An Agent-based virtual organization for risk control in

large enterprises

M Lourdes Borrajo1 and Juan M. Corchado2,3

1 Universityof Vigo lborrajo@uvigo.es
2 Universityof Salamanca corchado@usal.es
3Osaka Institute of Technology

Abstract. At present, business decision making is a crucial task in every

enterprise as it allows to minimize risks and maximize benefits. For effective
decision making, large corporations and enterprises need tools that will help
them detect inefficient activities in their internal processes. This article presents
a virtual organization of agents designed to detect risky situations and provide
recommendations to the internal auditors of large corporations. Each agent
within the virtual organization facilitates the interconnection of enterprises with
the central decision node of the corporation. The core of the agent-based virtual
organization consists of two agents: one that is specialized in detecting risky
situations in all aspects of business enterprise and an advisor agent which
communicates with the evaluator agents of the different departments of a
business and provides decision support services. This paper presents a real-case
scenario which includes small and medium enterprises, the results demonstrate
the feasibility of the proposed architecture.

Keywords: Agent-based virtual organizations, internal audit, case

1. Introduction

Systems used in the organization of enterprises are becoming increasingly complex.

In addition, several regulatory standards have been introduced which can make it
more difficult for enterprises to make decisions. As a result, it is necessary to propose
decision support systems which will complement existing business models and will
help enterprises improve their functioning. Such tools and methods can further reduce
risks by providing recommendations based on previous experiences. The periodic
realization of internal audits is necessary. However, due to their dynamism, the
evaluation and prediction of financial systems is, in general, a complicated task. It is
necessary to construct models that facilitate the analysis of changing environments.
The wide range of processes involved in the running a business can be classified into
different functional areas[1]. Each one of these areas is denominated a “Function”. A
Function is a group of coordinated and related activities which are necessary to reach
the objectives of the firm and are carried out in a systematic and reiterated way [2].
Functions are divided into activities, which are closely related to clearly defined
objectives. Commonly performed functions in a firm are: Purchases, Treasury, Sales,
Information Technology, Immobilized, Legal Normative Execution and Personal
Politics. In turn, each one of these functions is broken down into a series of activities.
For example, the function Information Technology is divided into the following:
Computer Plan Development, Study of Systems, Installation of Systems, Treatment of
the Information Flows and Security Management. Each activity is composed of a
number of tasks, for example: register, authorise, approve, harmonise, separate
obligations, operate, etc. Control procedures are conducted to ensure that the
established objectives are achieved. This article proposes an innovative approach
which uses agent-based virtual organizations to propose a model for risk management
in business corporations.
In problems like the one presented in this document, standard monitoring and
prediction techniques cannot be applied due to the complexity of the problem, the
existence of preliminary knowledge, the great dynamism of the system, etc. Thus, it is
necessary to use models that combine the advantages of several problem solving
mechanisms, capable of addressing the numerous elements of a general problem and
also of solving singular problems. In this regard, an adaptive system has been
developed. Its flexibility allows it to behave in different ways and to evolve,
depending on the state of the environment in which it operates. Virtual organizations
of agents are the most prevalent solution to constructing Artificial Intelligence
distributed systems [3]. Agents are computational entities that are characterized by
their autonomy, reactivity, pro-activity, social abilities, reasoning, learning and
mobility [4, 5, 6, 7, 8]. These characteristics make agent-based virtual organizations
very suitable for constructing intelligent environments. An agent can act as an
interface between the user and the rest of the elements of an intelligent environment
[9, 10, 11, 12, 13]. Moreover, intelligent agents can incorporate advanced artificial
intelligence models to predict risky situations. In this study we propose a distributed
approach where the components of a large enterprise are modelled as agent-based
virtual organizations which collaborate to create models that can evolve over time and
adapt to the changing conditions of the environment. Thus, they make it possible to
detect situations that imply a risk for the enterprises and providing suggestions and
recommendations that can help to avoid undesirable situations. The core of the agent-
based virtual organization are the evaluator and advisor agents, they incorporate new
techniques for analysing enterprise data, extract the relevant information, and detect
possible failures or inefficiencies in the operation processes.
The article is structured as follows: the next section briefly introduces the problem
that motivates this research. Section 3 presents the designed agent-based virtual
organization. Sections 4 and 5 outline the results obtained after testing the system and
the conclusions drawn from this study.

2. Risk management in Enterprises

“Risk Management” is a broad term for a business discipline that is focused on

protecting the assets and profits of an organization. - This is done by reducing the
possibility of unfortunate events occurring. In case an unfortunate event does occur,
its impact can be mitigated, and measures are taken to recover from the losses rapidly.
Risk management encompasses a series of steps that include risk identification, the
measurement and evaluation of exposures, exposure reduction or elimination, risk
reporting, and risk transfer and/or financing for losses that may occur. All
organizations manage risks in multiple ways, depending on the exposure they must
address [14]. Changes in the economic environment are increasingly pressuring
companies to address risks at the highest levels of the organization. Companies that
incorporate a strategic approach to risk management use specialized tools and have
more structured and frequent reporting on risk management. As such, they are in a
better position to ensure that risk management provides relevant and applicable
information that meets the needs of the organization and the executive team. But no
matter what an organization’s approach is, the tools used must be backed up by solid,
actionable reporting. To get their voice heard, risk managers should not only conduct
their own analysis, but they should also work on forging strong relationships with
internal auditors and other departments that will provide them with additional risk
reports [15].
Enterprise Risk Management (ERM) is defined as "a process, effected by an
entity's board of directors, management and other personnel, applied in strategy-
setting and across the enterprise, designed to identify potential events that may affect
the entity, and manage risk to be within its risk appetite, to provide reasonable
assurance regarding the achievement of entity objectives." [16]. The management of
risks and uncertainties is central to the survival and performance of organizations.
Enterprise risk management (ERM) is an emerging approach to managing risks across
different business functions in an organisation and it represents a shift from
paradigms that focus on managing specific risks [17, 18]. This paper provides a web
intelligent model to ERM, which will subsequently lead to better organisational
performance.

3. Agent-based virtual organization for Risk management

Over the last decades, agent and multi-agent systems have become important tools
in a range of fields [19, 20]. Agent-based virtual organizations are distributed
organizations where the components (agents) collaborate to achieve a series of goals
[21, 22]. Agents are autonomous entities that are characterized by their capabilities:
autonomy, proactivity, reactivity, social skills, organization, etc. These capabilities
make agents ideal, problem solving tools in business intelligence and risk
management scenarios. In these scenarios the agents can play different roles and
establish an organizational model which emulates human behaviours and management
processes. In this regard, it is possible to obtain different agent types, specialized in
concrete tasks and behaviours, that can collaborate to increase the productivity of the
business and manage risks effectively. Agents can act as an interface between human
users and systems, trying to provide advanced facilities and personalize the access to
the system, but also can act as autonomous entities that are proactive and can make
decisions independently. This paper leverages virtual organizations of agents to
provide advanced capacities for risk management in large enterprises. The system
provides a web system interface to facilitate remote interaction with the human users
involved in the risk management process. The core of each agent-based virtual
organization is a special type of agent, called CBR-BDI agent [23, 24, 25]. This agent
type integrates a case-based reasoning mechanism (CBR) in its internal structure to
take advantage of the reasoning abilities of the CBR paradigm [26, 27, 28, 29, 30, 31,
32, 33]. CBR-BDI agents are characterized by their learning capacities and their
ability to adapt to dynamic environments. These agent types are used to evaluate the
business' status and to generate recommendations that can help the business avoid
risky situations. CBR-BDI agents collaborate with other deliberative agents [34] in
the system to find optimum models for risk management.
The agents in the system allow the users to access the system through distributed
applications, which run on different types of devices and interfaces (e.g. computers,
cell phones, etc). Figure 1 shows the basic schema of the proposed architecture, where
all requests and responses are handled by the agents in the platform. The system is
modelled as a modular multi-agent architecture, where deliberative BDI agents can
cooperate, propose solutions to very dynamic environments, and face real problems,
even when they have a limited description of the problem and few resources
available. There are various kinds of agents in the architecture, each one with specific
roles, capabilities and characteristics:
- Business Agent. This agent was assigned to each of the departments in an
enterprise, . These agents collect new data and allow consultations. The
enterprise can interact with the system by means of this agent, by introducing its
information and receiving predictions.
- Evaluator Agent. It is responsible for evaluating and predicting situations that
may entail risks. This agent goes through the four CBR stages that allow it to
estimate the state of an activity. In the retrieval stage, the multi-agent system
identifies situations from the past which are similar to the current situation and
retrieves them from the case base. The information contained in these cases is
adapted and reused in the current situation, in this way an initial estimate of the
state of the activity is generated. Old situations serve as a basis for detecting
inefficient processes within the activity and allow to correct them by choosing the
best options. In this way, an enterprise can establish the extent to which an
activity is risky, define clear objectives and mitigate unfortunate events. its
function, and the company itself, to develop in a more positive way. The retain
phase allows the system to learn from past cases and as a result it evolves
together with the company by basing its corrective actions on the calculation of
previous errors.
- Advisor agent. The objective of this agent is to provide recommendations to help
the internal auditor decide which actions they should take and in this way,
improve the company’s internal and external processes.
- Expert Agent. This agent helps the auditors and enterprise control experts that
collaborate in the project to provide information and feedback to the multi-agent
system. These experts generate prototypical cases from their experience and they
receive assistance in developing the Store agent case-base.
- Storage Agent. This agent’s memory has been fed with cases constructed with
information provided by the enterprise (through its agent) and with prototypical
 cases identified by 34 enterprises control experts, using personal agents who have
collaborated and supervised the developed model.
- The Evaluator and Advisor agents are CBR-BDI agents with advanced reasoning
abilities that provided great adaptation and learning capacities.
- The Evaluator and Advisor agent use the same type of case and share the same
memory of cases. The data for the cases were obtained by Expert agents from the
surveys conducted with enterprise experts in the different functional areas of
various enterprises. This type of survey attempts to reflect the experience of the
experts in the different fields. For each activity, the survey presents two possible
situations: the first one tries to reflect the situation of an activity with an incorrect
activity state, and the second one tries to reflect the situation of an activity with a
satisfactory activity state. Both situations will be evaluated by a human expert
using a percentage. Each activity is composed of tasks, and each task has an
importance rate, and values of realization for both incorrect and satisfactory
activity state. The data acquired by means of surveys were used to build the
prototype cases for the initial Storage agent case base.

Figure 1. Virtual organization of agent.

The internal structure of the CBR-BDI Evaluator and Advisor agents is similar to
the architectures proposed in previous works [23, 24, 25]. In the present work, we use
a Maximum Likelihood Hebbian Learning (MLHL) based model [35] to automate the
process of case indexing and retrieval in the evaluator agent. The reuse stage
incorporates an innovative mixture of experts that makes use of multilayer perceptron,
support vector regression and radial basis function neural network [36, 37]. The revise
and retain stages implement a decision support system for experts. Moreover, the
knowledge obtained during the prediction process is of great importance for
subsequent predictions. On the other hand, the advisor agent is specialized in
providing recommendations that help avoid risky situations and improve the overall
functioning of the company. The retrieve phase recovers similar cases and their
corresponding solutions. The reuse phase incorporates a novel approach based on
decision trees and probabilistic gain functions to assess efficient and inefficient tasks.
The revise and retain stages also implement a decision support system for experts.
The use of Maximum Likelihood Hebbian Learning Based Method derives from the
work of several authors [38, 39, 40, 41, etc]. in the field of pattern recognition as an
extension of Principal Component Analysis (PCA) [42, 43]. The network operation is:
N (1)
Feedforward: y i = ∑W
j =1
ij x j , ∀i

M (2)
Feedback: e j = x j − ∑Wi =1
ij yi

Weights change: ∆Wij = η. yi .sign e j | e j | ( ) p −1 (3)

Applying equations 1 to 3 to the Case-base, the MLHL algorithm groups the cases in
clusters automatically. The proposed indexing mechanism classifies the
cases/instances automatically, clustering together those of similar structure. This
technique attempts to find interesting low dimensional projections of the data so that
humans can investigate the structure of the data without any tools. One of the greatest
advantages of this technique is that it is an unsupervised method, so we do not need to
have any information about the data beforehand. When a new case is presented to the
CBR system, its type is classified by applying equations 1 to 3 to it. This mechanism
may be used as a universal retrieval and indexing mechanism for any similar problem
to the one presented in this work.

4. Experimental results and case study

A case study aimed at providing innovative web business intelligence tools for the
management of large Corporations was carried out in the region of Castile and León,
Spain. The experiment consisted in constructing the initial prototype from case
memory and then in predicting unfortunate events for the enterprises taken into
considerations and providing recommendations. The case study presented in this work
was oriented to detect possible risky situations in large corporations, considering the
market crisis. A multi-agent system was implemented and 26 branches/subsidiary
companies that belong to 2 large corporations, participated in the experiment and
were assigned a personal business agent. The enterprises were situated in different
sectors of the Spanish market. The economic context is the same for all the
Companies. The system was tested for 24 months, from January 2015 to January
2017, tuned and improved given that knowledge was acquired from a total of 758
cases. The evolution of the enterprise was monitored by analysing its internal
activities and predictions were made based on the previous experiences and on the
situation of the market (the possible crisis that affect the market). The experts
complete a survey to provide information about the enterprise.
To validate the overall functioning of the system it was necessary to individually
assess the performance of the Evaluator and Advisor agents. These virtual agents
provide predictions on the performance of the activities and detect the processes that
can be improved within each activity, achieving an overall improvement. In the
following paragraphs we will focus on the evaluation of the CBR-BDI agents and
their influence on the multi-agent system. To validate the performance of the
Evaluator agent, an estimation of the efficiency of the predictions provided by this
agent was carried out. To evaluate the significance of the different techniques
integrated within the Evaluator agent, a cross validation was established, following
the Dietterich's 5x2- Cross-Validation Paired t-Test algorithm [44]. Value 5 in the
algorithm represents the number of replications of the training process and value 2 is
the number of sets in which the global set is divided. Thus, for each of the techniques,
the global dataset S was divided into two groups S1 and S2 as follows:
S = S1 ∪ S 2 and S1 ∩ S 2 = φ . Then, the learning and estimation processes were
performed. They were repeated 7 times and had the following steps: the system was
trained using S1 and then it was used to classify S1 and S2. In the second step, the
system was trained using S2 and then it was used to classify S1 and S2. To assess the
performance of the Advisor agents, it was necessary to take into account that the aim
of this agent which is to detect inefficient tasks by means of gain functions, as
explained in the previous section. To assess the functioning of the Advisor agent the
tasks which obtained higher values for the gain function were selected. The selected
tasks were used to estimate the different scenarios for different execution values for
the task. The estimation was performed using the values provided by the Evaluator
agent, obtaining a concrete value for the task.
The obtained results demonstrate that the organization based system contributed to
the positive development of all enterprises. This development was reflected in the
reduction of inefficient processes. The indicator used to determine the positive
evolution of the companies was the state of each of the activities analysed. After
analysing the company’s activities, it was necessary to prove that the state of the
activity (valued between 1 and 100) had increased beyond the state obtained in the
previous four-month period. The system considered small changes in the tasks
performed by the corporations, and all the experts that participated in the experiments
considered four months as a significant time to evaluate the evolution of a company
related to these changes. Figure 2 shows a global positive evolution of the different
activities of the enterprise and a set of recommendations for each of the activities,
consisting of suggestions that try to improve the efficiency of the activities. The
proposed virtual organization of agents can be considered as a unique system, since it
is useful for dynamic environments and open enough to be used in other enterprise
environments, the subsidiary companies of the corporation shared results via the
organization system, benefiting from this cooperation. The experts noted that the
behaviour of the system improved as the number of cases in the case base grows.
 Figure 2. Information on the global activities of a company provided by the
Evaluator Agent and recommendations provided by the Advisor Agent.

The system facilitates the identification of company inefficiencies and establishes

rational plans aimed at improving these processes. An average of 17 processes where
identified by the virtual organization as risky situations and among them an average
of 76% were successfully reconfigured with the help of the virtual agents with the
four-month window stablished for it.

5. Conclusions

In conclusion, the use of innovative tools in business intelligence can help detect
situations that involve potential risks and to get a better understanding of their internal
functioning. This article presented a virtual organization based multi-agent system
which models the behaviour of companies and provides a decision support tool, which
helps prevent unfortunate events from occurring by analysing all the processes that
compose each of the activities of a business. The proposed approach presents many
advantages over other proposals as it incorporates innovative techniques for
predicting risky situations and providing recommendations. Both strategies are based
on the integration of intelligent BDI agents within CBR systems and multiple experts
provide risk predictions. Moreover, the recommender system incorporates a new
strategy based on the use of probabilistic gain functions of decision trees for the
identification of inefficient tasks. The developed approach was implemented in
several Companies (considering the particular problem/casuistic of each company
within a large corporation). The case study demonstrated a high percentage of success
as the performance of the participating companies improved. Additional work is still
required to improve the adaptation of the CBR-BDI agents when the size of the
memory of cases is very high and to learn from the cases with invalid solutions.
Besides, it is necessary to increase the sample size by including more Companies in
the experiments as well as explore different sets of samples. These are our next
challenges.

Acknowledgements

This work has been funded by the Spanish Ministry of Science and Innovation
(TIN2015-65515-C4-3-R).

6. References

[1] Yañez, J.C., Borrajo, L. and Corchado, J.M. (2001). A Case-based Reasoning
System for Business Internal Control. Fourth International ICSC Symposium. Soft
Computing and Intelligent Systems for Industry. Paisley, Scotland, United
Kingdom, June 26-29, 2001.
[2] Mas, J. and Ramió, C. (1997). La Auditoría Operativa en la Práctica. Ed.
Marcombo, Barcelona.
[3]Rodriguez, S., Julián, V., Bajo, J., Carrascosa C., Botti, V., Corchado, J.M. (2011)
Agent-based virtual organization architecture, Engineering Applications of
Artificial Intelligence, 24(5), pp. 895-910
[4] Bajo, J., De Paz, Y., De Paz, J.F., Corchado, JM. (2009). Integrating Case
Planning and RPTW Neuronal Networks to Construct an Intelligent Environment
for Health Care. Expert Systems with Applications 36:3, 5844-5858.
[5] García Coria, J. A., Castellanos-Garzón, J. A., Corchado, J. M. (2014). Intelligent
business processes composition based on multi-agent systems. Expert Systems
with Applications, 41(4 PART 1), 1189–1205.
https://doi.org/10.1016/j.eswa.2013.08.003
[6] Tapia, D. I., Fraile, J. A., Rodríguez, S., Alonso, R. S., Corchado, J. M. (2013).
Integrating hardware agents into an enhanced multi-agent architecture for
Ambient Intelligence systems. Information Sciences, 222, 47–65.
https://doi.org/10.1016/j.ins.2011.05.002
[7] Costa, Â., Novais, P., Corchado, J. M., Neves, J. (2012). Increased performance
and better patient attendance in an hospital with the use of smart agendas. Logic
Journal of the IGPL, 20(4), 689–698. https://doi.org/10.1093/jigpal/jzr021
[8] García, E., Rodríguez, S., Martín, B., Zato, C., Pérez, B. (2011). MISIA:
Middleware infrastructure to simulate intelligent agents. Advances in Intelligent
and Soft Computing (Vol. 91). https://doi.org/10.1007/978-3-642-19934-9_14
[9] Tapia, D.I., Fraile, J.A., Rodríguez, S., Alonso, R.S., Corchado, J.M. (2013)
Integrating hardware agents into an enhanced multi-agent architecture for
Ambient Intelligence systems, Information Sciences, 222, pp. 47-65
[10] Rodríguez, S., De La Prieta, F., Tapia, D. I., Corchado, J. M. (2010). Agents and
computer vision for processing stereoscopic images. Lecture Notes in Computer
Science (including subseries Lecture Notes in Artificial Intelligence and Lecture
Notes in Bioinformatics) (Vol. 6077 LNAI). https://doi.org/10.1007/978-3-642-
13803-4_12
[11] Rodríguez, S., Gil, O., De La Prieta, F., Zato, C., Corchado, J. M., Vega, P., &
Francisco, M. (2010). People detection and stereoscopic analysis using MAS. In
INES 2010 - 14th International Conference on Intelligent Engineering Systems,
Proceedings. https://doi.org/10.1109/INES.2010.5483855
[12] Baruque, B., Corchado, E., Mata, A., Corchado, J. M. (2010). A forecasting
solution to the oil spill problem based on a hybrid intelligent system. Information
Sciences, 180(10), 2029–2043. https://doi.org/10.1016/j.ins.2009.12.032
[13] Tapia, D. I., & Corchado, J. M. (2009). An ambient intelligence based multi-
agent system for alzheimer health care. International Journal of Ambient
Computing and Intelligence, v 1, n 1(1), 15–26.
https://doi.org/10.4018/jaci.2009010102
[14] Valanarasu, R., Christy, A. (2018) Risk assessment and management in
enterprise resource planning by advanced system engineering theory. International
Journal of Business Intelligence and Data Mining, 13 (1-3), pp. 3-14.
[15] Namatame, A. (2018) Agent-based modeling of economic instability. Studies in
Computational Intelligence, 753, pp. 255-265.
[16] Ai, J., Brockett, P.L., Wang, T. (2017) Optimal Enterprise Risk Management
and Decision Making With Shared and Dependent Risks. Journal of Risk and
Insurance, 84 (4), pp. 1127-1169.
[17]Callahan, C., Soileau, J. (2017) Does Enterprise risk management enhance
operating performance?. Advances in Accounting, 37, pp. 122-139.
[18] Raschke, R.L., Mann, A. (2017) Enterprise content risk management: A
conceptual framework for digital asset risk management. Journal of Emerging
Technologies in Accounting, 14 (1), pp. 57-62.
[19] Tapia, D.I., Rodríguez, S., Bajo, J., Corchado, J.M. (2009) A SOA-based multi-
agent architecture, FUSION, Advances in Soft Computing, 50, pp. 99-107
[20] Jörg Bremer, Sebastian Lehnhoff. (2017) Decentralized Coalition Formation
with Agent-based Combinatorial Heuristics. ADCAIJ: Advances in Distributed
Computing and Artificial Intelligence Journal, Salamanca, v. 6, n. 3
[21] Rodríguez, S., Pérez-Lancho, B., De Paz, J.F., Bajo, J., Corchado, J.M. (2009)
Ovamah: Multiagent-based adaptive virtual organizations, 12th International
Conference on Information Fusion, FUSION 2009, 5203822, pp. 990-997
[22] Zato, C., Villarrubia, G., Sánchez, A., Barri I., Rubión E., Fernandez A., Rebate
C., Cabo J. A., Álamo R., Sanz J., Seco J., Bajo, J., Corchado, J.M. (2012)
PANGEA - Platform for automatic coNstruction of organizations of intelligent
agents, Advances in Intelligent and Soft Computing, 151 AISC, pp. 229-239
[23] Javier Bajo, María L. Borrajo, Juan F. De Paz, Juan M. Corchado, María A.
Pellicer (2012) A multi-agent system for web-based risk management in small and
medium business. Expert Systems with Applications, V. 39, Issue 8, Pages 6921-
6931
[24] Corchado, J. M., Pavón, J., Corchado, E. S., Castillo, L. F. (2004). Development
of CBR-BDI agents: A tourist guide application. In Lecture Notes in Computer
 Science (including subseries Lecture Notes in Artificial Intelligence and Lecture
Notes in Bioinformatics) (Vol. 3155, pp. 547–559). https://doi.org/10.1007/978-3-
540-28631-8
[25] Laza, R., Pavon, R., & Corchado, J. M. (2004). A reasoning model for
CBR_BDI agents using an adaptable fuzzy inference system. In Lecture Notes in
Computer Science (including subseries Lecture Notes in Artificial Intelligence and
Lecture Notes in Bioinformatics) (Vol. 3040, pp. 96–106). Springer, Berlin,
Heidelberg.
[26] Mata, A., & Corchado, J. M. (2009). Forecasting the probability of finding oil
slicks using a CBR system. Expert Systems with Applications, 36(4), 8239–8246.
https://doi.org/10.1016/j.eswa.2008.10.003
[27] Glez-Peña, D., Díaz, F., Hernández, J. M., Corchado, J. M., Fdez-Riverola, F.
(2009). geneCBR: A translational tool for multiple-microarray analysis and
integrative information retrieval for aiding diagnosis in cancer research. BMC
Bioinformatics, 10. https://doi.org/10.1186/1471-2105-10-187
[28] Corchado, J. A., Aiken, J., Corchado, E. S., Lefevre, N., Smyth, T. (2004).
Quantifying the Ocean’s CO2 budget with a CoHeL-IBR system. In Advances in
Case-Based Reasoning, Proceedings (Vol. 3155, pp. 533–546).
[29] Corchado, J. M., Borrajo, M. L., Pellicer, M. A., Yáñez, J. C. (2004). Neuro-
symbolic System for Business Internal Control. In Industrial Conference on Data
Mining (pp. 1–10). https://doi.org/10.1007/978-3-540-30185-1_1
[30] Fdez-Riverola, F., Corchado, J. M. (2003). CBR based system for forecasting
red tides. Knowledge-Based Systems, 16(5–6 SPEC.), 321–328.
https://doi.org/10.1016/S0950-7051(03)00034-0
[31] Corchado, J. M., Aiken, J. (2002). Hybrid artificial intelligence methods in
oceanographic forecast models. Ieee Transactions on Systems Man and
Cybernetics Part C-Applications and Reviews, 32(4), 307–313.
https://doi.org/10.1109/tsmcc.2002.806072
[32] Fyfe, C., Corchado, J. (2002). A comparison of Kernel methods for instantiating
case based reasoning systems. Advanced Engineering Informatics, 16(3), 165–
178. https://doi.org/10.1016/S1474-0346(02)00008-3
[33] Fyfe, C., & Corchado, J. M. (2001). Automating the construction of CBR
systems using kernel methods. International Journal of Intelligent Systems, 16(4),
571–586. https://doi.org/10.1002/int.1024
[34] Glez-Bedia, M., Corchado, J. M., Corchado, E. S., Fyfe, C. (2002). Analytical
model for constructing deliberative agents. International Journal of Engineering
Intelligent Systems for Electrical Engineering and Communications, 10(3).
[35] Corchado, J. M., Corchado, E. S., Aiken, J., Fyfe, C., Fernandez, F., &
Gonzalez, M. (2003). Maximum likelihood hebbian learning based retrieval
method for CBR systems. In Lecture Notes in Computer Science (including
subseries Lecture Notes in Artificial Intelligence and Lecture Notes in
Bioinformatics) (Vol. 2689, pp. 107–121). https://doi.org/10.1007/3-540-45006-
8_11
[36] Corchado, J. M., Fyfe, C. (1999). Unsupervised neural method for temperature
forecasting. Artificial Intelligence in Engineering, 13(4), 351–357.
https://doi.org/10.1016/S0954-1810(99)00007-2
[37] Corchado, J., Fyfe, C., Lees, B. (1998). Unsupervised learning for financial
forecasting. In Proceedings of the IEEE/IAFE/INFORMS 1998 Conference on
Computational Intelligence for Financial Engineering (CIFEr) (Cat.
No.98TH8367) (pp. 259–263). https://doi.org/10.1109/CIFER.1998.690316
[38] Corchado E., MacDonald D. and Fyfe C. (2002) Optimal Projections of High
Dimensional Data. ICDM '02. The 2002 IEEE International Conference on Data
Mining, IEEE Computer Society; Maebashi TERRSA, Maebashi City, Japan
December 9 - 12, 2002
[39] Fyfe C. and Corchado E. (2002). Maximum Likelihood Hebbian Rules. 10th
European Symposium on Artificial Neural Networks, ESANN’2002, Bruges,
April 24-25-26, 2002.
[40] Fyfe C. and Corchado E., (2002) A New Neural Implementation of Exploratory
Projection Pursuit. IDEAL2002 Third International Conference on Intelligent Data
Engineering and Automated Learning. Manchester, UK. 12-14 August, 2002.
[41] Fyfe C. and MacDonald D. (2001) ε-Insensitive Hebbian learning, Neuro
Computing.
[42] Oja E. (1989). Neural Networks, Principal Components and Subspaces,
International Journal of Neural Systems, 1:61-68.
[43] Oja E., Ogawa H. and Wangviwattana J. (1992) Principal Components Analysis
by Homogeneous Neural Networks, part 1, The Weighted Subspace Criterion,
IEICE Transaction on Information and Systems, E75D: 366-375. 1992.
[44] Dietterich, T.G. (1998). Approximate statistical tests for comparing supervised
classification learning algorithms. Neural Computation. 1895 - 1923.