Bio-Inspired AI

The goal of this course is to study two of the main paradigms of Bio-Inspired Artificial Intelligence, namely: Evolutionary Computation, inspired by evolutionary biology, and Swarm Intelligence, inspired by collective behaviors of social animals. First, the main theories and algorithms will be introduced. Then, it will be shown how these techniques can be applied, for instance, for solving complex optimization problems, training data-driven models, generating new contents (video-games, websites, art), finding bugs in software, automatically synthesizing or fixing computer programs, or finding innovative solutions in robotics, logistics, and engineering. Finally, it will be shown how these techniques can help the understanding of biological systems, in order to close the loop between biology and AI.

At the end of this course, students will be familiar with the most important Evolutionary Computation and Swarm Intelligence techniques, and will be able to apply them to different contexts in industry, research, or even entertainment. They will also know the fundamentals for developing new algorithms and adapt them to new problems.

Material

All the course material is available (for UNITN students) on Moodle (course "Bio-Inspired Artificial Intelligence [145763]").

Student projects 2022-2023

A collection of some of the student projects that were presented as part of the exam:

• A Quality-Diversity Co-Evolutionary Scenario

• Ant Colony Optimization for Train Mapping System Optimization - Leveraging Existing Train-Line Network

• Bio-Inspired Image Vectorization - an Algorithm for Efficient Image Reconstruction

• Bio-Inspired Predator-Prey Simulation

• Combining Ant Colony Optimization with NeuroEvolution of Augmenting Topologies

• D2BL - Dynamic Distribution-driven Backpropagation Learning

• Efficient Public Transport Planning Using Ant Colony Optimization

• Evolution in Social Dilemma Games, Social Pressure and Emergent Behaviours

• Forward Forward with Low Memory Matrix Adaptation Exploiting Gradient Information

• Hot Genetic Wheels

• Learning driving behaviors through Evolutionary Algorithms

Student projects 2021-2022

A collection of some of the student projects that were presented as part of the exam:

• Automatically solving games with AI using Neuro-evolution

• Autonomously playing Yellow-Spaceship using NEAT and GP

• Cryptocurrencies Automatic Trading Using Evolutionary Computation

• DarkrAI - a Pareto ε-greedy policy Improving Pokémon AI Training With NSGA-II

• Dual-Objective Scheduling of Rescue Vehicles to Distinguish Forest Fires via Differential Evolution and Particle Swarm Optimization Combined Algorithm

• EA for Hyperparameter Optimization of Graph Neural Networks - a comparative study

• Evolution of Cooperative Behaviour - A Computational Perspective

• F1 Race Strategy

• Genetic Strategies or Reinforcement Learning?

• HyperNEAT Evolution for Foraging Tasks

• Interactive Evolution for Melody Generation

• N-Steps Ahead Forecasting of Financial Time Series using Genetic Programming

• Neuro-Evolved Cellular Automata

• PS-PSO - Paul Signac Particle Swarm Optimization

• Spiralis Fractalis - the Evolution of Generative Art

• Trading Assets with Genetic Programming and Advantage Actor Critic - A comparison

• Using Genetic Algorithms for Predicting the Parameters of a COVID-19 SEIR Model

Student projects 2020-2021

A collection of some of the student projects that were presented as part of the exam:

• 3D Model Cutting via Evolutionary Algorithms with Blender

• Autonomous driving using neural networks trained by evolutionary algorithms

• Evolutionary Image Vectorization

• Exploring NEAT’s capabilities at evolving agents

• Genetic Algorithms for Evolving Convolutional Neural Networks for Music Genre Classification

• Multi Objective Optimization in Graph Influence Maximization

• Multi-objective approach to sustainability in fashion

• Neural Architecture Search with genetic algorithms and DARTS

• Solving the School Timetabling Problem with Genetic Algorithm

Student projects 2019-2020

A collection of some of the student projects that were presented as part of the exam:

• 3D Printing Unit Cost Optimization

• Comparing LSTM, NEAT and GP for solar radiation forecasting

• DeepSwarm Analysis

• EC::GoL

• Gene-rate

• Genetic K-Means Algorithm for the Land Cover Classification

• Multi Objective optimization wind turbines placement

• Population optimization of prokaryotic predator-prey agents simulation

• Soft Robots Evolution

• Solving a Job-Shop Scheduling Problem with Genetic Algorithm

• Solving benchmarks with AI

Student projects 2018-2019

A collection of some of the student projects that were presented as part of the exam:

• An Analysis of DeepSwarm Performances on MNIST Dataset

• Events Relevance Optimization

• Evolving CGP programs to play Atari games

• Exploring a resource map with a PSO algorithm

• Fooling Convolutional Neural Network by using Evolutionary Algorithms

• Generating Art Following the Approach of Karl Sims

• NN auto-encoders - backpropagation and bio-inspired algorithms in a limited parameters context

• PCB design optimization

• Solving games with AI

• The Container Loading Problem Solved by a Genetic Algorithm

• Traffic Junction Tuner - an evolutionary traffic junction optimizer