Jigsaw Puzzle, Hidden layers, New SoC, Algorithm basics, HTC
Week35 (09/01)
第一次完成520片的拼圖, 花了4個晚上的時間, 大約12小時吧?!
下次記得顏色鮮明的圖會比較愜意些.
Week36 (09/08)
網路文章
Credit firm Equifax says 143m Americans’ social security numbers exposed in hack
在新聞公開前高層就己賣掉市值1.8m的股票!!
網路資源
Google Developer Documentation Style Guide
Google公佈了內部文件撰寫的格式及注意事項.
Neural Networks and Deep Learning - Week 3
Tuning hidden layer size
iterations: 5000
learning_rate: 1.2
activation: sigmoid
Accuracy for 1 hidden units: 67.5 %
Accuracy for 2 hidden units: 67.25 %
Accuracy for 3 hidden units: 90.75 %
Accuracy for 4 hidden units: 90.5 %
Accuracy for 5 hidden units: 91.25 %
Accuracy for 20 hidden units: 90.0 %
Accuracy for 50 hidden units: 90.25 %
Interpretation:
- The larger models (with more hidden units) are able to fit the training set better, until eventually the largest models overfit the data.
- The best hidden layer size seems to be around n_h = 5. Indeed, a value around here seems to fits the data well without also incurring noticable overfitting.
- You will also learn later about regularization, which lets you use very large models (such as n_h = 50) without much overfitting.
Optional questions:
- What happens when you change the tanh activation for a sigmoid activation or a ReLU activation?
- Play with the learning_rate. What happens?
- What if we change the dataset?
這個case在hidden layers為3,4,5時看起來差異不大, 15層時就有overfit發生.
Week37 (09/15)
每次有新的IC出來暫存器多少會有些改動, 讓tool透過chip id去解決相容性問題, 到最後多少會有些歷史的包袱.
之後再多出個几顆就需要重構, 否則爾後解決相容性的問題會越來越花時間.
C -> python -> javascript -> HTML 這之間資料的處理是個頭痛的問題.
這週就在看spec, 開發新功能及相容性上渡過了, 處理相容性的時間佔了10%左右, 可喜可賀.
Week38 (09/22)
網路文章
The definitive guide on how to use static, class or abstract methods in Python
What is the unbound method in python2 and python3??
decorators for method: static, abstract, class.
網路資源
Sketchpad
可以在browser上直接畫圖的工具, 很多筆刷可以選.
目前是4.1版, export出來的Drawing.shetchpad 似乎不能再import回去, 這檔unzip後是sketch.json看起來像是記錄座標及功能, upload sketch.json也沒有反應.
讀書心得
一般在學習演算法會用到的:
Mathematics
- Algebra: linear equations, invert a matrix, find the rank of a matrix, compute the determinat of a matrix, all of the above can be done with Gaussian elimination.
- Number Theory: Greatest Common Divisor(GCD), Euclidean Algorithm, Congruence & Modulo Operation, Multiplicative Inverse, Chinese Remainder Theorem
- Combinatorics: Binomial Coefficients, Fibonacci Sequence, Closed Form
- Geometry: 2D Vector Operations, Line-Line Intersection, Circumcircle of a Triangle, Area of a Triangle, Area of a Simple Polygon
Data Structures
- Queue & Stack
- Heap & Priority Queue
- Union-Find Structure (Disjoint-set data structure)
- Binary Search Tree (BST)
- Fenwick Tree
- Lowest Common Ancestor (LCA)
Reference: CS 97SI: Introduction to Programming Contests
Week35 (09/29)
網路文章
Google signs agreement with HTC, continuing our big bet on hardware
Motorola redux? Google appears set to buy HTC
HTC員工去Google薪水多50萬？聘書洩底
Oops….Google might buy a dying Android OEM again.
總之看好唱衰的都有…Google也不是第一次買硬體了, 這次向"世界工廠"又邁了一大步.
之前M公司12.5B轉手價2.9B給L公司.
這次H入手價1.1B, 說好不是’acquisition’, 買人不買技術??
讀書心得
Power Verbs in English
Power Verbs for Resume Writing
DROVE, LAUNCHED, JUMP-STARTED, PIONEERED, EXECUTED
SPEARHEADED, CREATED, INITIATED, ACCELERATED, CONSOLIDATED
Action Verbs
Power/Action verbs 要有基本的了解