Cambios

              
    

          
          
        
        

            
f
1
{
f
1
{
            
2
  "author": "SI Valdez, A Rojas-Dom\u00ednguez", 
2
  "author": "SI Valdez, A Rojas-Dom\u00ednguez", 
            
3
  "author_email": null, 
3
  "author_email": null, 
            
4
  "creator_user_id": "a3da3ec9-3fd4-47a4-8d04-0a90b09614e0", 
4
  "creator_user_id": "a3da3ec9-3fd4-47a4-8d04-0a90b09614e0", 
            
5
  "extras": [
5
  "extras": [
            
6
    {
6
    {
            
7
      "key": "A\u00f1o", 
7
      "key": "A\u00f1o", 
            
8
      "value": "2019"
8
      "value": "2019"
            
9
    }, 
9
    }, 
            
10
    {
10
    {
            
11
      "key": "Google Scholar URL", 
11
      "key": "Google Scholar URL", 
            
12
      "value": 
12
      "value": 
            
13
gesize=100&sortby=pubdate&citation_for_view=MG1jyREAAAAJ:70eg2SAEIzsC"
13
gesize=100&sortby=pubdate&citation_for_view=MG1jyREAAAAJ:70eg2SAEIzsC"
            
14
    }, 
14
    }, 
            
15
    {
15
    {
            
16
      "key": "Identificador hash", 
16
      "key": "Identificador hash", 
            
17
      "value": "968ea34084e1"
17
      "value": "968ea34084e1"
            
18
    }, 
18
    }, 
            
19
    {
19
    {
            
20
      "key": "Lugar de publicaci\u00f3n", 
20
      "key": "Lugar de publicaci\u00f3n", 
            
21
      "value": "arXiv preprint arXiv:1907.05951, 2019"
21
      "value": "arXiv preprint arXiv:1907.05951, 2019"
            
22
    }, 
22
    }, 
            
23
    {
23
    {
            
24
      "key": "Tipo", 
24
      "key": "Tipo", 
            
25
      "value": "Publicaci\u00f3n"
25
      "value": "Publicaci\u00f3n"
            
26
    }, 
26
    }, 
            
27
    {
27
    {
            
28
      "key": "Tipo de publicaci\u00f3n", 
28
      "key": "Tipo de publicaci\u00f3n", 
            
29
      "value": "Otro"
29
      "value": "Otro"
            
30
    }, 
30
    }, 
            
31
    {
31
    {
            
32
      "key": "URL directo", 
32
      "key": "URL directo", 
            
33
      "value": "https://arxiv.org/pdf/1907.05951"
33
      "value": "https://arxiv.org/pdf/1907.05951"
            
34
    }
34
    }
            
35
  ], 
35
  ], 
            
36
  "groups": [
36
  "groups": [
            
37
    {
37
    {
            
38
      "description": "", 
38
      "description": "", 
            
39
      "display_name": "Publicaciones", 
39
      "display_name": "Publicaciones", 
            
40
      "id": "8be672a5-4640-455e-a4f3-46b52b66c07b", 
40
      "id": "8be672a5-4640-455e-a4f3-46b52b66c07b", 
            
41
      "image_display_url": "", 
41
      "image_display_url": "", 
            
42
      "name": "publicaciones", 
42
      "name": "publicaciones", 
            
43
      "title": "Publicaciones"
43
      "title": "Publicaciones"
            
44
    }
44
    }
            
45
  ], 
45
  ], 
            
46
  "id": "ec18fcf0-ee96-4e83-a619-035aa9898085", 
46
  "id": "ec18fcf0-ee96-4e83-a619-035aa9898085", 
            
47
  "isopen": false, 
47
  "isopen": false, 
            
48
  "license_id": null, 
48
  "license_id": null, 
            
49
  "license_title": null, 
49
  "license_title": null, 
            
50
  "maintainer": null, 
50
  "maintainer": null, 
            
51
  "maintainer_email": null, 
51
  "maintainer_email": null, 
            
52
  "metadata_created": "2025-10-21T09:02:02.444448", 
52
  "metadata_created": "2025-10-21T09:02:02.444448", 
            
n
53
  "metadata_modified": "2025-10-21T09:02:02.444461", 
n
53
  "metadata_modified": "2025-10-21T09:02:03.000431", 
            
54
  "name": "968ea34084e1", 
54
  "name": "968ea34084e1", 
            
55
  "notes": "The performance of deep neural networks, such as Deep 
55
  "notes": "The performance of deep neural networks, such as Deep 
            
56
Belief Networks formed by Restricted Boltzmann Machines (RBMs), 
56
Belief Networks formed by Restricted Boltzmann Machines (RBMs), 
            
57
strongly depends on their training, which is the process of adjusting 
57
strongly depends on their training, which is the process of adjusting 
            
58
their parameters. This process can be posed as an optimization problem 
58
their parameters. This process can be posed as an optimization problem 
            
59
over n dimensions. However, typical networks contain tens of thousands 
59
over n dimensions. However, typical networks contain tens of thousands 
            
60
of parameters, making this a High-Dimensional Problem (HDP). Although 
60
of parameters, making this a High-Dimensional Problem (HDP). Although 
            
61
different optimization methods have been employed for this goal, the 
61
different optimization methods have been employed for this goal, the 
            
62
use of most of the Evolutionary Algorithms (EAs) becomes prohibitive 
62
use of most of the Evolutionary Algorithms (EAs) becomes prohibitive 
            
63
due to their inability to deal with HDPs. For instance, the Covariance 
63
due to their inability to deal with HDPs. For instance, the Covariance 
            
64
Matrix Adaptation Evolutionary Strategy (CMA-ES) which is regarded as 
64
Matrix Adaptation Evolutionary Strategy (CMA-ES) which is regarded as 
            
65
one of the most effective EAs, exhibits the enormous disadvantage of 
65
one of the most effective EAs, exhibits the enormous disadvantage of 
            
66
requiring $O(n^2)$ memory and operations, making it unpractical for 
66
requiring $O(n^2)$ memory and operations, making it unpractical for 
            
67
problems with more than a few hundred variables. In this paper, we 
67
problems with more than a few hundred variables. In this paper, we 
            
68
introduce a novel EA that requires $O(n)$ operations and memory, but 
68
introduce a novel EA that requires $O(n)$ operations and memory, but 
            
69
delivers competitive solutions for the training stage of RBMs with 
69
delivers competitive solutions for the training stage of RBMs with 
            
70
over one million variables, when compared against CMA-ES and the 
70
over one million variables, when compared against CMA-ES and the 
            
71
Contrastive Divergence algorithm, which is the standard method for 
71
Contrastive Divergence algorithm, which is the standard method for 
            
72
training RBMs.", 
72
training RBMs.", 
            
n
73
  "num_resources": 0, 
n
73
  "num_resources": 1, 
            
74
  "num_tags": 15, 
74
  "num_tags": 15, 
            
75
  "organization": {
75
  "organization": {
            
76
    "approval_status": "approved", 
76
    "approval_status": "approved", 
            
77
    "created": "2022-05-19T00:10:30.480393", 
77
    "created": "2022-05-19T00:10:30.480393", 
            
78
    "description": "Observatorio Metropolitano CentroGeo", 
78
    "description": "Observatorio Metropolitano CentroGeo", 
            
79
    "id": "b3b3a79d-748a-4464-9471-732b6c74ec53", 
79
    "id": "b3b3a79d-748a-4464-9471-732b6c74ec53", 
            
80
    "image_url": 
80
    "image_url": 
            
81
"2022-05-19-001030.456616FullColor1280x1024LogoOnly.png", 
81
"2022-05-19-001030.456616FullColor1280x1024LogoOnly.png", 
            
82
    "is_organization": true, 
82
    "is_organization": true, 
            
83
    "name": "observatorio-metropolitano-centrogeo", 
83
    "name": "observatorio-metropolitano-centrogeo", 
            
84
    "state": "active", 
84
    "state": "active", 
            
85
    "title": "Observatorio Metropolitano CentroGeo", 
85
    "title": "Observatorio Metropolitano CentroGeo", 
            
86
    "type": "organization"
86
    "type": "organization"
            
87
  }, 
87
  }, 
            
88
  "owner_org": "b3b3a79d-748a-4464-9471-732b6c74ec53", 
88
  "owner_org": "b3b3a79d-748a-4464-9471-732b6c74ec53", 
            
89
  "private": false, 
89
  "private": false, 
            
90
  "relationships_as_object": [], 
90
  "relationships_as_object": [], 
            
91
  "relationships_as_subject": [], 
91
  "relationships_as_subject": [], 
            
t
92
  "resources": [], 
t
92
  "resources": [
            
93
    {
            
94
      "cache_last_updated": null, 
            
95
      "cache_url": null, 
            
96
      "created": "2025-10-21T09:02:03.040430", 
            
97
      "datastore_active": false, 
            
98
      "description": "The performance of deep neural networks, such as 
            
99
Deep Belief Networks formed by Restricted Boltzmann Machines (RBMs), 
            
100
strongly depends on their training, which is the process of adjusting 
            
101
their parameters. This process can be posed as an optimization problem 
            
102
over n dimensions. However, typical networks contain tens of thousands 
            
103
of parameters, making this a High-Dimensional Problem (HDP). Although 
            
104
different optimization methods have been employed for this goal, the 
            
105
use of most of the Evolutionary Algorithms (EAs) becomes prohibitive 
            
106
due to their inability to deal with HDPs. For instance, the Covariance 
            
107
Matrix Adaptation Evolutionary Strategy (CMA-ES) which is regarded as 
            
108
one of the most effective EAs, exhibits the enormous disadvantage of 
            
109
requiring $O(n^2)$ memory and operations, making it unpractical for 
            
110
problems with more than a few hundred variables. In this paper, we 
            
111
introduce a novel EA that requires $O(n)$ operations and memory, but 
            
112
delivers competitive solutions for the training stage of RBMs with 
            
113
over one million variables, when compared against CMA-ES and the 
            
114
Contrastive Divergence algorithm, which is the standard method for 
            
115
training RBMs.", 
            
116
      "format": "HTML", 
            
117
      "hash": "", 
            
118
      "id": "7efd7444-518b-4fab-aae7-da271d1b50b1", 
            
119
      "last_modified": null, 
            
120
      "metadata_modified": "2025-10-21T09:02:03.004487", 
            
121
      "mimetype": null, 
            
122
      "mimetype_inner": null, 
            
123
      "name": "An evolutionary algorithm of linear complexity: 
            
124
application to training of deep neural networks", 
            
125
      "package_id": "ec18fcf0-ee96-4e83-a619-035aa9898085", 
            
126
      "position": 0, 
            
127
      "resource_type": null, 
            
128
      "size": null, 
            
129
      "state": "active", 
            
130
      "url": 
            
131
esize=100&sortby=pubdate&citation_for_view=MG1jyREAAAAJ:70eg2SAEIzsC", 
            
132
      "url_type": null
            
133
    }
            
134
  ], 
            
93
  "state": "active", 
135
  "state": "active", 
            
94
  "tags": [
136
  "tags": [
            
95
    {
137
    {
            
96
      "display_name": "2019", 
138
      "display_name": "2019", 
            
97
      "id": "856c48cd-5e9b-4ee9-9342-d17b6fe761a5", 
139
      "id": "856c48cd-5e9b-4ee9-9342-d17b6fe761a5", 
            
98
      "name": "2019", 
140
      "name": "2019", 
            
99
      "state": "active", 
141
      "state": "active", 
            
100
      "vocabulary_id": null
142
      "vocabulary_id": null
            
101
    }, 
143
    }, 
            
102
    {
144
    {
            
103
      "display_name": "algorithm", 
145
      "display_name": "algorithm", 
            
104
      "id": "da224d7c-d794-48c8-8316-05d6b5fdc51c", 
146
      "id": "da224d7c-d794-48c8-8316-05d6b5fdc51c", 
            
105
      "name": "algorithm", 
147
      "name": "algorithm", 
            
106
      "state": "active", 
148
      "state": "active", 
            
107
      "vocabulary_id": null
149
      "vocabulary_id": null
            
108
    }, 
150
    }, 
            
109
    {
151
    {
            
110
      "display_name": "artificial-intelligence", 
152
      "display_name": "artificial-intelligence", 
            
111
      "id": "47cb06d3-c2a4-42af-b633-03eea6181083", 
153
      "id": "47cb06d3-c2a4-42af-b633-03eea6181083", 
            
112
      "name": "artificial-intelligence", 
154
      "name": "artificial-intelligence", 
            
113
      "state": "active", 
155
      "state": "active", 
            
114
      "vocabulary_id": null
156
      "vocabulary_id": null
            
115
    }, 
157
    }, 
            
116
    {
158
    {
            
117
      "display_name": "artificial-neural-network", 
159
      "display_name": "artificial-neural-network", 
            
118
      "id": "d63d29ea-7b70-4ed5-83cc-66f3b9586c1c", 
160
      "id": "d63d29ea-7b70-4ed5-83cc-66f3b9586c1c", 
            
119
      "name": "artificial-neural-network", 
161
      "name": "artificial-neural-network", 
            
120
      "state": "active", 
162
      "state": "active", 
            
121
      "vocabulary_id": null
163
      "vocabulary_id": null
            
122
    }, 
164
    }, 
            
123
    {
165
    {
            
124
      "display_name": "boltzmann-machine", 
166
      "display_name": "boltzmann-machine", 
            
125
      "id": "5eee20bd-596c-4b7f-a187-efb2a9478358", 
167
      "id": "5eee20bd-596c-4b7f-a187-efb2a9478358", 
            
126
      "name": "boltzmann-machine", 
168
      "name": "boltzmann-machine", 
            
127
      "state": "active", 
169
      "state": "active", 
            
128
      "vocabulary_id": null
170
      "vocabulary_id": null
            
129
    }, 
171
    }, 
            
130
    {
172
    {
            
131
      "display_name": "cma-es", 
173
      "display_name": "cma-es", 
            
132
      "id": "352ed64b-c0c4-4780-a4f1-871f7df7691f", 
174
      "id": "352ed64b-c0c4-4780-a4f1-871f7df7691f", 
            
133
      "name": "cma-es", 
175
      "name": "cma-es", 
            
134
      "state": "active", 
176
      "state": "active", 
            
135
      "vocabulary_id": null
177
      "vocabulary_id": null
            
136
    }, 
178
    }, 
            
137
    {
179
    {
            
138
      "display_name": "computer-science", 
180
      "display_name": "computer-science", 
            
139
      "id": "29cae056-cd7e-43f7-be5b-b25869a3fbf2", 
181
      "id": "29cae056-cd7e-43f7-be5b-b25869a3fbf2", 
            
140
      "name": "computer-science", 
182
      "name": "computer-science", 
            
141
      "state": "active", 
183
      "state": "active", 
            
142
      "vocabulary_id": null
184
      "vocabulary_id": null
            
143
    }, 
185
    }, 
            
144
    {
186
    {
            
145
      "display_name": "deep-neural-networks", 
187
      "display_name": "deep-neural-networks", 
            
146
      "id": "4f23ca30-4b12-401d-a93c-958cf047df6e", 
188
      "id": "4f23ca30-4b12-401d-a93c-958cf047df6e", 
            
147
      "name": "deep-neural-networks", 
189
      "name": "deep-neural-networks", 
            
148
      "state": "active", 
190
      "state": "active", 
            
149
      "vocabulary_id": null
191
      "vocabulary_id": null
            
150
    }, 
192
    }, 
            
151
    {
193
    {
            
152
      "display_name": "divergence-linguistics", 
194
      "display_name": "divergence-linguistics", 
            
153
      "id": "34d5602a-4b6e-464f-bc6e-c6636fc7cc58", 
195
      "id": "34d5602a-4b6e-464f-bc6e-c6636fc7cc58", 
            
154
      "name": "divergence-linguistics", 
196
      "name": "divergence-linguistics", 
            
155
      "state": "active", 
197
      "state": "active", 
            
156
      "vocabulary_id": null
198
      "vocabulary_id": null
            
157
    }, 
199
    }, 
            
158
    {
200
    {
            
159
      "display_name": "evolution-strategy", 
201
      "display_name": "evolution-strategy", 
            
160
      "id": "836dbf9c-a84f-424f-8df2-441a41df838c", 
202
      "id": "836dbf9c-a84f-424f-8df2-441a41df838c", 
            
161
      "name": "evolution-strategy", 
203
      "name": "evolution-strategy", 
            
162
      "state": "active", 
204
      "state": "active", 
            
163
      "vocabulary_id": null
205
      "vocabulary_id": null
            
164
    }, 
206
    }, 
            
165
    {
207
    {
            
166
      "display_name": "evolutionary-algorithm", 
208
      "display_name": "evolutionary-algorithm", 
            
167
      "id": "22db3110-8a27-435b-8d9a-c910fe7201ca", 
209
      "id": "22db3110-8a27-435b-8d9a-c910fe7201ca", 
            
168
      "name": "evolutionary-algorithm", 
210
      "name": "evolutionary-algorithm", 
            
169
      "state": "active", 
211
      "state": "active", 
            
170
      "vocabulary_id": null
212
      "vocabulary_id": null
            
171
    }, 
213
    }, 
            
172
    {
214
    {
            
173
      "display_name": "machine-learning", 
215
      "display_name": "machine-learning", 
            
174
      "id": "99557dd8-2914-4e28-8278-7a042cfb23d0", 
216
      "id": "99557dd8-2914-4e28-8278-7a042cfb23d0", 
            
175
      "name": "machine-learning", 
217
      "name": "machine-learning", 
            
176
      "state": "active", 
218
      "state": "active", 
            
177
      "vocabulary_id": null
219
      "vocabulary_id": null
            
178
    }, 
220
    }, 
            
179
    {
221
    {
            
180
      "display_name": "mathematical-optimization", 
222
      "display_name": "mathematical-optimization", 
            
181
      "id": "b33a86ee-df39-4619-8165-c8d73afbf5a4", 
223
      "id": "b33a86ee-df39-4619-8165-c8d73afbf5a4", 
            
182
      "name": "mathematical-optimization", 
224
      "name": "mathematical-optimization", 
            
183
      "state": "active", 
225
      "state": "active", 
            
184
      "vocabulary_id": null
226
      "vocabulary_id": null
            
185
    }, 
227
    }, 
            
186
    {
228
    {
            
187
      "display_name": "process-computing", 
229
      "display_name": "process-computing", 
            
188
      "id": "d07d93c6-53f7-48a4-bded-e205ec0bc2c2", 
230
      "id": "d07d93c6-53f7-48a4-bded-e205ec0bc2c2", 
            
189
      "name": "process-computing", 
231
      "name": "process-computing", 
            
190
      "state": "active", 
232
      "state": "active", 
            
191
      "vocabulary_id": null
233
      "vocabulary_id": null
            
192
    }, 
234
    }, 
            
193
    {
235
    {
            
194
      "display_name": "svaldez", 
236
      "display_name": "svaldez", 
            
195
      "id": "f42a8210-1cef-4f03-98a0-a6d3d0d4f848", 
237
      "id": "f42a8210-1cef-4f03-98a0-a6d3d0d4f848", 
            
196
      "name": "svaldez", 
238
      "name": "svaldez", 
            
197
      "state": "active", 
239
      "state": "active", 
            
198
      "vocabulary_id": null
240
      "vocabulary_id": null
            
199
    }
241
    }
            
200
  ], 
242
  ], 
            
201
  "title": "An evolutionary algorithm of linear complexity: 
243
  "title": "An evolutionary algorithm of linear complexity: 
            
202
application to training of deep neural networks", 
244
application to training of deep neural networks", 
            
203
  "type": "dataset", 
245
  "type": "dataset", 
            
204
  "url": 
246
  "url": 
            
205
esize=100&sortby=pubdate&citation_for_view=MG1jyREAAAAJ:70eg2SAEIzsC", 
247
esize=100&sortby=pubdate&citation_for_view=MG1jyREAAAAJ:70eg2SAEIzsC", 
            
206
  "version": null
248
  "version": null
            
207
}
249
}