Home » Blog » Notes

Notes

I regularly take down notes and paste them on the shields of my desk. When the shields are full, I remove some and transcribe the contents here.

04-09-2017

  • The graphical model captures the causal process by which the observed data was generated. For this reason, such models are called generative models.
  • Explain away: to dismiss or minimise the significance of something by explanation.
  • clique: a small close-knit group of people who do not readily allow others to join them.
  • Interquartile range (IQR) = middle 50% = Q3 - Q1
  • Point beyond 1.5IQR mild outlier
  • Point beyond 3IQR extreme outlier
  • 联合熵 joint entropy H(X,Y) = - sum_x sum_y p(x,y)logp(x,y)
  • H(X,Y) >= max{H(X),H(Y)}
  • H(X,Y) ⇐ H(X) + H(Y)
  • H(X,Y) = H(X) + H(Y) if X and Y are independent.
  • H(Y|X) = H(X,Y) - H(X)
  • Conditional entropy 条件熵
  • H(Y|X) entropy of Y conditioned on X
  • H(Y|X) = sum_x p(x) H(Y|X=x) = - sum_x,y p(x,y)logp(y|x)
  • H(Y|X) = 0 if Y completely determined by X
  • H(Y|X) = H(Y) if Y,X are independent
  • Mutual information 互信息
  • I(X;Y) = H(X) - H(X|Y) = H(Y) - H(Y|X) = H(X) + H(Y) - H(X,Y) = H(X,Y) - H(X|Y) - H(Y|X)
  • I(X;Y) = D_KL(p(x,y)||p(x)p(y))
  • Gini(X) = 1 - sum p_i^2
  • Entropy H(X) = -sum p_i logp_i
  • 两者均衡量不确定性,趋势相同,Gini加1保证非负
  • H(Y|X) = H(X|Y) + H(Y) - H(X) ←- Bayes' rule
  • H(Y|X) ⇐ H(Y)
  • H(X,Y) = H(X|Y) + H(Y|X) + I(X;Y)
  • H(X,Y) = H(X) + H(Y) - I(X;Y)
  • I(X;Y) ⇐ H(X)
  • from sklearn.model_selection import train_test_split
  • from sklearn.model_selection import cross_val_score
  • K-Folds Cross Validation
  • Leave One Out Cross Validation
  • inherent conflict: degeneracy of the covariance function is good for computational reasons, but bad for modelling reasons.
  • VIM 快捷键
  • ctrl+d: page down
  • ctrl+u: page up
  • *: next occurance
  • #: prev occurance
  • \+: at least once
  • shift+Enter 五笔
  • dfu-util -a 0 –dfuse-address 0x08000000 -D x.bin

03-08-2017

  • logit: logistic probit, coined by J. Berkson in 1944, inverse of logistic function.
  • odds: p/(1-p)
  • quantile function: inverse of CDF
  • Beta distribution: support x\in[0,1], f(x) = 1/B(a,b)*x^(a-1)(1-x)^(b-1)
  • E[x] = a/(a+b)
  • suitable for random behaviour of percentages and proportions.
  • Binary Search Tree: left subtree smaller, right subtree bigger.
  • Gini impurity: (G as in Gene) I_G = sum_{i=1}^m p_i(1-p_i) = 1 - \sum p_i^2
  • docker exec -it ID bash, run a command (bash) in a running container
  • docker run -it image, run a command in a new container, RUN instead of EXEC
  • docker ps -l, list containers
  • docker commit containerID, create a new image from container's change
  • docker run -e DISPLAY=192.168.xx.xx:0 -v /tmp/.X11-unix:/tmp/.X11-unix:rw osrf/ros:kinetic-desktop rqt
  • docker images
  • docker search
  • numpy matrix multiplication, a.dot(b)
  • element-wise multiplication, a*b
  • comma indicates a tuple should be constructed
  • hostname -i, show IP address

12-07-2017

  • 假设属性之间相互独立
  • Naive Bayes, x = [x_1, \ldots, x_n] feature vector, C_k: class k, p(C_k|x) \prop p(x|C_k)p(C_k).
  • By assuming each feature is conditionally independent of every other one given class C_k, then p(x|C_k) = \prod p(x_i|C_k).
  • :v/pattern/p, vim find lines not matching a pattern
  • sudo iptables -L -n, display firewall status
  • which, whereis, locate programs
  • NP-Hardness: non-deterministic polynomial-time hard, i.e. at least as hard as the hardest problems in NP.
  • ReLU: Rectified linear unit
  • if the posterior distribution p(\theta|x) is in the same family as the prior distribution, then the prior and posterior are called conjugate distributions, and the prior is called the conjugate prior for the likelihood function.

25-04-2017

  • DELETE FROM table WHERE id BETWEEN 10 and 200
  • git remote set-url origin git@github.com:…. if username and password are asked.
  • mount -o remount, rw /
  • mount -o remount, ro /
  • X band, 8-12GHz, 3.75-2.5cm
  • sudo dd bs=4M if=/dev/sdb | gzip > pathtoimage.gz
  • gzip -dc pathtoimage.gz | sudo dd bs=4M of=/dev/sdb
  • sudo dd bs=4M if=/dev/sdb | gzip > pathtoimage.gz
  • gzip -dc pathtoimage.gz | sudo dd bs=4M of=/dev/sdb
 git remote -v
 git remote add upstream https:// ....git
 git fetch upstream
 git checkout master
 git merge upstream/master
 git push origin master
  • COMMAND_LONG(#76)
  • COMMAND_INT(#75)
  • heter- other alter
  • laser pulse time-of-flight(TOF)
  • AP6212 = STM32F205 + BCM43438, 2MB Flash

08-09-2015

  • 2693451744
  • ss -aut4
  • ip.src == 192.168.1.130 and http
  • OPPORTUNIST: A person who exploits circumstances to gain immediate advantage rather than being guided by consistent principles or plans.

Use OpenBUGS in R

library(BRugs)
writeLines(modelString,con="model.txt") 
modelCheck( "model.txt" ) 
modelData(bugsData(list))
modelCompile()
modelGenInits() or ModelInits(bugsInits(initfunc()))
samplesSet(c("",""))
modelUpdate(burnin)
samplesSample("")
samplesStats("")

top

  • a - PID
  • c - absolute path
  • P - CPU
  • M - memory

Two modifications of the PIXHAWK code

  1. /src/modules/px4iofirmware/safety.c safety_button_pressed = 1
  2. commander.c line 1338 if updated safety.safety_switch_available = false
set root=(hd0,7)
linux /vmlinuz-linux root=/dev/sda5
initrd /boot /initramfs-linux.img
boot