Presentation on a paper which was published in the British International Conference on Databases in 2015. Experimental framework for selecting deep learning hyper-parameters.
A Configurable Deep Network for High Dimensional Clinical Trial AnalysisJim O' Donoghue
Presentation on a paper which compared shallow and deep methods for analysing high-dimensional clinical trial data, given at the International Joint Conference on Neural Networks in 2015
Civil Protection operators and Public Administrations, engaged in urban planning, resource & environmental management, need spatio-temporal processing of GI to support decision-making. Current SDIs and the ESDI, only partially address user needs as they offer no or very limited time variable management . The integration between INSPIRE-compliant geographic datasets and operational databases, essential in domains such as environmental risk management and civil protection, is poor.
Thus the present scope of services SDI can offer is somewhat limited. It is the aim of BRISEIDE to build on existing SDI’s in order to provide users with more complete and adequate data and processing tools.
A Configurable Deep Network for High Dimensional Clinical Trial AnalysisJim O' Donoghue
Presentation on a paper which compared shallow and deep methods for analysing high-dimensional clinical trial data, given at the International Joint Conference on Neural Networks in 2015
Civil Protection operators and Public Administrations, engaged in urban planning, resource & environmental management, need spatio-temporal processing of GI to support decision-making. Current SDIs and the ESDI, only partially address user needs as they offer no or very limited time variable management . The integration between INSPIRE-compliant geographic datasets and operational databases, essential in domains such as environmental risk management and civil protection, is poor.
Thus the present scope of services SDI can offer is somewhat limited. It is the aim of BRISEIDE to build on existing SDI’s in order to provide users with more complete and adequate data and processing tools.
Rachel Sheppard and Jason Frazier from Mason-McDuffie Mortgage attended the Inman Connect Conference in San Francisco in 2015. This presentation includes several helpful tips we learned from top real estate agents and professionals.
n all the dialects where the Industry 4.0 language is spoken, Industrial Internet of Things, Additive Manufacturing and Robotics from the technology side and Mass Customization, Product-Service Systems and Sustainable Manufacturing from the business side, represent key cornerstones and top priority challenges.
FASTEN will demonstrate an open and standardized framework to produce and deliver tailored-designed products, capable to run autonomously and deliver fast and low-cost additive manufactured products.
Rachel Sheppard and Jason Frazier from Mason-McDuffie Mortgage attended the Inman Connect Conference in San Francisco in 2015. This presentation includes several helpful tips we learned from top real estate agents and professionals.
n all the dialects where the Industry 4.0 language is spoken, Industrial Internet of Things, Additive Manufacturing and Robotics from the technology side and Mass Customization, Product-Service Systems and Sustainable Manufacturing from the business side, represent key cornerstones and top priority challenges.
FASTEN will demonstrate an open and standardized framework to produce and deliver tailored-designed products, capable to run autonomously and deliver fast and low-cost additive manufactured products.
EU H2020 HIT2GAP - SmarTABCD’15 Workshop - VERYSchool NavigatorArantico Ltd
Presentation given on the EU VERYSchool project at the SmarTABCD’15, Workshop on Smart Technologies and Applications on Buildings, Cities and Districts. The workshop was organized in parallel to the 1th International Conference on Artificial Intelligence Applications and Innovations (AIAI 15) organized in Bayonne from the 14th to the 17th of September, and focused on results generated by recent European initiatives for the development of innovative smart technologies for buildings and districts efficiency.
Building energy consumption is strongly dependent on the functionality of the building, the occupancy behavior, the outdoor environment, the structure and building materials, and the operation management. Nowadays and to optimize energy consumption, buildings are using automation systems with equipment, metering systems, smart devices and IoT (Internet of Things) applications to optimize their usage. Today’s automation systems produce tremendous amounts of data. This data can be very hard to organize and use across different applications because it is stored in many different formats; has inconsistent naming conventions and very limited data descriptors. In essence it lacks information to describe the meaning of the data. And without meaning a time–consuming manual effort is required before value creation can begin.
KTN, the Aerospace Technology Institute (ATI) and ADS organised on Thursday 20th May an online event to showcase the latest EU funding opportunities for the aerospace sector.
Although the UK has left the EU, it is still an Associate Member of Horizon Europe, the EU’s research and innovation programme that will run from 2021 to 2027. This means that UK scientists, researchers and businesses can continue to access funding under the programme on equivalent terms as organisations in EU countries. It is therefore fundamental that the UK innovation community is fully aware of the opportunities that remain open to them.
By coming to this event we aim to:
• Attract UK companies to submit proposals for European funding calls
• Provide clarity to UK organisations on the need and value to participate in EU competitions
• Attract non-aerospace organisations to the competitions
• Create an opportunity for new collaborations and connections
Metrology provides the basis for the determination of absolute values, comparability of measurements and the use of data in models. Thus, traceability to SI units is a fundamentally necessary quality assurance for any scientifically reliable result.
The traceability chain for a new metrological challenge does not always already exist. For example, the focus of metrology for radon activity concentration was in the area of radiation protection and the need to measure the relatively high radon activity concentrations in buildings. However, this type of traceability is not sufficient to provide traceability for high-resolution measurements of the very low radon activity concentration in the atmosphere. The traceRadon project aims to solve this gap in metrology, by extending the traceability chain from 100 Bq m-3 to 1 Bq m-3. This involves the development of new sources, new reference atmospheres and new transfer standards. By this, the use of radon as a tracer in climate observation is supported.
This presentation gives an overview on the basic needs for low level radon activities in regard to traceability, relevant standards and available facilities. Early results achieved by the traceRadon1 project will be included.
1This project 19ENV01 traceRadon has received funding from the EMPIR programme co-financed by the Participating States and from the European Union's Horizon 2020 research and innovation programme.
Horizon 2020 Space: Information and Consortia Building Event - SlidesKTN
The workshop gave an overview of H2020 Space Call topics, and support available for UK organisations in how to apply for funding, as well as information on Brexit and the continuation of UK participation in H2020. This enabled attendees to gain an insight into the benefits of participating, guidelines for preparing a project outline and the support and collaboration tools available.
In summary, the event covered:
1. Gather information on forthcoming 2020 topics;
2. Discuss and refine your project ideas with potential partners;
3. Join consortia forming around forthcoming 2020 topics.
Find out more about the EU Programmes Interest Group at https://ktn-uk.co.uk/programmes/eu-programmes
Follow the KTN EU team on Twitter for news on EU funding: https://twitter.com/ktnuk_eu
Rare earth free motor designs - ReFreeDrive projectLeonardo ENERGY
Webinar recording at https://youtu.be/PziGDLfE5_w
ReFreeDrive objective is to develop motor designs avoiding the use of rare earth magnets. Instead, abundant materials such as steel, ferrite and copper are used.
As Europe's leading economic powerhouse and the fourth-largest hashtag#economy globally, Germany stands at the forefront of innovation and industrial might. Renowned for its precision engineering and high-tech sectors, Germany's economic structure is heavily supported by a robust service industry, accounting for approximately 68% of its GDP. This economic clout and strategic geopolitical stance position Germany as a focal point in the global cyber threat landscape.
In the face of escalating global tensions, particularly those emanating from geopolitical disputes with nations like hashtag#Russia and hashtag#China, hashtag#Germany has witnessed a significant uptick in targeted cyber operations. Our analysis indicates a marked increase in hashtag#cyberattack sophistication aimed at critical infrastructure and key industrial sectors. These attacks range from ransomware campaigns to hashtag#AdvancedPersistentThreats (hashtag#APTs), threatening national security and business integrity.
🔑 Key findings include:
🔍 Increased frequency and complexity of cyber threats.
🔍 Escalation of state-sponsored and criminally motivated cyber operations.
🔍 Active dark web exchanges of malicious tools and tactics.
Our comprehensive report delves into these challenges, using a blend of open-source and proprietary data collection techniques. By monitoring activity on critical networks and analyzing attack patterns, our team provides a detailed overview of the threats facing German entities.
This report aims to equip stakeholders across public and private sectors with the knowledge to enhance their defensive strategies, reduce exposure to cyber risks, and reinforce Germany's resilience against cyber threats.
Adjusting primitives for graph : SHORT REPORT / NOTESSubhajit Sahu
Graph algorithms, like PageRank Compressed Sparse Row (CSR) is an adjacency-list based graph representation that is
Multiply with different modes (map)
1. Performance of sequential execution based vs OpenMP based vector multiply.
2. Comparing various launch configs for CUDA based vector multiply.
Sum with different storage types (reduce)
1. Performance of vector element sum using float vs bfloat16 as the storage type.
Sum with different modes (reduce)
1. Performance of sequential execution based vs OpenMP based vector element sum.
2. Performance of memcpy vs in-place based CUDA based vector element sum.
3. Comparing various launch configs for CUDA based vector element sum (memcpy).
4. Comparing various launch configs for CUDA based vector element sum (in-place).
Sum with in-place strategies of CUDA mode (reduce)
1. Comparing various launch configs for CUDA based vector element sum (in-place).
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...John Andrews
SlideShare Description for "Chatty Kathy - UNC Bootcamp Final Project Presentation"
Title: Chatty Kathy: Enhancing Physical Activity Among Older Adults
Description:
Discover how Chatty Kathy, an innovative project developed at the UNC Bootcamp, aims to tackle the challenge of low physical activity among older adults. Our AI-driven solution uses peer interaction to boost and sustain exercise levels, significantly improving health outcomes. This presentation covers our problem statement, the rationale behind Chatty Kathy, synthetic data and persona creation, model performance metrics, a visual demonstration of the project, and potential future developments. Join us for an insightful Q&A session to explore the potential of this groundbreaking project.
Project Team: Jay Requarth, Jana Avery, John Andrews, Dr. Dick Davis II, Nee Buntoum, Nam Yeongjin & Mat Nicholas
Techniques to optimize the pagerank algorithm usually fall in two categories. One is to try reducing the work per iteration, and the other is to try reducing the number of iterations. These goals are often at odds with one another. Skipping computation on vertices which have already converged has the potential to save iteration time. Skipping in-identical vertices, with the same in-links, helps reduce duplicate computations and thus could help reduce iteration time. Road networks often have chains which can be short-circuited before pagerank computation to improve performance. Final ranks of chain nodes can be easily calculated. This could reduce both the iteration time, and the number of iterations. If a graph has no dangling nodes, pagerank of each strongly connected component can be computed in topological order. This could help reduce the iteration time, no. of iterations, and also enable multi-iteration concurrency in pagerank computation. The combination of all of the above methods is the STICD algorithm. [sticd] For dynamic graphs, unchanged components whose ranks are unaffected can be skipped altogether.
Explore our comprehensive data analysis project presentation on predicting product ad campaign performance. Learn how data-driven insights can optimize your marketing strategies and enhance campaign effectiveness. Perfect for professionals and students looking to understand the power of data analysis in advertising. for more details visit: https://bostoninstituteofanalytics.org/data-science-and-artificial-intelligence/
A Framework for Selecting Deep Learning Hyper-Parameters
1. A Framework for Selecting
Hyper-Parameters
Jim O’ Donoghue
In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979
British International Conference on Databases
7th July 2015
2. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979
Background + Motivation
Algorithms + The CDN
Experiments + Results
Future Work
Conclusions
NEED TO FIX NUMBERS
3. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 3
1
4. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 4
1
5. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 5
1
Dementia Awareness
+ Prevention
6. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 6
1
Dementia Awareness + Prevention
Online Environment
7. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 7
1
Dementia Awareness + Prevention
Online Environment
Risk Prediction Algorithm
8. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 8
1
Dementia Awareness + Prevention
Online Environment
Risk Prediction Algorithm
- Validation
9. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 9
3
10. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 10
3
11. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 11
3
High-Dimensional
12. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 12
3
High-Dimensional
Variable
Interactions
13. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 13
3
High-Dimensional
Variable
Interactions
14. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 14
3
High-Dimensional
Variable
Interactions
Hyper-Parameter
Selection
15. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 15
3
High-Dimensional
Variable
Interactions
Hyper-Parameter
Selection
16. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 16
4
17. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 17
4
ClassClass
Input
Features
18. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 18
4
ClassClass
Input
Features
Learned
Features
19. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 19
4
Class
Connection
Weights
Class
Input
Features
Learned
Features
20. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 20
4
Class
Connection
Weights
Class
Input
Features
Learned
Features
21. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 21
4
Class
Connection
Weights
Class
Input
Features
Learned
Features
22. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 22
4
Class
Connection
Weights
Class
Input
Features
Learned
Features
23. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 23
4
Class
Connection
Weights
Class
Input
Features
Learned
Features
24. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 24
4
Class
Connection
Weights
Class
Input
Features
Learned
Features
25. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 25
4
Class
Connection
Weights
Class
Input
Features
Learned
Features
26. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 26
4
Class
Connection
Weights
Class
Input
Features
Learned
Features
27. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 27
4
Connection
Weights
Class
Input
Features
Learned
Features
28. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 28
5
Class
Connection
Weights
Class
Input
Features
29. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 29
6
Connection
Weights
Class
Input
Features
30. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 30
7
Connection
Weights
Input
Features
Learned
Features
31. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 31
8
Class
Connection
Weights
Input
Features
Learned
Features
32. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 32
9
33. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 33
9
MySql
File
System
Configurable Deep Network
Framework
34. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 34
9
MySql
File
System
Grid
Algorithm
Configurable Deep Network
Framework
35. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 35
9
MySql
File
System
Grid
Algorithm
Configurable Deep Network
Framework
e1, e2, … , en
36. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 36
9
MySql
File
System
Grid
Algorithm
Configurable Deep Network
Framework
Query
e1, e2, … , en
37. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 37
9
MySql
File
System
Grid
Algorithm
Configurable Deep Network
Framework
Final Model
Query
e1, e2, … , en
38. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 38
10
39. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 39
11
CDN
40. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 40
12
Subset of the Data – dimensions
What the variables are
What the predictor is
Purpose
41. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 41
13
To Choose:
42. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 42
13
To Choose:
learning rate α
43. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 43
13
To Choose:
learning rate α
weight decay term λ
44. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 44
13
To Choose:
learning rate α
weight decay term λ
training iterations t
45. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 45
13
The Grid:
46. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 46
13
The Grid:
α, λ:
[0.001, 0.003, 0.009, … , 0.1, 0.3, 0.9]
47. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 47
13
The Grid:
α, λ:
[0.001, 0.003, 0.009, … , 0.1, 0.3, 0.9]
t:
[100, 1000, 10000]
49. Alpha 0.9 0.3 0.09 0.003
0
5
10
15
20
25
30
35
40
45
50
Valid.
Cost
In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 49
13
53. 0
5
10
15
20
25
30
35
40
45
50
Valid.
Cost
In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 53
13
100
1,000
10,000
Training Iterations
Categorical Continuous Lambda 0.009 0.003 0.001
Alpha
0.9 0.3
0.09 0.003
54. 0
5
10
15
20
25
30
35
40
45
50
Valid.
Cost
In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 54
13
100
1,000
10,000
Training Iterations
Categorical Continuous Lambda 0.009 0.003 0.001
Alpha
0.9 0.3
0.09 0.003
55. 0
5
10
15
20
25
30
35
40
45
50
Valid.
Cost
In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 55
13
100
1,000
10,000
0.3046
Training Iterations
Categorical Continuous Lambda 0.009 0.003 0.001
Alpha
0.9 0.3
0.09 0.003
56. 0
5
10
15
20
25
30
35
40
45
50
Valid.
Cost
In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 56
13
100
1,000
10,000
0.3046
Training Iterations
Categorical Continuous Lambda 0.009 0.003 0.001
Alpha
0.9 0.3
0.09 0.003
57. 0
5
10
15
20
25
30
35
40
45
50
Valid.
Cost
In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 57
13
100
1,000
10,000
0.3046
Training Iterations
Categorical Continuous Lambda 0.009 0.003 0.001
Alpha
0.9 0.3
0.09 0.003
58. 0
5
10
15
20
25
30
35
40
45
50
Valid.
Cost
In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 58
13
100
1,000
10,000
0.3046 0.2815
Training Iterations
Categorical Continuous Lambda 0.009 0.003 0.001
Alpha
0.9 0.3
0.09 0.003
59. 0
5
10
15
20
25
30
35
40
45
50
Valid.
Cost
In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 59
13
100
1,000
10,000
0.3046 0.2815
Training Iterations
Categorical Continuous Lambda 0.009 0.003 0.001
Alpha
0.9 0.3
0.09 0.003
60. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 60
15
To Choose:
61. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 61
15
To Choose:
layer 1 nodes h(1)
n
62. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 62
15
To Choose:
layer 1 nodes h(1)
n
pre-training epochs e
63. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 63
15
The Grid:
h(1)
n:
[10, 30, 337, 900, 1300, 2000]
64. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 64
15
The Grid:
h(1)
n:
[10, 30, 337, 900, 1300, 2000]
e
[1, 5, 10, 15, 20]
65. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 65
15
Parameter Initialisation:
− 4
6
𝑓𝑎𝑛_𝑖𝑛 + 𝑓𝑎𝑛_𝑜𝑢𝑡
, + 4
6
𝑓𝑎𝑛_𝑖𝑛 + 𝑓𝑎𝑛_𝑜𝑢𝑡
72. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 72
14
To Choose:
Last layer nodes h(1)
n
73. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 73
14
To Choose:
Last layer nodes h(1)
n
The Grid:
[10, 30, 337, 900, 1300, 2000]
74. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 74
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Categorical Continuous
75. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 75
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Nodes
10 30 100
337 900 1300
2000
76. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 76
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Categorical Continuous
Nodes
10 30 100
337 900 1300
2000
77. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 77
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Categorical Continuous
Nodes
10 30 100
337 900 1300
2000
78. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 78
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Categorical Continuous
Nodes
10 30 100
337 900 1300
2000
79. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 79
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Categorical Continuous
Nodes
10 30 100
337 900 1300
2000
80. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 80
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Categorical Continuous
Nodes
10 30 100
337 900 1300
2000
0.232
81. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 81
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Categorical Continuous
Nodes
10 30 100
337 900 1300
2000
0.232
82. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 82
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Categorical Continuous
Nodes
10 30 100
337 900 1 1300
2000
0.232
83. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 83
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Categorical Continuous
Nodes
10 30 100
337 900 1 1300
2000
0.232
84. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 84
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Categorical Continuous
Nodes
10 30 100
337 900 1 1300
2000
0.232 0.291
85. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 85
14
0
10
20
30
40
50
60
70
80
90
Valid.
Cost
Categorical Continuous
Nodes
10 30 100
337 900 1 1300
2000
0.232 0.291
86. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 86
16
Lambda @ 0.03
10
2000
200
337
10
200
3567
10
200
3567
10
200
3567
10
200
3567
10
30
10
337
10
100
337
0
1
2
3
4
5
6
7
Alpha 0.001 0.01 0.9
87. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 87
16
Lambda @ 0.03
10
2000
200
337
10
200
3567
10
200
3567
10
200
3567
10
200
3567
10
30
10
337
10
100
337
0
1
2
3
4
5
6
7
Step 3000 1000 100
88. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 88
16
Lambda @ 0.03
10
2000
200
337
10
200
3567
10
200
3567
10
200
3567
10
200
3567
10
30
10
337
10
100
337
0
1
2
3
4
5
6
7
0.265 0.245
Alpha 0.001 0.01 0.9 Steps 3000 1000 100
0.272
89. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 89
16
Lambda @ 0.03
10
2000
200
337
10
200
3567
10
200
3567
10
200
3567
10
200
3567
10
30
10
337
10
100
337
Alpha 0.001 0.01 0.9 Steps 3000 1000 100
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
90. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 90
18
Activation functions
Algorithms
Inference
Framework – to Mongo and input from
Visualising learning
Implementing Early Stopping
Mini-batch Stochastic Gradient Descent
91. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 91
19
Much easier to model when you have
one extensible network that can handle
many type of data
Constituent models can be used to select
a starting point for deep learning
configurations
92. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 92
93. In-MINDD is funded under the European Union Seventh Framework Programme, Grant Agreement Number 304979 93
16
Lambda @ 0.03
10
2000
200
337
10
200
3567
10
200
3567
10
200
3567
10
200
3567
10
30
10
337
10
100
337
0
1
2
3
4
5
6
7
0.265 0.245
Alpha 0.001 0.01 0.9 Steps 3000 1000 100
0.272