围棋“人机大战”上演
<p>一场举世瞩目的机器人与人类大脑的对决正在韩国首尔上演。对战双方分别是谷歌的人工智能<span word="AlphaGo">AlphaGo</span>机器人与世界围棋高手李世石。</p><p><span word="The">The</span> <span word="prize">prize</span> <span word="is">is</span> $1<span word="m">m</span> (£700,000) <span word="but">but</span> <span word="there">there</span> <span word="is">is</span> <span word="a">a</span> <span word="lot">lot</span> <span word="more">more</span> <span word="at">at</span> <span word="stake">stake</span></p><p>请看相关报道：</p><p><span word="The">The</span> <span word="artificial">artificial</span> <span word="intelligence">intelligence</span> <span word="program">program</span> <span word="AlphaGo">AlphaGo</span> <span word="defeated">defeated</span> <span word="top">top</span> <span word="professional">professional</span> <span word="Go">Go</span> <span word="player">player</span> <span word="Lee">Lee</span> <span word="Sedol">Sedol</span> 9<span word="p">p</span> <span word="in">in</span> <span word="the">the</span> <span word="first">first</span> <span word="game">game</span> <span word="of">of</span> <span word="historic">historic</span> <span word="man">man</span> <span word="vs">vs</span> <span word="machine">machine</span> <span word="battle">battle</span>.</p><p>在具有历史性意义的“人机大战”首场对决中，人工智能程序<span word="AlphaGo">AlphaGo</span>打败了顶级专业棋手李世石九段。</p><p>此次“人机大战”（<span word="man">man</span> <span word="VS">VS</span> <span word="machine">machine</span> <span word="battle">battle</span>）对战双方分别是韩国著名棋手李世石和谷歌的人工智能（<span word="artificial">artificial</span> <span word="intelligence">intelligence</span>，<span word="AI">AI</span>）<span word="AlphaGo">AlphaGo</span>。</p><p>人工智能“<span word="AlphaGo">AlphaGo</span>” 中的<span word="Alpha">Alpha</span>是排位第一的希腊字母（<span word="a">a</span>），<span word="Go">Go</span>指的是围棋。围棋还可以用<span word="baduk">baduk</span> ，<span word="weiqi">weiqi</span>或<span word="the">the</span> <span word="game">game</span> <span word="of">of</span> <span word="Go">Go</span>表示，是源自中国的一种棋盘游戏（<span word="board">board</span> <span word="game">game</span>），比象棋（<span word="chess">chess</span>）要复杂的多。</p><p>大家对人工智能并不陌生，从苹果的<span word="Siri">Siri</span>，到日常浏览的搜索引擎（<span word="search">search</span> <span word="engine">engine</span>），再到网络的文章推荐和商品推荐系统，这些全都是人工智能。但<span word="AlphaGo">AlphaGo</span>跟这些常见的<span word="AI">AI</span>不同。</p><p><span word="AlphaGo">AlphaGo</span>的核心是两种不同的深度神经网络——策略网络（<span word="policy">policy</span> <span word="network">network</span>）和值网络（<span word="value">value</span> <span word="network">network</span>），<span word="AlphaGO">AlphaGO</span>利用这两个网络分析局势，判断每种下子策略的优劣。它具有强大的学习能力，能通过强化学习（<span word="reinforcement">reinforcement</span> <span word="learning">learning</span>）策略进行自我对弈来提高围棋水平。它的背后是一群杰出的机器学习（<span word="machine">machine</span> <span word="learning">learning</span>）算法领域的专家。</p>
页:
[1]