[interview] questions - dsindex/blog GitHub Wiki

recent questions
- NLP interviews
algorithm and data structure
- coding test samples
- sorting
  - quick, heap, merge, insertion, radix
  - find top k items : partial sort(using min heap), quick select)
- data structure
  - linked list
  - stack
  - queue, priority queue
  - binary tree
  - binary search tree
  - btree, trie, hash
  - graph
- edit distance
- longest common substring
- string search
  - BM
  - KMP, aho-corasick
  - suffix tree, suffix array
  - segment tree, range minimum query
- viterbi algorithm
- beam search
simple coding
- strcat, str reverse
- tree에서 common ancestor
- atoi
- utf-8 string에서 n-gram 생성
- 1 ~ 10까지 binary tree. 1이 root. main()안에서 함수 없이 tree 구성
machine learning
- 중심극한정리(central limit theorem)?
- why sample mean is unbiased estimator? 어째서 sample mean이 global mean의 unbiased estimator인가?
- bayesian 정리
  - prior, evidence, likelihood, posterior, maximum likelihood, maximum a posteriori
  - likelihood와 probability의 차이점?
- entropy, relative entropy, cross entropy, mutual information
- naive bayes와 hidden markov model의 차이점?
- maximum entropy model과 conditional random field의 차이점?
- hidden markov model을 crf로 표현한다면?
- maximum entropy markov model과 crf의 차이점?
  - label bias?
  - RNN에서 label bias?
- beam search와 viterbi search 차이점
  - best first search는?
- precision, recall, f-measure, accuracy
- support vector machine, decision tree, random forest
- linear algebra
  - eigen vector, eigen value, SVD
- linear/logistic/multinomial logistic regression
- multi layer perceptron
- sigmoid derivative
- softmax
  - softmax derivative
  - what will be happened for large output node?
  - hierarchical softmax and negative sampling
- cost function
- learning slow down problem, why?
- overfitting, underfitting
  - regularization, dropout
- weight initialization
  - why do we prefer sharpened truncated normal distribution?
- hyper parameters
  - epoch(early stoping), schedule learning rate, regularization factor, mini-batch size
- variation of gradient descent
  - Hessian technique(gradient of gradient)
  - momentum-based gradient descent(velocity)
- vanishing gradient, why?
  - relu activation function
- cnn
  - what is convolutional operator?
  - differences b/w conv1d, conv2d, conv3d
  - why does each filters learn different features?
- rnn
- lstm
- attention
- language model
- word2vec
  - objective function의 직관적인 의미 설명? 문제는? negative sampling을 한다면?
  - k-nearest neighbor
    - how to implement?
  - subword information을 이용하려면?
  - lookup-based word embedding과 RNN-based LM word embedding의 차이
other question list
- ace the coding interview every time
- top-10 algorithms for coding interview
- top-10 algorithms in interview questions
- careercup
- numberworks
- deep mind
- amazon
- airbnb
- etc
  - 두개의 string A, B에 대해서 A에만 있는 character를 추출
  - 주어진 string을 역순으로 출력하는 코드 작성
  - BST(Binary Search Tree) traversal(http://minjang.github.io/2016/05/03/finding-subtle-mistake/)
  - quick sort와 heap sort의 차이점 설명. in-place 알고리즘이란?
  - Given a random generator that produces a number 1 to 5 uniformly, write a function that produces a number from 1 to 7 uniformly.
  - Segment a long string into a set of valid words using a dictionary. Return false if the string cannot be segmented. What is the complexity of your solution?
  - Find the second largest element in a Binary Search Tree
  - Find out k most frequent numbers from incoming stream of numbers one the fly
  - What is the optimization problem for a SVM?
  - Implement the pow function.
  - Generate a sorted vector from two sorted vectors.
  - Describe the different classification and prediction models. (k-means clustering, bayesian model, decision trees )
job list
- at ACL 2019
etc