從iPhone、汽車到香蕉的貿易之旅

 曹嬿恆、簡萓靚譯，從iPhone、汽車到香蕉的貿易之旅：一本破解關於貿易逆差、經貿協定與全球化迷思，商周出版，2020

Fred P. Hochberg, Trade Is Not a Four-Letter Word: How Six Everyday Products Explain Global Trade—And Destroy the America First Myth, Avid Reader Press, 2020.

iPhone如何推翻美中「貿易逆差」的說法？

本田休旅車居然比福特和通用汽車更「美國」？(*)

失去國外市場，《冰與火之歌》甚至整個好萊塢都會消失殆盡？

塔可沙拉、本田汽車、香蕉、iPhone、大學文憑、《冰與火之歌》，

用六項商品的故事，逐步解開貿易的謎團！

貿易決定了我們餐桌上的選項，決定了購物的所有價格，

決定了哪家工廠將會慘澹關門，決定了哪個集團將統御世界；

但對於這個市場依舊存在著太多誤解，包括：

‧中國是糟糕的貿易夥伴？

‧貿易赤字代表國家嚴重損失？

‧關稅是外國人要付的？

‧進口品買得愈少，我們的日子過得愈好？

其實，貿易赤字根本無法反映雙邊的經濟狀況，

甚至會因通貨升值與貶值而波動，

而且商品價值只計入最終完成組裝的供應商所在國家，

使得號稱美國設計發明、風靡世界的iPhone，

身上所有瑞士陀螺儀、日本視網膜螢幕、以及美國玻璃的價值，

完全計入中國經濟！

 (*) pdf.

產品化物件偵測技術

徐宏民，產品化物件偵測技術 (一)，電子時報，2021-04-07

幾十年來電腦視覺研究試著在這關鍵的物件偵測技術上帶來突破。可以想像一下，電腦如何在由一堆影像畫素值中標定可能的物件？框列出可能位置，再逐一判斷是否有物件存在，是工程上「較容易」實現的方式。一般而言有三個主要步驟：候選區域(region proposal)計算、物件分類、以及後處理。 

徐宏民，產品化物件偵測技術(二)，電子時報，2021-04-13

最關鍵的問題是正確率。正確率的描述非常籠統，一般我們會更細分為precision(P)以及recall(R)，前者代表所回報的物件中有多少比例是正確的，比如說畫面中框列了10輛車子，有幾輛是對的；後者代表實際的物件標的中找到多少比例，例如畫面中有10輛車子，實際框列了幾台。...

我們很難設計單一演算法P跟R都是完美無缺。一般在檢測環境(AOI、自駕時)中比較在乎recall，所以會刻意將所有可能物件挑出，但是會造成P下降(多了假警報)，解決方法是接續使用其他演算法再進行過濾，剔除誤判，或是利用其他訊號源再確認，比如說使用雷達訊號標定可能物件之後，再使用攝影機辨認是否為車輛。

在某些應用中比較在乎precision，可以犧牲recall。例如搜尋系統中。尋找大量照片時，因為使用者不清楚有多少真實標的(例如：狗)存在，我們只需將有把握的標的呈現出來，並按照信心度排序，就能滿足使用者的需要。一般推薦系統也是採用這樣的策略，確保使用者的滿意度。...

解決anchor在實際場域上的限制，可以試著修改或是增減需要的anchor種類。不過另一種常見的作法是直接使用anchor-free的策略(如FCOS)，不使用預設模板，在偵測時，以某個基準點，往外推估可能物件的長寬，在實際使用上有不錯的效能。

Machine Learning Faces a Reckoning in Health Research

Megan Scudellari, Machine Learning Faces a Reckoning in Health Research, IEEE Spectrum, 29 Mar 2021.

In a paper describing her team’s analysis of 511 other papers, Ghassemi’s team reported that machine learning papers in healthcare were reproducible far less often than in other machine learning subfields. The group’s findings were published this week in the journal Science Translational Medicine. And in a systematic review published in Nature Machine Intelligence, 85 percent of studies using machine learning to detect COVID-19 in chest scans failed a reproducibility and quality check, and none of the models was near ready for use in clinics, the authors say.

“We were surprised at how far the models are from being ready for deployment,” says Derek Driggs, co-author of the paper from the lab of Carola-Bibiane Schönlieb at the University of Cambridge. “There were many flaws that should not have existed.”

liquefied natural gas (LNG) portfolio optimization

Alessandro Agosta, Dumitru Dediu, Timo Leenman, and Marijn van Diessen, LNG portfolio optimization: Putting the business model to the test, McKinsey, April 12, 2021.

The analysis indicates that in third quarter 2020, traditional marketers saw a 47 percent decline in realized prices compared with third quarter 2019, whereas the portfolio optimizers saw a 31 percent decline in realized prices over the same period (Exhibit 3). Although there are some obvious short­comings in this comparison—for example, the declines could be due to different geographic presences or a time lag on indexation—the findings do suggest that in the gas industry, portfolio opti­mizers were less affected by the recent downturn.

第三代半導體

林宏達，什麼是第三代半導體？一文看懂車用、通訊、充電通吃的殺手級應用，財訊，2021-04-20

(高中) 數學與資訊工程

馬來西亞的學校提案，準備今年 5 月，開授相關線上演講，以便吸引高中生就讀資訊相關科系。去年12月中開校內協調會的時候，學校長官指派我，負責「數學與資訊工程 」。 月底到了，忙著提科技部計畫和撰寫研究的程式，但是，各種點子不斷地進入我的腦袋裡，只好趕緊把它寫下來，不然半夜進入我的夢鄉，擾人清夢。花了兩天，寫下初稿；最近又多次修正，前後花了不下 20 小時，決定提前定稿 ，以便改作其他教學和研究事務。

後來想一想， 既然這是一個有意義的工作，就準備把他錄成影片，並上傳 YouTube，以幫助有需要的年輕人。追求新知並傳授給學生，一直是我當老師快樂的泉源。

歡迎指正和提供寶貴意見。(pdf in 1) (pdf in 4，如果需要列印， 請雙面列印此版本，環保救地球)

初稿 2021/1/15。

Domino's Pizza launched a pilot program with autonomous vehicle provider Nuro

Inside AI, 2021/4/14

Domino's Pizza launched a pilot program with autonomous vehicle provider Nuro and will begin deliveries in Houston this week. The deliveries will be made using Nuro's R2 robots, the first fully autonomous on-road delivery vehicle (without a human safety driver) that's received permission to operate in the U.S.

Dataflow-as-a-Service of SambaNova

Inside AI, 2021/4/14

SambaNova, a developer of AI hardware and software systems, raised $676M in financing that values the startup at over $5.1B. The Nvidia competitor makes chips for AI processes, which it uses to build servers and AI software that it leases to other businesses.

Charting a business course for reinforcement learning

Jacomo Corbo, Oliver Fleming, and Nicolas Hohn, It’s time for businesses to chart a course for reinforcement learning, McKinsey, April 1, 2021.

Broadly speaking, we see reinforcement learning delivering this value across the business, with potential applications in every business domain and industry (Exhibit 2). Some of the near-term applications for reinforcement learning fall into three categories: speeding design and product development, optimizing complex operations, and guiding customer interactions.

 Exhibit 2 some applications.

To be sure, implementing reinforcement learning is a challenging technical pursuit. A successful reinforcement learning system today requires, in simple terms, three ingredients:

1. A well-designed learning algorithm with a reward function. 
2. A learning environment.  
3. Compute power. 

Computing power is better than compute power. 

Feedback Control in Programmatic Advertising

N. Karlsson, Feedback Control in Programmatic Advertising: The Frontier of Optimization in Real-Time Bidding, IEEE Control Systems Magazine, vol. 40, no. 5, pp. 40-77, Oct. 2020.

Feedback control is critical in the scalable optimization of Internet advertising, and it is, therefore, an enabling technology. However, it is challenging to model the plant and design the controller because the plant is nonlinear, time varying, stochastic, and poorly known. A closed-loop system model easily becomes unrealistic or extremely complicated and intractable to analyze.

歡迎光臨人類學 (Invitation to Anthropology)

郭禎麟、吳意琳、黃宛瑜、金家琦、呂環延、方怡潔、黃恩霖譯，歡迎光臨人類學，群學，2010

Luke Eric Lassiter, Invitation to Anthropology, 3rd edition, AltaMira Press, 2009.

種族差異是天生的？民族中心主義的傾向是必然的？性別不平等的文化根源為何？當代的婚姻與家庭該何去何從？不同宗教信仰背景的人能否共處共榮？這些問題背後有著許多的故事，既關乎他者，也關乎我們自己。

本書作者拉斯特深信故事的力量，因此他以說故事的方式，一步步引領讀者進入人類學豐富而多元的故事中。包括人類學的誕生起源、理論視野的開展與研究方法的深化，以及當代重要的跨文化議題。拉斯特提醒我們：即便有著不同的形貌、說著不同的語言，甚至發展出不同的社會價值觀，但同處於一個世界體系中，人們終究彼此相連。唯有更深刻地看見文化的作用、認識人類的同異之處，真正的相互理解與溝通才可能到來。

How we use AutoML, Multi-task learning and Multi-tower models for Pinterest Ads

Ernest Wang, How we use AutoML, Multi-task learning and Multi-tower models for Pinterest Ads, Aug 20, 2020.

People come to Pinterest in an exploration mindset, often engaging with ads the same way they do with organic Pins. Within ads our mission is to help Pinners go from inspiration to action by introducing them to the compelling products and services that advertisers have to offer. A core component of the ads marketplace is predicting engagement of Pinners based on the ads we show them. In addition to click prediction, we look at how likely a user is to save or hide an ad. We make these predictions for different types of ad formats (image, video, carousel) and in context of the user (e.g., browsing the home feed, performing a search, or looking at a specific Pin.)

ICLR 2021 Outstanding Paper Awards

Ivan Titov, Naila Murray, Alice Oh, and Katja Hofmann,  Announcing ICLR 2021 Outstanding Paper Awards, ICLR 2021.

每篇文章都有審稿者的意見。

晶圓製造產能短缺的三個原因

洪友芳，劉德音：產能短缺有3原因 跟集中台灣生產無關，自由時報，2021/03/30

台灣半導體產業協會（TSIA）理事長劉德音指出，目前晶圓製造的產能短缺並非跟在哪裡生產有關，主要來自3個主要原因，疫情導致供應鏈庫存提高成為常態、美中貿易戰使供應鏈與市場占有率轉移的不確定因素導致重複下單、疫情加速數位轉型，前2個原因是暫時性的，疫情加速數位轉型是會持續下去，工作與生活型態改變將會使人們更加依賴科技。...