bayesian / boosting Goto Github PK

Fast & Simple implementation of GBM

GBM is the generally regarded as best perform supervised learning algorithms before recent DL revolution. It is robust but not scalable.

Goal:

1. Fast (Handle 40M rows * 500 features within 10 hours)
2. Simple (The less lines of code, the better) <= 3000
3. Mudular/Extensible for further improvements

Algorithms:

1. pre-bucketing (data compression)
2. bucket sort to build histogram, then linear scan to find best split
3. hints and intelligent of using #buckets
4. stochastic gradient boosting machine

features:

1. correctness (model + fimps)
2. deterministic randomness
3. easily extensible for wide varieties of similar algorithms: random forest, bagging, gbm, for both classification and regression methods, regression takes priority

new features:

1. byte/short: two layer of storage. (save both memory and cpu)
2. taking hints based on previous fimps (top 1/3 using short, rest using byte)

Prameters:

m: number trees n: number of leaves per tree r: example sampling rate s: feature sampling rate

d: number of data points f: number of features

k: number of buckets ml: minimum number of datapoints per leave

Complexity: Memory: max(f * d1 * 8, [f * d, f * d * 2))

Algorithmic:

1. Bucketization: O(f * d1 * log(d1))
2. Continue reading: O(f * d2 * log(k))

3: Single Best Split: O(f' * d' + f' * k) 4a: depth-k balanced tree: k * S 4b: single n-leaves tree: #splits: (2n - 3), O(S * n * log(n)) (roughly)

D: 20M, exampling sampling: 4M feature sampling rate:

Components:

Config: (specify data format and training parameters) DataSet: (column-wise storage, with Self Compression) Tree: (works both in compressed/raw) TreeRegressor: (k-leaf regression tree) GbmFun: (function to extend to different types of loss) Gbm: (gradient boosting machine)