-
Notifications
You must be signed in to change notification settings - Fork 4
/
BatchPoissonPure.h
424 lines (350 loc) · 12.5 KB
/
BatchPoissonPure.h
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
//
// Created by eliezer on 07.11.16.
//
#ifndef POISSON_SCR_CPP_BATCHPOISSONPURE_H
#define POISSON_SCR_CPP_BATCHPOISSONPURE_H
#include <ostream>
#include <memory>
#include <vector>
#include <list>
#include <iostream>
#include <utility>
#include <string>
#include <boost/random.hpp>
#include <boost/random/gamma_distribution.hpp>
#include <boost/unordered_map.hpp>
#include <unordered_map>
using namespace std;
enum class vars {
a_beta=0, a_theta ,a_epsilon,a_eta,a_tau,
b_beta,b_theta,b_epsilon,b_eta, b_tau,
e_beta,e_theta,e_epsilon,e_eta, e_tau,
elog_beta,elog_theta,elog_epsilon,elog_eta, elog_tau, phi, xi_M,xi_N,xi_S};
static const char * EnumStrings[] = { "a_beta", "a_theta" ,"a_epsilon","a_eta","a_tau",
"b_beta","b_theta","b_epsilon","b_eta", "b_tau",
"e_beta","e_theta","e_epsilon","e_eta", "e_tau",
"elog_beta","elog_theta","elog_epsilon","elog_eta", "elog_tau", "phi", "xi_M","xi_N","xi_S" };
vector<vars> static vec_vars={vars::a_beta, vars::a_theta ,vars::a_epsilon,vars::a_eta,vars::a_tau,
vars::b_beta,vars::b_theta,vars::b_epsilon,vars::b_eta, vars::b_tau,
vars::e_beta,vars::e_theta,vars::e_epsilon,vars::e_eta, vars::e_tau,
vars::elog_beta,vars::elog_theta,vars::elog_epsilon,vars::elog_eta, vars::elog_tau, vars::phi,
vars::xi_M,vars::xi_N,vars::xi_S};
enum class variables_aux {phi=0, xi_M=1, xi_N=2, xi_S=3};
size_t sizes_var(vars v,size_t n_ratings, size_t n_wd_entries, size_t n_users, size_t n_items,size_t k_feat, size_t n_words, size_t n_max_neighbors);
size_t total_memory(size_t n_ratings, size_t n_wd_entries, size_t n_users, size_t n_items,size_t k_feat, size_t n_words, size_t n_max_neighbors);
template<class T> class Array{
public:
T* data;
size_t ncol;
size_t nrow;
T* data_end;
size_t incr;
Array(){
data_end=data=NULL;
ncol=nrow=0;
}
Array(T* data, size_t nrow, size_t ncol, T value) : data(data), ncol(ncol), nrow(nrow) {
incr=1;
data_end=data+ncol*nrow;
for(T* data_start=data;data_start!=data_end;data_start++){
data_start[0]=value;
}
}
Array(T* data, size_t nrow, size_t ncol) : data(data), ncol(ncol), nrow(nrow) {
incr=1;
data_end=data+ncol*nrow;
}
size_t cols(){
return ncol;
}
size_t rows(){
return nrow;
}
T get(size_t row, size_t column) const
{
return (row * ncol + data)[column];
}
T& operator ()(size_t row, size_t column)
{
return (row * ncol + data)[column];
}
T& operator ()(size_t column)
{
return data[column];
}
Array<T>& operator=(const T rhs) {
for(T* data_start=data;data_start!=data_end;data_start++){
data_start[0]=rhs;
}
return *this;
}
Array<T> & operator+=(T a){
for(T* data_start=data;data_start!=data_end;data_start++){
data_start[0]+=a;
}
return *this;
}
Array<T> & operator*=(T a){
for(T* data_start=data;data_start!=data_end;data_start++){
data_start[0]*=a;
}
return *this;
}
Array<T> & operator/=(T a){
for(T* data_start=data;data_start!=data_end;data_start++){
data_start[0]/=a;
}
return *this;
}
void row_multiply(size_t row,T a){
T* pt;
int i;
for (i = 0, pt=(row * ncol + data); i < ncol; ++i,++pt) {
pt[0]*=a;
}
}
void row_add(size_t row,T a){
T* pt;
int i;
for (i = 0, pt=(row * ncol + data); i < ncol; ++i,++pt) {
pt[0]+=a;
}
}
Array<T> row(size_t irow){
return Array<T>(irow*ncol+data,1,ncol);
}
T col_sum(size_t col){
T* pt;
int i;
T sum=0;
for (i = 0, pt=(col+ data); i < nrow; ++i,pt+=ncol ) {
sum+=pt[0];
}
return sum;
}
void row_normalize(Array<T>& a,Array<T>& b){
// normalize this, a and b (ncol and nrow should be the same) using the row sum of this, a and b as
// normalizing factor
if(nrow==a.nrow && a.nrow==b.nrow && ncol==a.ncol && a.ncol == b.ncol){
for (int i = 0; i < nrow; ++i) {
T sum = 0;
for (int j = 0; j < ncol; ++j) {
sum+=(*this)(i,j);
}
for (int k = 0; k < a.ncol; ++k ) {
sum+=a(i,k);
}
for (int l = 0; l < b.ncol; ++l) {
sum+=b(i,l);
}
row_multiply(i,1.0/sum);
a.row_multiply(i,1.0/sum);
b.row_multiply(i,1.0/sum);
}
}
}
void row_normalize(){
for (int i = 0; i < nrow; ++i) {
T sum = 0;
for (int j = 0; j < ncol; ++j) {
sum+=(*this)(i,j);
}
row_multiply(i,1.0/sum);
}
}
void init_gamma_row_normalized(double shape=1.1){
init_gamma_row_normalized(shape,1);
}
void init_gamma_row_normalized(double shape, double rate){
// shape= alpha
// rate = beta
// the gamma normalized with rate=1 is actually a dirichlet with concentration parameter = shape
boost::mt19937 rng=boost::mt19937(time(0));
boost::gamma_distribution<> gd( shape );
boost::variate_generator<boost::mt19937&,boost::gamma_distribution<> > var_gamma( rng, gd );
for (int i = 0; i < nrow; ++i) {
T sum = 0;
for (int j = 0; j < ncol; ++j) {
(*this)(i,j)=var_gamma()/rate;
sum+=(*this)(i,j);
}
//row_multiply(i,1.0/sum);
}
}
friend std::ostream &operator<<(std::ostream &os, const Array<T> &arr1) {
os << "{\"nrow\":" << arr1.nrow
<< ", \"ncol\":" << arr1.ncol
<<", \"data\":[" ;
for (size_t i = 0; i < arr1.nrow; ++i) {
os << "[";
for (size_t j = 0; j < arr1.ncol; ++j) {
os << arr1.get(i,j);
if ((j + 1) < arr1.ncol)
os << ",";
}
os << "]";
if((i+1)<arr1.nrow)
os << ",";
os << endl;
}
os << "] "<<endl<<"}";
return os;
}
};
template<class T> class ArrayManager{
private:
T* data;
T* next_pointer;
size_t total_capacity;
size_t used_capacity;
public:
ArrayManager(size_t total_capacity) : total_capacity(total_capacity) {
used_capacity=0;
data = new T[total_capacity];
next_pointer=data;
}
Array<T> makeArray(size_t nrow,size_t ncol){
if((used_capacity+(nrow*ncol))<total_capacity) {
Array<T> ret = Array<T>(next_pointer, nrow, ncol);
used_capacity += ncol * nrow;
next_pointer += (ncol * nrow);
return (ret);
}else
std::invalid_argument( " not enough pre-allocated space for next array" );
}
Array<T> makeArray(size_t nrow,size_t ncol,T value){
if((used_capacity+(nrow*ncol))<total_capacity) {
Array<T> ret = Array<T>(next_pointer, nrow, ncol,value);
used_capacity += ncol * nrow;
next_pointer += (ncol * nrow);
return (ret);
}else
std::invalid_argument( " not enough pre-allocated space for next array" );
}
Array<T> makeArray(size_t ncol){
size_t nrow=1;
if((used_capacity+(nrow*ncol))<total_capacity) {
Array<T> ret = Array<T>(next_pointer, nrow,ncol);
used_capacity += ncol * nrow;
next_pointer += (ncol * nrow);
return (ret);
}else
std::invalid_argument( " not enough pre-allocated space for next array" );
}
Array<T> makeArray(size_t ncol, T value){
size_t nrow=1;
if((used_capacity+(nrow*ncol))<total_capacity) {
Array<T> ret = Array<T>(next_pointer, nrow,ncol,value);
used_capacity += ncol * nrow;
next_pointer += (ncol * nrow);
return (ret);
}else
std::invalid_argument( " not enough pre-allocated space for next array" );
}
virtual ~ArrayManager() {
delete[] data;
}
};
typedef Array<double> Arrayf;
typedef unordered_map< pair<size_t, size_t >,size_t,boost::hash< std::pair<size_t, size_t> > > pairmap;
class gamma_latent {
public:
Arrayf a_latent;
Arrayf b_latent;
Arrayf e_expected;
Arrayf elog_expected;
double a;
double b;
size_t nvars;
size_t kdim;
gamma_latent(const Arrayf &a_latent, const Arrayf &b_latent, const Arrayf &e_expected, const Arrayf &elog_expected,
double a, double b);
gamma_latent( ArrayManager<double>* arrman, size_t nrows, size_t ncols, double a, double b);
gamma_latent(){
;
}
void update_expected();
double elbo_term();
double elbo_term(vector<gamma_latent*> vars);
double elbo_term_prod_linear_expectations(vector<gamma_latent*> vars);
void init_b_latent();
void init_a_latent();
friend std::ostream &operator<<(std::ostream &os, const gamma_latent &var){
os << "{\"a\" : " << var.a
<<", \"b\":" << var.b
<< ", \"nvars\":" << var.nvars
<< ", \"kdim\": " << var.kdim<<endl;
os << ",\"a_latent\":" << var.a_latent<<endl;
os << ",\"b_latent\":" << var.b_latent<<endl;
os << ",\"exp\":" << var.e_expected<<endl;
os << ",\"logexp\":" << var.elog_expected<<endl;
os << "}"<<endl;
return os;
}
};
class BatchPoissonNewArray {
public:
ArrayManager<double>* arrman;
gamma_latent beta;
gamma_latent theta;
gamma_latent epsilon;
gamma_latent eta;
gamma_latent tau;
Arrayf phi;
Arrayf xi_M;
Arrayf xi_N;
Arrayf xi_S;
list<double > elbo_lst;
list<long > iter_time_lst;
vector<tuple<size_t,size_t,size_t>> r_entries; // tuple<user,item,feedback>
vector<tuple<size_t,size_t,size_t>> w_entries; // tuple<word,item,word-count-in-item>
vector< list < pair<size_t, size_t > > > user_items_neighboors; // list<<pair<user_neighbor_i,index_in_r_entries>>
pairmap user_items_map; // map<pair<user_id,item_id>,index_in_r_entries>>
vector< vector < size_t > > user_neighboors;
vector< pair<size_t,size_t>> user_items_index;
// pair<u_i,v_i>, where u_i is the beginning index and v_i the ending index
// of items for user i in the rating matrix
size_t _n_users;
size_t _n_items;
size_t _k_feat;
size_t _n_words;
size_t _n_ratings;
size_t _n_wd_entries; // number of word-document non-zero counts
size_t _n_max_neighbors;
double a=0.1; double b=0.1; double c=0.1; double d=0.1; double e=0.1; double f=0.1; double g=0.1;
double h=0.1; double k=0.1; double l=0.1;
size_t mem_use=0;
BatchPoissonNewArray(size_t n_ratings,size_t n_wd_entries,size_t n_users, size_t n_items, size_t k_feat, size_t n_words,size_t n_max_neighbors,
double a=0.1, double b=0.1, double c=0.1, double d=0.1, double e=0.1, double f=0.1, double g=0.1,
double h=0.1, double k=0.1, double l=0.1);
BatchPoissonNewArray(){
;
}
void train(size_t n_iter, double tol);
virtual void init_train(vector<tuple<size_t, size_t, size_t>> r_entries,
vector<tuple<size_t, size_t, size_t>> w_entries,vector< vector < size_t > > user_neighboors);
virtual void update_latent();
virtual void update_aux_latent();
void init_aux_latent();
virtual double compute_elbo();
virtual vector<vector<double>> estimate();
vector<vector<size_t>> recommend(size_t m);
virtual double tau_elbo_expected_linear_term();
virtual ~BatchPoissonNewArray();
friend std::ostream &operator<<(std::ostream &os, BatchPoissonNewArray &var){
for(auto v : var.estimate())
{
cout << v.size() << "||";
std::copy (v.begin(), v.end(), std::ostream_iterator<double>(os, "\t"));
os << endl;
}
cout << endl;
return os;
}
void recommend(ostream &output, size_t m);
};
// auxiliary numerical functions
long double digammal(long double x);
double LogFactorial(size_t n);
double gamma_term(double a, double b, double a_latent, double b_latent, double e_latent, double elog_latent);
#endif //POISSON_SCR_CPP_BATCHPOISSONPURE_H