Implementing Immune Repertoire Models Using Weighted Finite State Machines

Gijs Schröder,Inge MN Wortel,Johannes Textor

2023-08-07

Abstract:The adaptive immune system's T and B cells can be viewed as large populations of simple, diverse classifiers. Artificial immune systems (AIS) $\unicode{x2013}$ algorithmic models of T or B cell repertoires $\unicode{x2013}$ are used in both computational biology and natural computing to investigate how the immune system adapts to its changing environments. However, researchers have struggled to build such systems at scale. For string-based AISs, finite state machines (FSMs) can store cell repertoires in compressed representations that are orders of magnitude smaller than explicitly stored receptor sets. This strategy allows AISs with billions of receptors to be generated in a matter of seconds. However, to date, these FSM-based AISs have been unable to deal with multiplicity in input data. Here, we show how weighted FSMs can be used to represent cell repertoires and model immunological processes like negative and positive selection, while also taking into account the multiplicity of input data. We use our method to build simple immune-inspired classifier systems that solve various toy problems in anomaly detection, showing how weights can be crucial for both performance and robustness to parameters. Our approach can potentially be extended to increase the scale of other population-based machine learning algorithms such as learning classifier systems.

Neural and Evolutionary Computing

What problem does this paper attempt to address?

### What problems does this paper attempt to solve?

This paper aims to solve the challenges encountered by artificial immune systems (AIS) when dealing with large - scale data, especially problems in representing and modeling immune cell repertoires. Specifically:

1. **Scale problem**: Traditional artificial immune systems based on finite - state machines (FSM) can efficiently compress and represent immune cell repertoires, but they are unable to handle the multiplicity in input data, that is, the frequency at which each string or pattern appears. This limits their performance and robustness in practical applications.

2. **Information complexity**: Since traditional FSM - based AIS treats all detectors as having equal weights and ignores the differences in the importance of different detectors, these models are unable to store and process more complex information.

3. **Algorithm scalability**: Existing AIS models are difficult to scale to larger - scale data sets, especially when a large number of pattern - matching tasks need to be handled. This makes them perform poorly when facing complex problems in the real world.

To solve these problems, the author proposes a new weighted finite - state machine (WFSM) method, which improves existing models by assigning weights to each detector. This method can not only effectively handle the multiplicity in input data but also improve the performance and robustness of the model. In addition, the author also shows how to use WFSM to implement weighted positive selection and negative selection algorithms and verifies its advantages in tasks such as anomaly detection through experiments.

### Main contributions

1. **Define weighted selection algorithms**: Propose weighted positive selection and negative selection algorithms, in which each detector is given a different weight according to the number of input samples it recognizes.

2. **Efficient implementation**: Show how to efficiently implement these weighted selection algorithms using compressed weighted finite - state machines (WFSM).

3. **Performance verification**: Compare the performance of weighted and non - weighted models through simple string anomaly detection problems, and prove the advantages of weighted models in terms of performance and parameter robustness.

4. **Expansion potential**: Discuss the potential applications of this method in other population - based machine - learning algorithms, such as learning classifier systems (LCS).

Through these improvements, the author hopes to effectively simulate and understand the functions of the adaptive immune system in larger - scale and more complex scenarios.