Percorrer por autor "Correia, Manuel"
A mostrar 1 - 3 de 3
Resultados por página
Opções de ordenação
- A Hybrid AIS-SVM Ensemble Approach for Text ClassificationPublication . Antunes, Mário; Silva, Catarina; Ribeiro, Bernardete; Correia, ManuelIn this paper we propose and analyse methods for expanding state-of-the-art performance on text classification. We put forward an ensemble-based structure that includes Support Vector Machines (SVM) and Artificial Immune Systems (AIS). The underpinning idea is that SVM-like approaches can be enhanced with AIS approaches which can capture dynamics in models. While having radically different genesis, and probably because of that, SVM and AIS can cooperate in a committee setting, using a heterogeneous ensemble to improve overall performance, including a confidence on each system classification as the differentiating factor. Results on the well-known Reuters-21578 benchmark are presented, showing promising classification performance gains, resulting in a classification that improves upon all baseline contributors of the ensemble committee.
- TAT-NIDS: An Immune-Based Anomaly Detection Architecture for Network Intrusion DetectionPublication . Antunes, Mário; Correia, Manuel; Antunes, Mário;One emergent, widely used metaphor and rich source of inspiration for computer security has been the vertebrate Immune System (IS). This is mainly due to its intrinsic nature of having to constantly protect the body against harm inflicted by external (non-self) harmful entities. The bridge between metaphor and the reality of new practical systems for anomaly detection is cemented by recent biological advancements and new proposed theories on the dynamics of immune cells by the field of theoretical immunology. In this paper we present a work in progress research on the deployment of an immune-inspired architecture, based on Grossman's Tunable Activation Threshold (TAT) hypothesis, for temporal anomaly detection, where there is a strict temporal ordering on the data, such as network intrusion detection. We start by briefly describing the overall architecture. Then, we present some preliminary results obtained in a production network. Finally, we conclude by presenting the main lines of research we intend to pursue in the near future.
- Towards an immune-inspired temporal anomaly detection algorithm based on tunable activation thresholdsPublication . Antunes, Mário; Correia, Manuel; Carneiro, JorgeThe detection of anomalies in computer environments, like network intrusion detection, computer virus or spam classification, is usually based on some form of pattern search on a database of "signatures" for known anomalies. Although very successful and widely deployed, these approaches are only able to cope with anomalous events that have already been seen. To cope with these weaknesses, the "behaviour" based systems has been deployed. Although conceptually more appealing, they have still an impractical high rate of false alarms. The vertebrate Immune System is an emergent and appealing metaphor for new ideas on anomaly detection, being already adopted some algorithms and theoretical theories in particular fields, such as network intrusion detection. In this paper we present a temporal anomaly detection architecture based on the Grossman's Tunable Activation Threshold (TAT) hypothesis. The basic idea is that the repertoire of immune cells is constantly tuned according to the cells temporal interactions with the environment and yet retains responsiveness to an open-ended set of abnormal events. We describe some preliminary work on the development of an anomaly detection algorithm derived from TAT and present the results obtained thus far using some synthetic data-sets.