Artificial Immune systems Flashcards
Artificial Immune Systems
What AISs (usually) do is detect rare/suspicious events, by borrowing computational ideas from the IS
The Role of the Immune System
It protects our bodies from infection:
Nonspecific defences :
body’s first line against disease. they are not directed against a particular pathogen. they guard against all
nfections, regardless of their cause.
‒ A first line nonspecific line of defence:barriers
‒ A second nonspecific line of defence: general attack
Specific defences are attempts by the body to defend itself against particular pathogens:
‒ specific defence: Primary Immune respose
‒ specific defence: Secondary Immune respose
Pathogen
is any agent (bacterium, virus, etc) that can cause us
trouble
Lymphocytes
- B-Lymphocytes (B Cells)
the Bone marrow, are responsible for producing antibodies - T-Lymphocytes (T Cells) mature in the Thymus and have a variety of roles
Important Properties of the Immune System
- Recognition
- Distributed & Self Regulating
- Diversity
- Learning & Memory
- Metadynamics - non-used cells discarded, new cells created continuously
Artificial Immune Recognition System (AIRS)
Uses immune system metaphors to classify examples in a training dataset.
An immune system approach to computer security therefore needs to include:
- Distributed nature – lymphocytes find evidence of infection locally
- Diversity – keeping systems diverse means less likely to spread a virus/be vulnerable to attack
- Adaptability – new viruses/methods of attack should be recognised by the system
- Anomaly Detection
- Numbers – human immune systems must react quickly to large numbers of pathogens
Principles – Self/Non-Self Distinction
- Each machine/copy of software has a unique copy of the protection algorithm
- Detection is probabilistic
- A robust system – should look for any foreign activity not known signatures
self/non self distinction how it works
Antibodies
Training Phase
Affinity Maturation
Self Recognition
Non-Self Detection
Anomaly Detection and Classification
Antibodies
computational entities that mimic the role of immune system antibodies. Antibodies are designed to recognize and bind to specific antigens, which represent patterns or entities to be identified or classified.
Training Phase
During the training phase of AIS, the system is exposed to a set of training patterns that are labeled as either self or non-self. The antibodies learn to differentiate between these patterns and build a repertoire of self and non-self recognition.
Affinity Maturation:
The antibodies undergo an affinity maturation process, similar to the immune system’s clonal selection and affinity maturation processes. Through iterative cloning, mutation, and selection, the antibodies with higher affinity towards self patterns are amplified, while those with lower affinity are suppressed or eliminated.
Self Recognition:
As the AIS evolves, the antibodies become specialized in recognizing the self patterns encountered during the training phase. They develop a strong affinity for self patterns and exhibit a lower response to non-self patterns.
Non-Self Detection
When the AIS encounters a new, unseen pattern or entity, the antibodies compare its characteristics to the learned self patterns. If the pattern deviates significantly from the learned self patterns, the antibodies exhibit a stronger response.
Anomaly Detection and Classification:
The self/non-self distinction allows AIS to detect anomalies or deviations from the learned self patterns. By comparing the characteristics of an input pattern to the stored self patterns, AIS can classify the input as either self (normal) or non-self (anomalous or potentially harmful).
