|
| 1 | +export const theoryData = { |
| 2 | + populationSize: { |
| 3 | +title: "Population Size", |
| 4 | +explanation: "The number of individuals in each generation. A larger population increases genetic diversity and the chance of finding a global optimum but requires more computational power.", |
| 5 | + }, |
| 6 | + generations: { |
| 7 | +title: "Generations", |
| 8 | +explanation: "The total number of iterations the evolutionary process will run. More generations allow the population to converge on better solutions but increase execution time.", |
| 9 | + }, |
| 10 | + cxpb: { |
| 11 | +title: "Crossover Probability", |
| 12 | +explanation: "The probability that two parent individuals will exchange genetic material. High values promote the recombination of good traits from different individuals.", |
| 13 | + }, |
| 14 | + mutpb: { |
| 15 | +title: "Mutation Probability", |
| 16 | +explanation: "The probability that an individual's genes will be randomly altered. This introduces new genetic material, helping the population escape local optima.", |
| 17 | + }, |
| 18 | + hof: { |
| 19 | +title: "Hall of Fame", |
| 20 | +explanation: "A special archive that stores the absolute best individuals found throughout the entire run, ensuring that the top solutions are never lost during evolution.", |
| 21 | + }, |
| 22 | + mu: { |
| 23 | +title: "Mu (μ)", |
| 24 | +explanation: "The number of individuals to be selected as parents for the next generation. It defines the size of the breeding pool.", |
| 25 | + }, |
| 26 | + lambda: { |
| 27 | +title: "Lambda (λ)", |
| 28 | +explanation: "The number of children to be generated in each step. This determines the exploratory capacity of the algorithm per generation.", |
| 29 | + }, |
| 30 | + tournamentSize: { |
| 31 | +title: "Tournament Size", |
| 32 | +explanation: "Used in selection: 'k' individuals are picked at random, and the best one wins. A larger tournament size increases selection pressure.", |
| 33 | + }, |
| 34 | + phi1: { |
| 35 | +title: "Cognitive Coefficient (φ1)", |
| 36 | +explanation: "In PSO, this scales the influence of a particle's personal best performance on its current velocity.", |
| 37 | + }, |
| 38 | + phi2: { |
| 39 | +title: "Social Coefficient (φ2)", |
| 40 | +explanation: "In PSO, this scales the influence of the swarm's global best performance on a particle's current velocity.", |
| 41 | + }, |
| 42 | + dimensions: { |
| 43 | + title: "Dimensions", |
| 44 | +explanation: "The number of variables or parameters that the algorithm is trying to optimize simultaneously.", |
| 45 | + }, |
| 46 | + algorithmStrategy: { |
| 47 | +title: "Algorithm Strategy", |
| 48 | +explanation: "The high-level logic governing evolution. 'eaSimple' is a basic generational model, while 'eaMuPlusLambda' and 'eaMuCommaLambda' use more advanced selection and replacement mechanics common in Evolution Strategies.", |
| 49 | + }, |
| 50 | + psoStrategy: { |
| 51 | +title: "PSO Strategy", |
| 52 | +explanation: "Defines the velocity and position update logic for particles. 'original' uses the standard PSO formulas, while variants like 'multiswarm' or 'speciation' help the algorithm handle more complex optimization landscapes.", |
| 53 | + }, |
| 54 | + benchmarkFunction: { |
| 55 | +title: "Benchmark Function", |
| 56 | +explanation: "Mathematical functions used to evaluate optimization performance. Each function (like Ackley, Rastrigin, or Rosenbrock) represents a unique 'landscape' with different challenges for the algorithm to solve.", |
| 57 | + }, |
| 58 | + weights: { |
| 59 | + title: "Weights", |
| 60 | + explanation: "Coefficients that determine the importance of different objectives in the fitness function. If you have multiple goals, weights help the algorithm prioritize between them (e.g., maximizing accuracy vs. minimizing complexity).", |
| 61 | + }, |
| 62 | + matingFunction: { |
| 63 | + title: "Mating (Crossover)", |
| 64 | + explanation: "The mechanism of biological recombination. It combines genetic information from two parents to generate offspring, aiming to produce better solutions by merging high-performing traits.", |
| 65 | + }, |
| 66 | + mutationFunction: { |
| 67 | + title: "Mutation", |
| 68 | + explanation: "Introduces random changes to individuals. This prevents premature convergence and ensures the algorithm explores new areas of the search space, maintaining genetic diversity.", |
| 69 | + }, |
| 70 | + selectionFunction: { |
| 71 | + title: "Selection", |
| 72 | + explanation: "The 'survival of the fittest' phase. It determines which individuals from the current population will be kept as parents for the next generation based on their fitness scores.", |
| 73 | + }, |
| 74 | + datasetUrl: { |
| 75 | + title: "Dataset Source", |
| 76 | + explanation: "The raw data used for training and evaluation. In ML tuning, the algorithm uses this data to test how well different model parameters perform.", |
| 77 | + }, |
| 78 | + targetColumn: { |
| 79 | + title: "Target Variable", |
| 80 | + explanation: "The specific column in your dataset that you want the model to predict (the dependent variable).", |
| 81 | + }, |
| 82 | + mlEvalFunction: { |
| 83 | + title: "ML Evaluation", |
| 84 | + explanation: "The metric used to judge the performance of the machine learning model (e.g., Accuracy, F1-score, or MSE). The EA optimizes this value.", |
| 85 | + }, |
| 86 | + individualSize: { |
| 87 | + title: "Individual Size", |
| 88 | + explanation: "The length of the chromosome or the number of genes representing a solution. For example, in bit manipulation, it's the number of bits per individual.", |
| 89 | + }, |
| 90 | + minMaxBoundaries: { |
| 91 | + title: "Search Boundaries", |
| 92 | + explanation: "The constraints on the search space. Defines the minimum and maximum values that a gene (or particle position) can take.", |
| 93 | + }, |
| 94 | + gpPrimitiveSet: { |
| 95 | + title: "Primitive Set", |
| 96 | + explanation: "The building blocks for Genetic Programs. It includes functions (operators like +, -, *, /) and terminals (constants like 1, 2, or variables like 'x'). Crossing these creates complex tree structures.", |
| 97 | + }, |
| 98 | + treeGenerator: { |
| 99 | + title: "Tree Generator", |
| 100 | + explanation: "The algorithm used to create the initial programs. 'genFull' creates balanced trees where every branch reaches max depth, while 'genHalfAndHalf' provides a mix of short and deep trees for better diversity.", |
| 101 | + }, |
| 102 | + bloatLimits: { |
| 103 | + title: "Bloat Control", |
| 104 | + explanation: "Prevents programs from growing excessively large without improving fitness. By setting a height limit, we ensure the solutions remain computationally efficient and readable.", |
| 105 | + } |
| 106 | +}; |
0 commit comments