="color: black;">逐步</span><span style="color: black;">增多</span>冷却期的长度,以便顾客<span style="color: black;">调节</span>其价格参考点。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">序贯性定位营销(Sequential Targeting),即连续地对消费者开展促销<span style="color: black;">行径</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">科研</span><span style="color: black;">显示</span>,采用深度强化学习<span style="color: black;">办法</span><span style="color: black;">能够</span><span style="color: black;">处理</span>当前实施连续定位营销策略所面临的三大挑战:</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">01</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">前瞻性</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">平衡<span style="color: black;">机构</span>当前收入和<span style="color: black;">将来</span>收入</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">02</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">边探索学习边<span style="color: black;">得到</span>市场<span style="color: black;">报答</span></strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">最大化利润,<span style="color: black;">同期</span><span style="color: black;">经过</span>探索<span style="color: black;">研发</span><span style="color: black;">持续</span>学习</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">03</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">可扩展性</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">应对高维状态和营销政策空间</p><img src="https://mmbiz.qpic.cn/mmbiz_jpg/vk4jak4unGonVd9GCgmA8CUJ5dhFbWrO3iczjiafoEcmABq22ZQicgg9Mw4kF4EkcibuKzO3pwBzIuLz0NjkeaWNKQ/640?wx_fmt=jpeg&wxfrom=5&wx_lazy=1&wx_co=1&tp=webp" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">照片</span><span style="color: black;">源自</span>:©千库网</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">为了更好地使DRL能够适应<span style="color: black;">繁杂</span>的消费者<span style="color: black;">行径</span>维度,<span style="color: black;">科研</span>团队又提出了一种基于量化的<strong style="color: blue;">不确定性学习启发式算法</strong>,以实现<span style="color: black;">有效</span>的探索<span style="color: black;">研发</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">经过</span><span style="color: black;">评定</span>,其结果<span style="color: black;">显示</span>,平均<span style="color: black;">状况</span>下,采用这种新算法代理(agent)产生的<span style="color: black;">长时间</span>收入比采用传统<span style="color: black;">办法</span>所产生的收入多26.75%,学习速度<span style="color: black;">亦</span>比所有基准中,其他产业界常用算法模型的速度快76.92%。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">为了更好地理解<span style="color: black;">科研</span>结果<span style="color: black;">背面</span>的潜在机制,<span style="color: black;">科研</span>团队进行了多种可解释性算法的<span style="color: black;">科研</span>,能够<span style="color: black;">诠释</span>个体和群体水平上学习最优策略的<span style="color: black;">行径</span>模式。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">另外</span>,王小毅教授与团队合作者提出的“模拟”在线测试环境为DRL训练和测试构建了用户<span style="color: black;">行径</span>模拟器,为平台<span style="color: black;">供给</span>了一种节省成本的方式来学习DRL代理,而无需在现实世界中运行<span style="color: black;">海量</span>、测试。</p><img src="https://mmbiz.qpic.cn/mmbiz_png/vk4jak4unGonVd9GCgmA8CUJ5dhFbWrOWAJ4gs3sKEXCZVGicOFk4k6dBsM8biaRCgrKLwhkQQVxUMI5xY3ULBDA/640?wx_fmt=png&wxfrom=5&wx_lazy=1&wx_co=1&tp=webp" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">照片</span><span style="color: black;">源自</span>:©千库网</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">然而,不得不承认的是人工智能算法在营销<span style="color: black;">行业</span>的应用仍然面临着<span style="color: black;">有些</span>挑战。<span style="color: black;">因为</span>算法依靠训练数据和计算资源,需要进行<span style="color: black;">海量</span>的实验和优化来<span style="color: black;">加强</span>算法的效率和准确性;<span style="color: black;">因为</span>消费者<span style="color: black;">行径</span>的<span style="color: black;">繁杂</span>性和不确定性,需要<span style="color: black;">更加多</span>的数据和模型来解释和预测消费者<span style="color: black;">行径</span>;<span style="color: black;">因为</span><span style="color: black;">运用</span>DRL算法的实时决策和<span style="color: black;">调节</span>需要在<span style="color: black;">短期</span>内做出,<span style="color: black;">因此呢</span>需要<span style="color: black;">创立</span>一个实时决策系统来支持DRL算法的应用。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">为<span style="color: black;">认识</span>决这些挑战,王小毅教授团队正在进一步<span style="color: black;">奋斗</span>开展<span style="color: black;">开发</span>工作,有望在今年<span style="color: black;">供给</span>一个更加通用和普适的<strong style="color: blue;">DRL框架</strong>,框架<span style="color: black;">亦</span><span style="color: black;">能够</span>很容易地推广到其他<span style="color: black;">目的</span>市场营销场景,例如基于位置的服务、在线流<span style="color: black;">媒介</span>或在线教育的序贯<span style="color: black;">举荐</span>,可供平台和营销人员在<span style="color: black;">平常</span>实践中<span style="color: black;">容易</span><span style="color: black;">运用</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">该框架将基于量化的不确定性学习启发式<span style="color: black;">办法</span>,结合实时决策系统和用户<span style="color: black;">行径</span>模拟器,以<span style="color: black;">加强</span>DRL算法的效率和准确性,并<span style="color: black;">帮忙</span>企业更好地理解和预测消费者<span style="color: black;">行径</span>,制定更加智能和有效的营销策略。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">总体而言,王小毅教授团队的<span style="color: black;">科研</span>结果<span style="color: black;">表示</span>了“深度强化学习”<span style="color: black;">办法</span>在优化定位营销策略以最大化企业<span style="color: black;">长时间</span>收入方面的巨大<span style="color: black;">潜能</span>,证明了这种<span style="color: black;">办法</span>在数字营销<span style="color: black;">行业</span>能够产生颠覆性影响。</p><img src="https://mmbiz.qpic.cn/mmbiz_jpg/vk4jak4unGonVd9GCgmA8CUJ5dhFbWrOEzdYHmpVuOsLBkbE6dlIxC9m2nPh31ibspicN4rlicVbHBggPdY3oMZGQ/640?wx_fmt=jpeg&wxfrom=5&wx_lazy=1&wx_co=1&tp=webp" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">照片</span><span style="color: black;">源自</span>:©千库网</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;"><span style="color: black;">她们</span>的<span style="color: black;">科研</span>对</strong></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;">品牌价值和数字营销有何<span style="color: black;">道理</span>?</strong></span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">基于“深度强化学习”的营销策略对<span style="color: black;">提高</span>数字化营销效率有何<span style="color: black;">功效</span>?对助力企业制定数字化营销策略有何<span style="color: black;">道理</span>?</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">其实,这项听起来“<span style="color: black;">奥秘</span>”的基于算法驱动的营销<span style="color: black;">科研</span>成果,与品牌、企业的息息<span style="color: black;">关联</span>,且在企业数字化转型中<span style="color: black;">拥有</span>重要价值与战略<span style="color: black;">道理</span>。<span style="color: black;">详细</span><span style="color: black;">来讲</span>,它<span style="color: black;">能够</span><span style="color: black;">帮忙</span>企业更加<span style="color: black;">精细</span>地进行市场定位和<span style="color: black;">目的</span>营销,<span style="color: black;">加强</span>数字化营销的效率和<span style="color: black;">精细</span>度,实现业务增长和品牌价值的<span style="color: black;">提高</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">1</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">更好<span style="color: black;">认识</span>消费者的<span style="color: black;">需要</span>和偏好</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">“深度强化学习”<span style="color: black;">能够</span><span style="color: black;">帮忙</span>企业更好地<span style="color: black;">认识</span>消费者的<span style="color: black;">需要</span>和偏好,从而<span style="color: black;">精细</span>地进行市场定位和<span style="color: black;">目的</span>营销。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">经过</span>对消费者数据的深度分析,企业<span style="color: black;">能够</span><span style="color: black;">认识</span>消费者的<strong style="color: blue;"><span style="color: black;">行径</span>模式和心理特征</strong>,从而制定更加<span style="color: black;">精细</span>的营销策略。例如,企业<span style="color: black;">能够</span><span style="color: black;">按照</span>强化学习算法的分析结果,适时<span style="color: black;">调节</span><span style="color: black;">制品</span>定价、<span style="color: black;">制品</span><span style="color: black;">举荐</span>、<span style="color: black;">宣传</span>投放等策略,以<span style="color: black;">加强</span>消费者的消费忠诚度和购买转化率。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">2</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">优化营销<span style="color: black;">途径</span>,<span style="color: black;">提高</span>效率</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">“深度强化学习”<span style="color: black;">能够</span><span style="color: black;">帮忙</span>企业优化营销<span style="color: black;">途径</span>,<span style="color: black;">加强</span>数字化营销的效率和<span style="color: black;">精细</span>度。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">经过</span>对<span style="color: black;">区别</span>营销<span style="color: black;">途径</span>的数据进行分析和比较,企业<span style="color: black;">能够</span><span style="color: black;">认识</span><span style="color: black;">每一个</span><span style="color: black;">途径</span>的贡献度和效果,从而制定更加科学的<span style="color: black;">途径</span>策略。例如,企业<span style="color: black;">能够</span><span style="color: black;">按照</span>强化学习算法的分析结果,适时<span style="color: black;">调节</span>各个<span style="color: black;">途径</span>的投放比例和投放策略,以<span style="color: black;">加强</span>数字化营销的效率和<span style="color: black;">精细</span>度。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">3</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">实现数字化营销的精细化和智能化</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">深度强化学习算法<span style="color: black;">能够</span><span style="color: black;">帮忙</span>企业实现数字化营销的精细化和智能化。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">经过</span>对消费者数据的深度学习和分析,企业<span style="color: black;">能够</span><span style="color: black;">认识</span>消费者的<span style="color: black;">行径</span>和偏好,从而实现个性化的营销和服务。例如,企业<span style="color: black;">能够</span><span style="color: black;">按照</span>强化学习算法的分析结果,针对<span style="color: black;">区别</span>的消费者群体,<span style="color: black;">供给</span><span style="color: black;">区别</span>的营销<span style="color: black;">方法</span>和服务,以<span style="color: black;">加强</span>消费者的满意度和忠诚度。</p><img src="https://mmbiz.qpic.cn/mmbiz_jpg/vk4jak4unGonVd9GCgmA8CUJ5dhFbWrOO92fxXf0h8m20xf0m0A52jnOtrbAZeMKWskiazB9ib6VKlOdicHSWnrww/640?wx_fmt=jpeg&wxfrom=5&wx_lazy=1&wx_co=1&tp=webp" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">图片<span style="color: black;">源自</span>:©千库网</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">基于这项<span style="color: black;">科研</span>成果,王小毅团队与阿里巴巴合作改进数字营销经营<span style="color: black;">办法</span>论、与浙江中烟合作设计智慧营销大脑,为传统零售<span style="color: black;">商场</span>的制定自有品牌策略等,为企业数智创新<span style="color: black;">供给</span>了新的路径和解题方式。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">日前,王小毅教授联合阿里妈妈发布的DEEPLINK模型,<span style="color: black;">仔细</span>解析了消费者从种草,到首购,在到复购的消费路径。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">如果以“个性化定位营销策略”来分析<span style="color: black;">诠释</span>模型中看似<span style="color: black;">繁杂</span>无序的消费路径,<span style="color: black;">能够</span>将其拆解为:Discover<span style="color: black;">发掘</span>—Engage种草—Enthuse热爱—Perform行动—Initial首购—Numerous复购—Keen忠诚。该模型是基于过去围绕<span style="color: black;">营销</span>漏斗的AIPL模型(Awareness认知-Interest兴趣-Purchase购买-Loyalty忠实)的进一步升级。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">这般</span>的路径分析能够<span style="color: black;">帮忙</span>企业更好地理解和预测消费者<span style="color: black;">行径</span>,制定更加智能和有效的营销策略。</p><img src="https://mmbiz.qpic.cn/mmbiz_png/vk4jak4unGonVd9GCgmA8CUJ5dhFbWrOIYxjg5TUCr674kKibnslC5ickN03RB4bLYTcQzjE40IWf8GD2lfW9Zog/640?wx_fmt=png&wxfrom=5&wx_lazy=1&wx_co=1&tp=webp" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">附:论文摘要</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">Deep reinforcement learning (DRL) has opened up many unprecedented opportu-nities in revolutionizing the digital marketing field. In this study, we designed a DRL-based personalized targeting strategy in a sequential setting. We show that the strategy is able to address three important challenges of sequential targeting: (1) forward looking (balancing between a firm’s current revenue and future revenues), (2) earning while learning (maximizing profits while continuously learning through exploration-exploitation), and (3) scalability (cop-ing with a high-dimensional state and policy space). We illustrate this through a novel design of a DRL-based artificial intelligence (AI) agent. To better adapt DRL to complex consumer behavior dimensions, we proposed a quantization-based uncertainty learning heuristic for effi-cient exploration-exploitation. Our policy evaluation results through simulation suggest that the proposed DRL agent generates 26.75% more long-term revenues than can the non-DRL approaches on average and learns 76.92% faster than the second fastest model among all bench-marks. Further, in order to better understand the potential underlying mechanisms, we con-ducted multiple interpretability analyses to explain the patterns of learned optimal policy at both the individual and population levels. Our findings provide important managerial- relevant and theory-consistent insights. For instance, consecutive price promotions at the begin-ning can capture price-sensitive consumers’ immediate attention, whereas carefully spaced nonpromotional “cooldown” periods between price promotions can allow consumers to adjust their reference points. Additionally, consideration of future revenues is necessary from a long- term horizon, but weighing the future too much can also dampen revenues. In addition, analy-ses of heterogeneous treatment effects suggest that the optimal promotion sequence pattern highly varies across the consumer engagement stages. Overall, our study results demonstrate DRL’s potential to optimize these strategies’ combination to maximize long-term revenues.</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">本期素材 | ZJUSOM官微</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">本期责编 | ZUPUC品牌推广</span></p>
感谢你的精彩评论,为我的思绪打开了新的窗口。 这篇文章真的让我受益匪浅,外链发布感谢分享! “沙发”(SF,第一个回帖的人) i免费外链发布平台 http://www.fok120.com/ “板凳”(第三个回帖的人)
页:
[1]