Hidden Markov Models: Probabilistic Reasoning Over Time

1
Hidden Markov ModelsProbabilistic Reasoning
Over Time

• Natural Language Processing
• CMSC 25000
• February 23, 2006

2
Agenda

• Hidden Markov Models
• Uncertain observation
• Temporal Context
• Recognition Viterbi
• Training the model Baum-Welch
• Speech Recognition
• Framing the problem Sounds to Sense
• Speech Recognition as Modern AI

3
Hidden Markov Models (HMMs)

• An HMM is
• 1) A set of states
• 2) A set of transition probabilities
• Where aij is the probability of transition qi -gt
  qj
• 3)Observation probabilities
• The probability of observing ot in state i
• 4) An initial probability dist over states
• The probability of starting in state i
• 5) A set of accepting states

4
Three Problems for HMMs

• Find the probability of an observation sequence
  given a model
• Forward algorithm
• Find the most likely path through a model given
  an observed sequence
• Viterbi algorithm (decoding)
• Find the most likely model (parameters) given an
  observed sequence
• Baum-Welch (EM) algorithm

5
Bins and Balls

• Blue Blue Red

1 1 1 (0.90.3)(0.60.3)(0.60.7)0.0204
1 1 2 (0.90.3)(0.60.3)(0.40.4)0.0077
1 2 1 (0.90.3)(0.40.6)(0.30.7)0.0136
1 2 2 (0.90.3)(0.40.6)(0.70.4)0.0181
2 1 1 (0.10.6)(0.30.7)(0.60.7)0.0052
2 1 2 (0.10.6)(0.30.7)(0.40.4)0.0020
2 2 1 (0.10.6)(0.70.6)(0.30.7)0.0052
2 2 2 (0.10.6)(0.70.6)(0.70.4)0.0070
6
Answers and Issues

• Here, to compute probability of observed
• Just add up all the state sequence probabilities
• To find most likely state sequence
• Just pick the sequence with the highest value
• Problem Computing all paths expensive
• 2TNT
• Solution Dynamic Programming
• Sweep across all states at each time step
• Summing (Problem 1) or Maximizing (Problem 2)

7
Forward Probability
Where a is the forward probability, t is the time
in utterance, i,j are states in the
HMM, aij is the transition probability,
bj(ot) is the probability of observing ot in
state bj N is the max state, T is the last time
8
Forward Algorithm

• Idea matrix where each cell forwardt,j
  represents probability of being in state j after
  seeing first t observations.
• Each cell expresses the probability
  forwardt,j P(o1,o2,...,ot,qtjw)
• qt j means "the probability that the tth state
  in the sequence of states is state j.
• Compute probability by summing over extensions of
  all paths leading to current cell.
• An extension of a path from a state i at time t-1
  to state j at t is computed by multiplying
  together i. previous path probability from the
  previous cell forwardt-1,i, ii. transition
  probability aij from previous state i to current
  state j iii. observation likelihood bjt that
  current state j matches observation symbol t.

9
Forward Algorithm

• Function Forward(observations length T,
  state-graph) returns best-path
• Num-stateslt-num-of-states(state-graph)
• Create path prob matrix forwardinum-states2,T2
  
• Forward0,0lt- 1.0
• For each time step t from 0 to T do
• for each state s from 0 to num-states do
• for each transition s from s in
  state-graph
• new-scorelt-Forwards,tats,sbs(ot)
  
• Forwards,t1 lt- Forwards,t1new-score
  

10
Viterbi Algorithm

• Find BEST sequence given signal
• Best P(sequencesignal)
• Take HMM observation sequence
• gt seq (prob)
• Dynamic programming solution
• Record most probable path ending at a state i
• Then most probable path from i to end
• O(bMn)

11
Viterbi Code
Function Viterbi(observations length T,
state-graph) returns best-path Num-stateslt-num-of-
states(state-graph) Create path prob matrix
viterbinum-states2,T2 Viterbi0,0lt- 1.0 For
each time step t from 0 to T do for each state
s from 0 to num-states do for each
transition s from s in state-graph
new-scorelt-viterbis,tats,sbs(ot)
if ((viterbis,t10) (viterbis,t1ltnew-
score)) then viterbis,t1 lt-
new-score back-pointers,t1lt-s Backtrace
from highest prob state in final column of
viterbi return
12
Modeling Sequences, Redux

• Discrete observation values
• Simple, but inadequate
• Many observations highly variable
• Gaussian pdfs over continuous values
• Assume normally distributed observations
• Typically sum over multiple shared Gaussians
• Gaussian mixture models
• Trained with HMM model

13
Learning HMMs

• Issue Where do the probabilities come from?
• Solution Learn from data
• Trains transition (aij) and emission (bj)
  probabilities
• Typically assume structure
• Baum-Welch aka forward-backward algorithm
• Iteratively estimate counts of transitions/emitted
  
• Get estimated probabilities by forward computn
• Divide probability mass over contributing paths

14
Learning HMMs

• Issue Where do the probabilities come from?
• Supervised/manual construction
• Solution Learn from data
• Trains transition (aij), emission (bj), and
  initial (pi) probabilities
• Typically assume state structure is given
• Unsupervised
• Baum-Welch aka forward-backward algorithm
• Iteratively estimate counts of transitions/emitted
  
• Get estimated probabilities by forward computn
• Divide probability mass over contributing paths

15
Manual Construction

• Manually labeled data
• Observation sequences, aligned to
• Ground truth state sequences
• Compute (relative) frequencies of state
  transitions
• Compute frequencies of observations/state
• Compute frequencies of initial states
• Bootstrapping iterate tag, correct, reestimate,
  tag.
• Problem
• Labeled data is expensive, hard/impossible to
  obtain, may be inadequate to fully estimate
• Sparseness problems

16
Unsupervised Learning

• Re-estimation from unlabeled data
• Baum-Welch aka forward-backward algorithm
• Assume representative collection of data
• E.g. recorded speech, gene sequences, etc
• Assign initial probabilities
• Or estimate from very small labeled sample
• Compute state sequences given the data
• I.e. use forward algorithm
• Update transition, emission, initial probabilities

17
Updating Probabilities

• Intuition
• Observations identify state sequences
• Adjust probability of transitions/emissions
• Make closer to those consistent with observed
• Increase P(ObservationsModel)
• Functionally
• For each state i, what proportion of transitions
  from state i go to state j
• For each state i, what proportion of observations
  match O?
• How often is state i the initial state?

18
Estimating Transitions

• Consider updating transition aij
• Compute probability of all paths using aij
• Compute probability of all paths through i (w/
  and w/o i-gtj)

i
j
19
Forward Probability
Where a is the forward probability, t is the time
in utterance, i,j are states in the
HMM, aij is the transition probability,
bj(ot) is the probability of observing ot in
state bj N is the max state, T is the last time
20
Backward Probability
Where ß is the backward probability, t is the
time in sequence, i,j are states in
the HMM, aij is the transition probability,
bj(ot) is the probability of observing ot
in state bj N is the final state, and T is the
last time
21
Re-estimating

• Estimate transitions from i-gtj
• Estimate observations in j
• Estimate initial i

22
Speech Recognition

• Goal
• Given an acoustic signal, identify the sequence
  of words that produced it
• Speech understanding goal
• Given an acoustic signal, identify the meaning
  intended by the speaker
• Issues
• Ambiguity many possible pronunciations,
• Uncertainty what signal, what word/sense
  produced this sound sequence

23
Decomposing Speech Recognition

• Q1 What speech sounds were uttered?
• Human languages 40-50 phones
• Basic sound units b, m, k, ax, ey, (arpabet)
• Distinctions categorical to speakers
• Acoustically continuous
• Part of knowledge of language
• Build per-language inventory
• Could we learn these?

24
Decomposing Speech Recognition

• Q2 What words produced these sounds?
• Look up sound sequences in dictionary
• Problem 1 Homophones
• Two words, same sounds too, two
• Problem 2 Segmentation
• No space between words in continuous speech
• I scream/ice cream, Wreck a nice
  beach/Recognize speech
• Q3 What meaning produced these words?
• NLP (But thats not all!)

25
(No Transcript)
26
Signal Processing

• Goal Convert impulses from microphone into a
  representation that
• is compact
• encodes features relevant for speech recognition
• Compactness Step 1
• Sampling rate how often look at data
• 8KHz, 16KHz,(44.1KHz CD quality)
• Quantization factor how much precision
• 8-bit, 16-bit (encoding u-law, linear)

27
(A Little More) Signal Processing

• Compactness Feature identification
• Capture mid-length speech phenomena
• Typically frames of 10ms (80 samples)
• Overlapping
• Vector of features e.g. energy at some frequency
• Vector quantization
• n-feature vectors n-dimension space
• Divide into m regions (e.g. 256)
• All vectors in region get same label - e.g. C256

28
Speech Recognition Model

• Question Given signal, what words?
• Problem uncertainty
• Capture of sound by microphone, how phones
  produce sounds, which words make phones, etc
• Solution Probabilistic model
• P(wordssignal)
• P(signalwords)P(words)/P(signal)
• Idea Maximize P(signalwords)P(words)
• P(signalwords) acoustic model P(words) lang
  model

29
Language Model

• Idea some utterances more probable
• Standard solution n-gram model
• Typically tri-gram P(wiwi-1,wi-2)
• Collect training data
• Smooth with bi- uni-grams to handle sparseness
• Product over words in utterance

30
Acoustic Model

• P(signalwords)
• words -gt phones phones -gt vector quantizn
• Words -gt phones
• Pronunciation dictionary lookup
• Multiple pronunciations?
• Probability distribution
• Dialect Variation tomato
• Coarticulation
• Product along path

0.5
0.5
0.5
0.2
0.5
0.8
31
Pronunciation Example

• Observations 0/1

32
Acoustic Model

• P(signal phones)
• Problem Phones can be pronounced differently
• Speaker differences, speaking rate, microphone
• Phones may not even appear, different contexts
• Observation sequence is uncertain
• Solution Hidden Markov Models
• 1) Hidden gt Observations uncertain
• 2) Probability of word sequences gt
• State transition probabilities
• 3) 1st order Markov gt use 1 prior state

33
Acoustic Model

• 3-state phone model for m
• Use Hidden Markov Model (HMM)
• Probability of sequence sum of prob of paths

0.3
0.9
0.4
Transition probabilities
0.7
0.1
0.6
C3 0.3
C5 0.1
C6 0.4
C1 0.5
C3 0.2
C4 0.1
C2 0.2
C4 0.7
C6 0.5
Observation probabilities
34
ASR Training

• Models to train
• Language model typically tri-gram
• Observation likelihoods B
• Transition probabilities A
• Pronunciation lexicon sub-phone, word
• Training materials
• Speech files word transcription
• Large text corpus
• Small phonetically transcribed speech corpus

35
Training

• Language model
• Uses large text corpus to train n-grams
• 500 M words
• Pronunciation model
• HMM state graph
• Manual coding from dictionary
• Expand to triphone context and sub-phone models

36
HMM Training

• Training the observations
• E.g. Gaussian set uniform initial mean/variance
• Train based on contents of small (e.g. 4hr)
  phonetically labeled speech set (e.g.
  Switchboard)
• Training AB
• Forward-Backward algorithm training

37
Does it work?

• Yes
• 99 on isolated single digits
• 95 on restricted short utterances (air travel)
• 80 professional news broadcast
• No
• 55 Conversational English
• 35 Conversational Mandarin
• ?? Noisy cocktail parties

38
N-grams

• Perspective
• Some sequences (words/chars) are more likely than
  others
• Given sequence, can guess most likely next
• Used in
• Speech recognition
• Spelling correction,
• Augmentative communication
• Other NL applications

39
Corpus Counts

• Estimate probabilities by counts in large
  collections of text/speech
• Issues
• Wordforms (surface) vs lemma (root)
• Case? Punctuation? Disfluency?
• Type (distinct words) vs Token (total)

40
Basic N-grams

• Most trivial 1/tokens too simple!
• Standard unigram frequency
• word occurrences/total corpus size
• E.g. the0.07 rabbit 0.00001
• Too simple no context!
• Conditional probabilities of word sequences

41
Markov Assumptions

• Exact computation requires too much data
• Approximate probability given all prior wds
• Assume finite history
• Bigram Probability of word given 1 previous
• First-order Markov
• Trigram Probability of word given 2 previous
• N-gram approximation

Bigram sequence
42
Issues

• Relative frequency
• Typically compute count of sequence
• Divide by prefix
• Corpus sensitivity
• Shakespeare vs Wall Street Journal
• Very unnatural
• Ngrams
• Unigram little bigrams colloc trigramsphrase

43
Evaluating n-gram models

• Entropy Perplexity
• Information theoretic measures
• Measures information in grammar or fit to data
• Conceptually, lower bound on bits to encode
• Entropy H(X) X is a random var, p prob fn
• E.g. 8 things number as code gt 3 bits/trans
• Alt. short code if high prob longer if lower
• Can reduce
• Perplexity
• Weighted average of number of choices

44
Entropy of a Sequence

• Basic sequence
• Entropy of language infinite lengths
• Assume stationary ergodic

45
Cross-Entropy

• Comparing models
• Actual distribution unknown
• Use simplified model to estimate
• Closer match will have lower cross-entropy

46
Speech Recognition asModern AI

• Draws on wide range of AI techniques
• Knowledge representation manipulation
• Optimal search Viterbi decoding
• Machine Learning
• Baum-Welch for HMMs
• Nearest neighbor k-means clustering for signal
  id
• Probabilistic reasoning/Bayes rule
• Manage uncertainty in signal, phone, word mapping
• Enables real world application