Title: Boosted Augmented Naive Bayes Efficient discriminative learning of Bayesian network classifiers
1Boosted Augmented Naive Bayes Efficient
discriminative learning of Bayesian network
classifiers
- Yushi Jing
- GVU, College of Computing, Georgia Institute of
Technology - Vladimir Pavlovic
- Department of Computer Science, Rutgers
University - James M. Rehg
- GVU, College of Computing, Georgia Institute of
Technology -
2Contribution
- Boosting approach to Bayesian network
classification - Additive combination of simple models (e.g. Naïve
Bayes) - Weighted maximum likelihood learning
- Generalizes Boosted Naïve Bayes (Elkan 1997)
- Comprehensive experimental evaluation of BNB.
- Boosted Augmented Naïve Bayes (BAN)
- Efficient training algorithm
- Competitive classification accuracy
- Naïve Bayes, TAN, BNC (2004), ELR (2001)
3Bayesian network
- Modular and Intuitive graphical representation
- Explicit Probabilistic Representation
Bayesian network classifiers
- Joint distribution
- Conditional distribution
- Class Label
- How to efficiently train Bayesian network
discriminatively to improve its classification
accuracy?
4Parameter Learning
- Maximum Likelihood parameter learning
- Efficient parameter learning algorithm
- Maximizes LLG score
- No analytic solution for parameters that
maximizes CLLG
5Model selection
- ML does not optimize CLLA
- ELRA optimizes CLLA
- (Greiner and Zhou, 2002)
A
Excellent classification accuracy Computationally
expensive in training
B
- ML optimizes CLLB when B is optimal
- BNC algorithm searches for the optimal
structure - (Grossman and Domingos, 2004)
6Talk outline
Our Goal
- Combine parameter and structure optimization
- Avoid over-fitting
- Retain training efficiency
- Minimization function for Boosted Bayesian
network - Empirical Evaluation of Boosted Naïve Bayes
- Boosted Augmented Naïve Bayes (BAN)
- Empirical Evaluation of BAN
7Exponential Loss Function (ELF)
- Boosted Bayesian network classifier minimizes ELF
function.
- ELFF is an upper bound of CLLF
8Minimizing ELF via ensemble method
- Ensemble method
- Adaboost (Population version) constructs F(x)
additively to approximately minimizes ELFF - Discriminatively updates the data weights
- Tractable ML learning to train the parameters
9Results 25 UCI datasets (BNB)
BNB vs. NB 0.151 vs. 0.173
10Results 25 UCI datasets (BNB)
(13)
(14)
BNB (9)
BNB (10)
BNB vs. NB 0.151 vs. 0.173
- BNB vs. TAN
- 0.151 vs. 0.184
TAN (2)
NB (2)
BNB (5)
BNB (7)
(16)
(15)
BNB vs. ELR-NB 0.151 vs. 0.161
BNB vs. BNC-2P 0.151 vs. 0.164
ELR-NB (4)
BNC-2P (3)
11Evaluation of BNB
- Computationally Efficient method
- O(MNT) , T 520, O(MN)
- Good classification Accuracy
- Outperforms NB, TAN
- Competitive with ELR, BNC
- Sparse structure boosting competitive
accuracy - Potential drawbacks
- Strongly correlated features (Corral, etc)
12Structure Learning
- Challenge
- Efficiency
- NP-hard problem
- K-2, Hill Climbing search still examines
polynomial number of structures - Resisting overfitting
- Structure controls classifier capacity
- Our proposed solution
- Combines sparse model to form an ensemble
- Constrains edge selection
13Creating
- Step 1 (Friedman et al. 1999)
- Build pair-wise conditional mutual information
table - Create maximum spanning tree using conditional
mutual information as edge weight - Convert a undirected graph into a directed graph
1
2
3
4
14Initial structure
- Select Naïve Bayes
- Create BNB via AdaBoost
- Evaluate BNB
2
1
3
4
15Iteratively adding edges
Ensemble CLL -0.75
2
1
3
4
16Final BAN structure
Ensemble of the final structure produced by
17Analysis of BAN
- BAN
- The base structure is sparser than BNC model
- BAN uses an ensemble of sparser models to
approximate a densely connected structure
Example of BAN model
Example of BNC-2P model
18Computational complexity of BAN
- Training Complexity O(MN2 MNTS)
- O (MN2) G_tree
- O (MNTS) Structure Search
- T gt boosting iteration per structure
- S gt number of structure examined
- S lt N
- Empirical training time
- T 525, S 05
- Approximately 25-100 times the training of NB
19Result (simulated dataset)
True structure
Naïve Bayes
- 25 different distribution
- CPT table
- Number of features
- 4000 samples, 5 fold cross validation
20Results (simulated dataset)
(6)
BAN(19)
NB (0)
21Results (simulated dataset)
True structure
BAN (3)
22
- Correct edges added under BAN
BNB achieved optimal error in 22 datasets BAN
outperforms BNB in the remaining 3
22Results 25 UCI datasets (BAN)
- Standard datasets for Bayesian network
classifiers - Friedman et. al. 1999
- Greiner and Zhou 2002
- Grossman and Domingos 2004
- 5 fold cross validation
- Implemented NB, TAN, BAN, BNB, BNC-2P
- Obtained results for ELR-NB, ELR-TAN
23Results BAN vs. Standard method
(13)
BAN (10)
BAN (10)
NB (2)
TAN (2)
BAN VS NB 0.141 VS 0.173
BAN VS TAN 0.141 VS 0.184
24Results BAN vs. Structure Learning
BAN (7)
BNC (1)
BAN VS BNC-2P 0.141 VS 0.164
BAN contains 0-5 augmented edges BNC-2P contains
4-16 augmented edges
25Results BAN vs. ELR
(13)
BAN (8)
(14)
BAN (5)
BAN (6)
BAN (4)
- BAN VS ELR-TAN 0.141 vs. 0.155
- BAN VS ELR-NB 0.141 vs. 0.161
Error stats directly taken from published
results BAN is more efficient to train
26Evaluation of BAN vs. BNB
- Comparison under significance test
- BAN outperforms BNB (7)
- Corral
- 2 - 5
- BNB outperforms BAN (2)
- 0.5-2
- Not significant 13
- BAN choose BNB as base structure
- IRIS, MOFN
- Average testing error
- 0.141 vs. 0.151
- BAN outperforms BNB (16)
- BNB outperforms BAN (6)
BAN (7)
(14)
BNB (2)
- BAN VS BNB 0.141 VS 0.151
27Conclusion
- An ensemble of sparse model as an alternative to
structure and parameter optimization - Simple to implement
- Very efficient in training
- Competitive classification accuracy
- NB, TAN, HGC
- BNC
- ELR
28Future Work
- Extend BAN to handle sequential data
- Analyze the class of Bayesian network classifiers
that can be approximated with an ensemble of
sparse structures. - Can the BAN model parameters be obtained through
parameter learning given the final model
structure? - Can we use BAN approach to learn generative
models?