compressed-transform-stats.h
3.12 KB
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
// transform/compressed-transform-stats.h
// Copyright 2012 Johns Hopkins University (author: Daniel Povey)
// See ../../COPYING for clarification regarding multiple authors
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// THIS CODE IS PROVIDED *AS IS* BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION ANY IMPLIED
// WARRANTIES OR CONDITIONS OF TITLE, FITNESS FOR A PARTICULAR PURPOSE,
// MERCHANTABLITY OR NON-INFRINGEMENT.
// See the Apache 2 License for the specific language governing permissions and
// limitations under the License.
#ifndef KALDI_TRANSFORM_COMPRESSED_TRANSFORM_STATS_H_
#define KALDI_TRANSFORM_COMPRESSED_TRANSFORM_STATS_H_
#include <vector>
#include "transform/transform-common.h"
namespace kaldi {
// The purpose of this class is to compress the AffineXformStats into less
// memory for easier storage and transmission across the network. It was a
// feature requested by particular user of Kaldi. It's based on the
// CompressedMatrix class, which compresses a matrix into around one byte per
// element, but before applying that, we first use various techniques to
// normalize the range of elements of the stats and to make it so that the
// compressed G matrices will still be positive definite. [Basically, we
// compress the Cholesky of each G_i, and we first normalize all the G_i to have
// the same trace.] We also mess with the K stats a bit, to ensure that the
// derivative of the "compressed" transform taken where the transformation
// matrix is the "default" matrix, is the same as the derivative of the
// un-compressed matrix. [I.e. we correct the stored K to account for the
// compression of G.]
class CompressedAffineXformStats {
public:
CompressedAffineXformStats(): beta_(0.0) { }
CompressedAffineXformStats(const AffineXformStats &input) {
CopyFromAffineXformStats(input);
}
void CopyFromAffineXformStats(const AffineXformStats &input);
void CopyToAffineXformStats(AffineXformStats *output) const;
void Write(std::ostream &os, bool binary) const;
void Read(std::istream &is, bool binary);
private:
// Note: normally we don't use float, only BaseFloat. In this case
// it seems more appropriate to use float (since the stuff in G_ is
// already a lot more approximate than float.)
float beta_;
Matrix<float> K_;
CompressedMatrix G_; // This dim x [ 1 + (0.5*(dim+1)*(dim+2))] matrix
// stores the contents of the G_ matrix of the AffineXform Stats, in a
// compressed form.
// Convert one G matrix into linearized, normalized form ready
// for compression.
static void PrepareOneG(const SpMatrix<double> &Gi, double beta,
SubVector<double> *linearized);
// Reverse the process of PrepareOneG.
static void ExtractOneG(const SubVector<double> &linearized, double beta,
SpMatrix<double> *Gi);
};
} // namespace kaldi
#endif // KALDI_TRANSFORM_COMPRESSED_TRANSFORM_STATS_H_