立委博士，问问副总裁，聚焦大模型及其应用。Netbase前首席科学家10年，期间指挥研发了18种语言的理解和应用系统，鲁棒、线速，scale up to 社会媒体大数据，语义落地到舆情挖掘产品，成为美国NLP工业落地的领跑者。Cymfony前研发副总八年，曾荣获第一届问答系统第一名（TREC-8 QA Track），并赢得17个小企业创新研究的信息抽取项目（PI for 17 SBIRs）。
I showed the First Lady's news pictures to my daughter. She was so intrigued, "Dad, Mom told me that you used to teach First Lady many years ago, is that true?" "It is true, but that was only a short time, one or two semesters, and it was not her major course. As a part-time lecturer, I was teaching Advanced English to graduate students in the music conservatory and she happened to be one in my class. She was already famous then as a new star for folk songs." Tanya got excited, "Well, you never know, maybe her English training in graduate school helps her in state visits today. My Dad is cool." She continued, "Dad, Mom also told me that you were interpreter for foreign minister when she dated you, is that true?" "Well, that was largely an accident, only happened once when I substituted some professor to act as interpreter for the former foreign minister and former Chinese congress vice-chairman Mr. Huang Hua. Your Mom agreed to date me partially because of her seeing a picture of me interporeting for the VIP Mr. Huang. So I guess I benefited from that 'accident'." Tanya was amused and felt very proud, "I have the coolest Dad in the world. He was so successful even when he was young, teaching future first lady and interpreting for the then foreign minister. Wow."
The personal story aside, Chinese social media are never short of coverage and fans of Chinese First Lady Mrs Peng Liyuan in the last few years. For too long China watched the western media covering first ladies in the US and other countries without being able to brag about its own. Since Mrs. Peng went on the spotlight and accompanied Chinese President Xi Jinping on world trips, the Chinese netters have been overjoyed to follow her all the way with compliments and amazement in her gracefulness. Mrs. Peng has been a star in the Chinese music industry for decades and knows how to present herself in the public. A more recent story came from APEC last year when the Russian president Putin was seen to stand up, gracefully placing a blanket around the shoulders of Chinese First Lady, too gentleman an act that triggered waves of online comments.
Using our own text mining tool, we collected one year Chinese social media data to see what the public image looks like for the First Lady. Overwhelming praises and admiration, on her grace, intelligence and personality, with almost no negative comments. The only eye-catching criticism that was uncovered involves early days of Peng Liyuan "wearing fat army trousers (穿肥大的军裤)", which seems not to be something that agrees with first lady's image in people's mind. (It turned out that this was a story about the First Lady's dating the president long ago when she wanted to test the present if he was only attracted to her appearance by wearing not as nice on purpose. The story got spread all over the net.) But look at the Photo News today, First Lady is now leading the fashion trend of China.
I showed the First Lady's news pictures to my daughter. Tanya was so intrigued, "Dad, Mom told me that you used to teach First Lady many years ago, is that true?" "It is true, but that was only a short time, one or two semesters, and it was not her major subject. As a part-time lecturer, I was teaching Advanced English to graduate students in the music conservatory and she happened to be one in my class. She was already famous then as a new star for folk songs." Tanya got excited, "Well, you never know, maybe her English training in graduate school helps her in state visits today. My Dad is cool." She continued, "Dad, Mom also told me that you were interpreter for foreign minister when she dated you, is that true?" "Well, that was largely an accident, only happened once when I substituted some professor to act as interpreter for the former foreign minister and former Chinese congresss vice-chairman Mr. Huang Hua. Your Mom agreed to date me partially because of her seeing a picture of me interporeting for Mr. Huang. So I guess I benefited from that 'accident'." Tanya was amused and felt very proud, "I have the coolest Dad in the world. He was so successful even when he was young, teaching future first lady and interpreting for the then foreign minister. Wow"
不过最近克林顿的选情是原地踏步，并没有明显进展。比较克林顿的三个圈可知，最淡的圈是过去30天的前10天，明显落后于川普，后两个圈是最近20天，基本原地，只是圈子变大了，说明竞选的投入和力度加大了，但效益并不明显。而从川普方面的三个圈圈看趋势，这老头儿实际的总体趋势是下跌，过去三十天，中间的十天舆情有改观，但最近的十天又倒回去了，虽然热议度有增长。（MD，这个分析没法细做，越做越惊心动魄，很难保持平和的心态，可咱是 data scientist 啊。朋友说，“就是要挖点惊心动魄的”，真心唯恐天下不乱啊。）看看川普的30天社煤的褒贬云图（Word Cloud for pros and cons）和情绪云图（Word Cloud for emotions）吧：
"But the entrepreneur admitted that there were limitations to the data in that sentiment around social media posts is difficult for the system to analyze. Just because somebody engages with a Trump tweet, it doesn't mean that they support him. Also there are currently more people on social media than there were in the three previous presidential elections."
haha，同行是冤家，他的AI能比我自然语言deep parsing支持的 I 吗？从文中看，他着重 engagement，这玩意儿的本质就是话题性、热议度吧。早就说了，川普是话题大王，热议度绝对领先。（就跟冰冰一样，话题女王最后在舆情上还是败给了舆情青睐的圆圆，不是？）不是码农相轻，他这个很大程度上是博眼球，大家都说川普要输，我偏说他必赢。两周后即便错了，这个名已经传出去了。川普团队也会不遗余力帮助宣传转发这个。
现在的纠结是，【大数据告诉我们，希拉里选情告急】，到底发还是不发？为了党派利益和反川立场，不能发。长老川志气，灭吾党威风。为了 data scientist 的职业精神，应该发。一切从数据和事实出发，是信息时代之基。中和的办法是，先发一篇批驳那篇流传甚广的所谓印度AI公司预测川普要赢，因为那一篇的调查区间与我此前做的调查区间基本相同，那是希拉里选情最好的一个月，他们居然根据 engagement alone 大嘴巴预测川普的胜选，根本就没有深度数据的精神，就是赌一把而已。也许等批完了伪AI，宣扬了真NLU，然后再发这篇 【大数据告诉我们，希拉里选情告急】
FBI director 说这次重启调查，需要很长时间才能厘清。现在只是有了新线索需要重启，不能说明希拉里有罪无罪。没有结论前，先弄得满城风雨，客观上就是给选情带来变数。虽然在 prove 有罪前，都应该假定无罪，但是只要有风声，人就不可能不受影响。所以说这个时间点是最关键的。如果这次重启调查另有黑箱，就更惊心动魄了。如果不是有背后的黑箱和势力，这个时间点的电邮门爆炸纯属与新线索的发现巧合，那就是希拉里的运气不佳，命无天子之福。一辈子强性格，卧薪尝胆，忍辱负重，功亏一篑，无功而返，保不准还有牢狱之灾。可以预测，大选失败就是她急剧衰老的开始。
记得他在博文上好几次提到了我，甚至说我是几千年出一个的伟人。哈哈，连我这样的也快被捧晕了。也许中文对人评价的calibration不同？这倒是个值得研究的语言现象。不管怎样，多一个崇拜者总是让人高兴的事。因为我有自知之明，所以我相信他的几百万读者中至少有８０％是第一次听到我的名字。所以我的知名度也因他而提高吧。如果他的读者中有１０％buy his statement，那我就征服了又一个国度了。
李：50步 100步。扭曲现实 与 虚拟现实。不过 trigger 不同罢了。反正都不是现实。更要命的问题是 谁也不知道现实的本尊是什么。甚至不知道是不是有恒定的真实性存在。如果现实本身就有不可知 或 测不准的特性，那么甚至扭曲都谈不上。扭曲的前提是假设有一个绝对的黄金标准或对象。就是这位教授写了本一定影响的书 书名就是 The Case against Reality：反现实辩。
李：化学家也可以鄙视物理学起码是宇宙学，说他们不接“地”气。生物学鄙视化学可以说他们不以人为本。语言学鄙视一下生物学可以拿智能的载体说事，以人为本的前提是以会说话的动物作为对象。说话（speech） is everything：这一点不仅有全才马克思背书，连当代哲学家明星赫拉利也强调再四。更不用说背后的乔姆斯基巨人身影了。
爱因斯坦把时空统一成了 SPACE TIME 的四维模型，说明了时空不可分割、时空相互作用，甚至时空弯曲等等，但实际上还是给了时间一个特殊的位置。（其他物理学家的确有人要取消这个特殊位置，认定时间与空间维度一样可以倒流。）后来的弦论（String Theory）主张10+1维度的宇宙模型，也是把时间打入另类。后加的7个维度总是在空间的延长线上做文章，visualize 为一个微观世界中增加的维度。反正微观看不见，在想象中可以任意增加，只要能凑出一个大一统理论（theory of everything）就行。
Chang uses the video to clarify the difference between TACC and Autopilot. TACC controlled the vehicle’s speed solely and did not have any warning apart from an icon on the central infotainment screen. Activating it demanded just a tap down on the right lever. Autopilot controls the steering and the speed and demands two taps on the same lever.
The Chinese government demanded Tesla to fix the operation of TACC because it could have a higher speed than the one in which the car was going. If the driver tapped down the right lever by accident—such as in a sharp turn—the vehicle could accelerate and lead to a "misjudgment of vehicle control" and crashes.
可以想象的解决办法是允许用户做基于地理定位的针对性配置，譬如根据GPS记忆某处的处理方式需要与模型不同。主要是老马压得太紧，很多时候是现在顾不上这些，不是不能实现。例如，高档特斯拉的空气 suspension 系统就可以根据GPS记忆实现用户指定的可调控的高度，增加驾驶舒适度，但 Model 3 和 Y 没有空气悬挂，所以开车很硬，颠簸厉害。
说到特斯拉小三和Y的短板，颠簸绝对算一条。这有个故事。我家特斯拉起名叫 big white （大白）。原来是刚开始买的基本型特斯拉模型3 起名叫 Xiaobai（小白），开了两天觉得颠簸，三天内无条件退换，就去换了辆双马达长续航。其实长得完全一样，尺寸不变，但加了几千块钱，于是叫大白。其实也依然颠簸，开车很硬。好路没感觉，遇到路况不好（加州公路常常失修，路段质量无法恭维），想起来就跟当年学手扶拖拉机驾驶时候的感觉一样。这毛病没治，特斯拉只有豪华版车型 S 和 X 才有空气悬挂，开起来才舒服。但人很奇怪 有心理因素 加了几千块钱 就似乎觉得不那么颠簸了。马斯克本人对此心知肚明，建议顾客不要把轮胎的气放一放，说这样就不会那么硬，瞧，这出的都是什么臭主意：轮胎气不足不仅耗电，降低续航，还有很多其他问题甚至危险。老马曾经一度说要给小三添加空气悬挂，结果回去一合计觉得成本压力太大，自食其言。
A Taxonomy of AV Myths
Myth: AV software is a singular “AI” that simply learns to drive over time.
Myth: Creating an AV is just a matter of collecting a large amount of data and putting it into a neural net.
Myth: The company with the most data is necessarily in the lead.
Myth: It’s possible to enumerate every situation an AV will ever encounter. Creating an AV is then just a matter of experiencing each possible situation one time and adding it to the data set.
Myth: An AV can only use its sensors, or its map. It cannot use both at the same time. If the sensor input disagrees with the map, it is an irreconcilable problem and the AV can’t work anymore.
Myth: Mapping is extremely expensive (billions of dollars) and/or time-consuming (years).
Myth: If an AV uses a map, then it cannot handle construction zones or other changes.
Myth: If an AV uses a map, then it cannot ever operate in unmapped areas.
Myth: If an AV uses a map, then it’s just a tram running on virtual rails.
Myth: Roads are designed for vision, so other sensor modalities like lidar and radar are useless.
Myth: Lidar uses a huge amount of power.
Myth: Lidar is so power-hungry that it can’t be used on a battery electric vehicle.
Myth: Computer vision is as good as lidar, so lidar is useless.
Myth: Humans drive just fine with two eyeballs, so other sensor modalities (radar and lidar) are useless.
Myth: Once your AV sort of works for a few miles at a time, it’s an easy process to improve it to superhuman reliability. It’s just the March of Nines, which requires nothing but time, or more data.
Myth: If an AV completes a trip without an intervention or disengagement, then it was L4 (or L5) for that trip.
Myth: It is possible for individuals to observe an AV system over the course of their typical personal driving needs and declare the system universally safe.
Myth: The trolley problem is of fundamental importance to the design of AVs.
Myth: AVs can’t work unless we put sensors or beacons in all the roads, and/or make all the cars talk to each other wirelessly.
Myth: Simulations are useless.
Myth: It’s all just a matter of finishing the software.
Myth: Regulations are the only thing holding back AVs.
Myth: Humans have to be sacrificed today in order for an AV system to improve, so it will eventually save lives.
Myth: This is a taxonomy, and not just a litany.
特斯拉一家抵抗一个世界，唯一貌似站得住的 argument 是说包含了激光雷达和高清地图的路线难以 scale up，但是它忘了，所有的其他厂家的自动驾驶系统都是真正的 hybrid AI，并不是只靠激光雷达和高清地图，他们一样用计算机视觉 CV。不能 scale up 的指责，在激光雷达太过昂贵难以量产普及的时候，还有些道理。现在，这种指责根本就站不住。混合系统在有高清的时候用高清（譬如大中城市 繁忙路段 downtowns），在没有高清的时候 backoff 到 CV，这怎么不可行 怎么就不能 scale up 呢？
guo：还真不是。这就像我们常说的深度学习与知识系统“并举”。反正终结围棋的是个纯粹的极端。
wei：嗯，这算是一个正例。围棋这种 game 与自驾的应用不好比 不一定是 apples to apples
视觉本质上是局限性很强的。特斯拉最常用的 argument 说人就是靠视觉驾驶 道路交通及其信号系统也主要是根据人的视觉来设计的。这个 argument 太弱 门槛设的太低，因为人的驾驶实在无法恭维，不应该是自动驾驶的标杆：每年太多交通事故了，一直高居人类杀手前三。绝对不是我们的理想目标。最多能做基线。
guo：看看特斯拉对拿掉 雷达 的解释吧。马斯克的提问：两者矛盾了听谁的？
wei：这正是我要 address 的：一个新近的 argument 是说 信息融合不好做 马斯克问：如果雷达和视觉产生矛盾 听谁的？这个 argument 不无道理，但已经退到了算法的水平之争了。Hybrid AI 当中的 information fusion 一直都有各种挑战，也一直都有不断进步。这个argument 不是致命的，因为 融合 有各种算法，除了紧耦合外，还可以松耦合，可以架构为 backoff，特斯拉没有办法对主流自动驾驶一剑锁喉。我的预测，最终不是自驾主流去迁就特斯拉，而是特斯拉拥抱和回归主流。其中，已经越来越平价化的激光雷达会是自动驾驶之重器：产业界对此市场的布局已经开始，中美都会有以激光雷达为核心业务的独角兽崛起。
When I was at college in Class 77, we were lucky to have Miss Wang as our teacher. Most of us did not know the difference between "ok/alright" and "Yes" when answering a request. One day, Miss Wang asked Student A:
"Would you do me a favor ...?"
A replied, "Ok."
Miss Wang asked Studet B the same question and B said: "Alright".
Miss Wang lost her patience:
"Can you guys be more positive?"
We were still lost, thinking that we were all positive.
Finally, she pointed out: "can you say something like 'Yes, I will'?"
王老师是我的恩师。本科阶段,师资紧缺,幸亏有王老师的加入。还记得20多年前的一个笑话。有一次，王老师重感冒，带病上课，打喷嚏不止，忍不住小声咕哝道：“Such nuisance！” 我坐前排，轻声回应道:”It’s really not a new-sance. It’s an old ’sance’. You have caught cold for days now.” (顺便一提,在西方,别人打喷嚏时最合适的话应该是,”Bless you!”) 王老师本想批评我不认真听讲,但终于忍俊不住笑了。同学中有听到的跟着笑, 有没听见的觉得莫名其妙。
王老师是我的恩师。本科阶段,师资紧缺,幸亏有王老师的加入。还记得20多年前的一个笑话。有一次，王老师重感冒，带病上课，打喷嚏不止，忍不住小声咕哝道：“Such nuisance！” 我坐前排，轻声回应道:”It’s really not a new-sance. It’s an old ’sance’. You have caught cold for days now.” (顺便一提,在西方,别人打喷嚏时最合适的话应该是,”Bless you!”) 王老师本想批评我不认真听讲,但终于忍俊不住笑了。同学中有听到的跟着笑, 有没听见的觉得莫名其妙。一晃30年了,祝愿王老师健康快乐,长寿无疆。
Is there a way, whatever, any such,
in all the world existing, to unlade,
unlock, unlatch, enliven by a touch
magical the long-lonely weighed-
down soul, that, waking like an apple tree
frost-fettered, now reviving, trembling,
with wind and rain rollicking in glee,
it brightens to a loveliness of spring?
Yes, there is one, one only, one alone –
not love uncomprehending, singeing, blind,
self-happy-seeking; but there is a known
way, thing, fine thing, consoling, gracious, kind –
What? Sympathy! Reverberating wide
and wider, when the asking is replied.
１９４２年，我插班进入金女大英文系。当时，金女大、燕京、南京、齐鲁四所教会大学，已先后迁至位于成都的华西大学宽阔的校园内。各校英文系的学生都可选听任何一校英文教授开的专业课，”资源共享”。南大英文系主任 Dr. William P. Fenn 开的”英诗概论”和燕大教授 Grace Boynton 开的”现代诗歌”课（注：２００３年１１月３日《中华读书报》所载章开沅的”教会大学在中国”文中，曾评价并引用过 Fenn 的言论。Boynton 的名字也曾在几年前”博览群书”某文中被提起过），都讲授得十分精彩。这就引发了我自中学始即跃跃欲试的”大胆妄为”的写诗冲动，照着我特别喜爱的一些诗的语言风格，”依样画葫芦”，课余逐渐写了几十首。后来，精选出１６首十四行，作为我的毕业论文（Dr. Fenn 就是我的导师），居然引起华西大校园中许多外籍教师的瞩目。
不过，”万事开头难”。一个小插曲至今留在记忆中，鲜活常青。Miss Boynton 除教”现代诗歌”外，还开过英文作文课。１９４３年某日，下课后我拿着自己的几首诗请她指教。她微笑着接过去，却不紧不慢、温和地说：”呀，你们中国学生能把英文句子写通就不错了，还写诗呀！”但诗页既已递到她手，我也不便再要回来。这样，我的”处女作”，就进入了她的审读视线。再次课余见面，Boynton 态度大变，她惊喜地评说了两三句。此后，她曾邀请我到她独居的小楼去喝过一二次午茶，谈天、说诗。作文课上，她常向全班朗读我写的散文。再后，华西大各校外籍教师中曾传来过对我实际上很幼稚（尤其在意韵方面）的作品的溢美之辞，比如说，”She’ll burst into print any time”……
想不到的是：那竟成了我此生仅有的一段原创性英文诗写作的实践了！至于”burst into print” (应该说, struggle into print) ，至少在有关英诗方面，那是在整整半个世纪以后，当陪衬”运动员”和沦为囚犯１９载、最终幸逢拨乱反正以后的事了。－－也就是你帮忙上网介绍的那本《古今爱国抒情诗词选》的出版之时的事了。
当年暑假在师院，我常跟王老师一起吃早饭，听她谈她一生的坎坷和传奇。她最怀念的是大学时代和她大学时代所写的英诗，当时出类拔萃，才气横溢，极受教授和校长器重。这一篇小诗就是当年她给我回忆记录下来的片断。还记得王老师跟我讲解诗句“He sits and reads the pool”时，说sits是短元音，reads是长元音，相互配合使得诗句抑扬顿挫。并做出鸟儿神情专著盯着池塘的神态，作为此句的演示。
王知还老师早期的另一首诗（不全）
The Kingfisher
He is a flame of emerald,
With a shining ruby breast,
Of all things most beautiful,
By Mother Nature drest;
Inhabits the deep woodland green,
By some water calm and cool,
And on a wayward stretching bough,
He sits and reads the pool.
Then darts he like a light’ning flash
With winning fleeting grace,
Dips his winglets in the pool,
And wrinkles its halcyon face.
Then the flame of emerald
Has vanished out of sight.
相见时难别亦难，东风无力百花残。
春蚕到死丝方尽，蜡炬成灰泪始干。
………
UNTITLED by Li Shang-yin
So hard for us to meet,
Harder still to part.
Languid though the east wind,
Faded flowers are blown apart.
The silkworm’s silk is exhausted
Only when its life is spent;
The candle’s tears are dried,
When itself to cinder’s burnt.
Successively, the battlefield,
The prison, I have all gone through;
And these have not deprived me yet
Of my pure, music-like virtue.
Not e’en with Death confronting me
Have I lost heart or bowed my head,
How, then, could poverty be brought
To bear on me, my nerves to shred?
Twenty ounces of cotton wool
Between tattered sheets for a quilt,
A virtually empty pot above
A fire with three pinewood sticks built;
E’en so, I do feel I have passed
A rigorous winter fine and swell,
And scorn the tenet that the world
Holds trump cards o’er us each and all!
A thunderbolt cracks out of the sky,
The earth collapses as though lost its stay.
All silence – the nation at once appears,
Dumb-founded, speechless, all in tears!
To tell the news, No one his voice could find,
About to speak, sobs shove the words behind.
Having heard, no one dares believe,
Believing, the heart’s broken in grief.
（四）
素纸黑纱含恸剪，苍松翠柏和泪扎。
谁言献花是旧俗，明朝她死定无花。
White tissue, black armbands,
In anguish we cut and wind.
green pine and cypress branch,
With tears we fondly bind.
Who says it’s a mode outdated
To offer flowers to those we admire?
Wait till she dies, surely she’s fated
To have no flowers at her pyre!
关于本诗集的资讯：
《古今爱国抒情诗词选》SELECTED LYRICS ON THEMES OF PATRIOTISM AND MORAL INTEGRITY(FROM
ANCIENT TO MODERN CHINESE CLASSIC POETRY) / 王知还 选译. - 北京: 中国对外翻译出版公司, 1995；239页;
19CM; 汉英对照；ISBN 7-5001-0378-6
……
Since I cannot read Chinese, I cannot comment on the accuracy of the translation.
But what I have read, I like it very much. The English translation reads
very well and worthy of publishing. Since this work is conceived as a
bilingual manuscript, I would very much like to see it published in Chinese
and English. This would entail certain problems. I recommend that you write
to the organization below, to ask for publishers who might be interested
in printing this book in two languages.
……
You might like to send a sample of the poems as well. You could mention
that their address was recommended by myself who is a friend of Luo Qing,
the Chinese painter and publisher.
……
it is our advice that the author directly contact UNESCO since they at
times provide financial support for publication of this sort. She should
contact
Mr. Edouard Maunick
Chief, Cultural Exchanges
UNESCO
……
With persistence, we are sure she would find a suitable publisher for it.
M.K.Jain
President
Asian Humanities Press
Santa Clara, California
May 30, 1988
冤案之源
