#### DMCA

## Evolutionary ensemble for stock prediction (2004)

Venue: | Proceedings of the Genetic and Evolutionary Computation Conference |

Citations: | 2 - 0 self |

### Citations

2968 | Some methods for classification and analysis of multivariate observations
- MacQueen
- 1967
(Show Context)
Citation Context ...s select a subset of the population as ensemble. It consists of some best individuals or representative ones from the whole population. In the latter, a clustering algorithm such as k-means algorithm =-=[22]-=- is used and a representative solution for each cluster is selected. In this paper, we devised an instance-based ensemble which is different from traditional ensembles. Traditional ensemble models do ... |

2017 | Finding structure in time
- Elman
- 1990
(Show Context)
Citation Context ...he final property ratio P as follows: P = CN + SN . C1 + S1 3 Evolutionary Ensemble 3.1 Artificial Neural Networks We use a recurrent neural network architecture which is a variant of Elman’s network =-=[6]-=-. It consists of input, hidden, and output layers as shown in Figure 3.Evolutionary Ensemble for Stock Prediction 1105 if ( signal is SELL ) { Ct+1 ← Ct + min(B,St) × (1 − T ) St+1 ← St − min(B,St) }... |

969 | Adaptive mixtures of local experts
- Jacobs, Jordan, et al.
- 1991
(Show Context)
Citation Context ...ations. 3.3 Instance-Based Ensemble Model An ensemble learning is to aggregate multiple subsystems to solve a complex problem. A number of approaches have been developed for ensemble learning [5] [8] =-=[12]-=- [23] [31]. The method is based on the fact that a solution with the smallest training error does not necessarily guarantee the most generalized one. It is usual to select the best individual as the f... |

667 |
Neural networks ensembles
- Hansen, Salamon
- 1990
(Show Context)
Citation Context ...enerations. 3.3 Instance-Based Ensemble Model An ensemble learning is to aggregate multiple subsystems to solve a complex problem. A number of approaches have been developed for ensemble learning [5] =-=[8]-=- [12] [23] [31]. The method is based on the fact that a solution with the smallest training error does not necessarily guarantee the most generalized one. It is usual to select the best individual as ... |

172 | On the scalability of parallel genetic algorithms
- Cantu-Paz, Goldberg
- 1999
(Show Context)
Citation Context ...n Section 2. Only one node exists in the output layer for x(t+1)−x(t) x(t) . 3.2 Parallel Genetic Algorithm We use a parallel GA to optimize the weights. It is a global single-population master-slave =-=[2]-=- and the structure is shown in Figure 4. In this neuro-genetic hybrid approach, the fitness evaluation is dominant in running time. To evaluate an offspring (a network) the backpropagation-based algor... |

117 |
Towards the genetic synthesis of neural networks
- Harp, Samad
(Show Context)
Citation Context ... where n, p, and q are the numbers of input, hidden, output units, respectively. In this work, the matrix size is 20×(75+20+1). We should note that most GAs for ANN optimization used linear encodings =-=[9]-=- [21]. We take the 2D encoding suggested in [17]. Figure 5 shows an example neural network and the corresponding chromosome. – Selection, crossover, and mutation: Roulette-wheel selection is used for ... |

97 |
Learning machines: Foundations of trainable pattern-classifying systems
- Nilsson
- 1965
(Show Context)
Citation Context ...s. 3.3 Instance-Based Ensemble Model An ensemble learning is to aggregate multiple subsystems to solve a complex problem. A number of approaches have been developed for ensemble learning [5] [8] [12] =-=[23]-=- [31]. The method is based on the fact that a solution with the smallest training error does not necessarily guarantee the most generalized one. It is usual to select the best individual as the final ... |

87 | Making use of population information in evolutionary artificial neural networks
- Yao, Liu
- 1998
(Show Context)
Citation Context ...3 Instance-Based Ensemble Model An ensemble learning is to aggregate multiple subsystems to solve a complex problem. A number of approaches have been developed for ensemble learning [5] [8] [12] [23] =-=[31]-=-. The method is based on the fact that a solution with the smallest training error does not necessarily guarantee the most generalized one. It is usual to select the best individual as the final solut... |

84 |
Boosting and other ensemble methods
- Drucker, Cortes, et al.
- 1994
(Show Context)
Citation Context ...of generations. 3.3 Instance-Based Ensemble Model An ensemble learning is to aggregate multiple subsystems to solve a complex problem. A number of approaches have been developed for ensemble learning =-=[5]-=- [8] [12] [23] [31]. The method is based on the fact that a solution with the smallest training error does not necessarily guarantee the most generalized one. It is usual to select the best individual... |

47 |
Genetic algorithms approach to feature discretization in artificial neural networks for the prediction of stock price index," Expert Systems with Applications
- Kyong-jae, Han
(Show Context)
Citation Context ...not only the buy-and-hold strategy but also other traditional ensemble approaches. 1 Introduction Stock prediction is a historically hot topic. There were a variety of studies on this topic [10] [14] =-=[18]-=- [19] [24] [26] [28] [29]. Early studies were mostly about deterministic measures to help the prediction [1] [11] [16] [20]. Since about the early nineties many approaches based on stochastic or heuri... |

43 |
New Trading Systems and Methods
- Kaufman
- 2005
(Show Context)
Citation Context ...m for the input variables as they are. We utilize a number of technical indicators being used by financial experts such as moving average, golden-cross, dead-cross, relative strength index, and so on =-=[15]-=-. We describe some of them which were not considered in [19] in the following: – Rate of change ( ROC ) • A ratio of price difference between the current price and the price a period of time ago. • RO... |

22 | Toward More Powerful Recombinations
- Kahng, Moon
- 1995
(Show Context)
Citation Context ...mple neural network and the corresponding chromosome. – Selection, crossover, and mutation: Roulette-wheel selection is used for parent selection. The offspring is produced by geographic 2D crossover =-=[13]-=-. It is known to create diverse new schemata and reflect well the geographical relationships among genes. It chooses a number of lines, divides the chromosomal domain into two equivalent classes, and ... |

15 | Evolving Artificial Neural Networks to Combine Financial Forecasts
- Harrald, Kamstra
- 1997
(Show Context)
Citation Context ...rage over not only the buy-and-hold strategy but also other traditional ensemble approaches. 1 Introduction Stock prediction is a historically hot topic. There were a variety of studies on this topic =-=[10]-=- [14] [18] [19] [24] [26] [28] [29]. Early studies were mostly about deterministic measures to help the prediction [1] [11] [16] [20]. Since about the early nineties many approaches based on stochasti... |

13 |
Data mining approach to policy analysis in a health insurance domain
- Chae, Ho, et al.
- 2001
(Show Context)
Citation Context ...iction [1] [11] [16] [20]. Since about the early nineties many approaches based on stochastic or heuristic models have been proposed. They include artificial neural networks [26] [29], decision trees =-=[4]-=-, rule induction [7], Bayesian belief networks [30], evolutionary algorithms [14] [18], classifier systems [25], fuzzy sets [3] [28], and association rules [27]. Hybrid models combining a few approach... |

12 | Financial Volatility Trading Using Recurrent Neural Networks
- Tino, Schittenkopf, et al.
- 2001
(Show Context)
Citation Context ...y-and-hold strategy but also other traditional ensemble approaches. 1 Introduction Stock prediction is a historically hot topic. There were a variety of studies on this topic [10] [14] [18] [19] [24] =-=[26]-=- [28] [29]. Early studies were mostly about deterministic measures to help the prediction [1] [11] [16] [20]. Since about the early nineties many approaches based on stochastic or heuristic models hav... |

11 |
Controlling Chaos by GA-Based Reinforcement Learning Neural Network
- Lin, Jou
(Show Context)
Citation Context ...re n, p, and q are the numbers of input, hidden, output units, respectively. In this work, the matrix size is 20×(75+20+1). We should note that most GAs for ANN optimization used linear encodings [9] =-=[21]-=-. We take the 2D encoding suggested in [17]. Figure 5 shows an example neural network and the corresponding chromosome. – Selection, crossover, and mutation: Roulette-wheel selection is used for paren... |

9 | Financial prediction and trading strategies using neurofuzzy approaches
- Pantazopoulos, Tsoukalas, et al.
- 1998
(Show Context)
Citation Context ...he buy-and-hold strategy but also other traditional ensemble approaches. 1 Introduction Stock prediction is a historically hot topic. There were a variety of studies on this topic [10] [14] [18] [19] =-=[24]-=- [26] [28] [29]. Early studies were mostly about deterministic measures to help the prediction [1] [11] [16] [20]. Since about the early nineties many approaches based on stochastic or heuristic model... |

9 |
Explorations in LCS Models of Stock Trading
- Schulenburg
(Show Context)
Citation Context ...els have been proposed. They include artificial neural networks [26] [29], decision trees [4], rule induction [7], Bayesian belief networks [30], evolutionary algorithms [14] [18], classifier systems =-=[25]-=-, fuzzy sets [3] [28], and association rules [27]. Hybrid models combining a few approaches are also popular [10] [24]. Kwon and Moon [19] proposed neuro-genetic hybrids for the stock prediction and s... |

6 |
Simulation and forecasting complex financial time series using neural networks and fuzzy logic
- Castillo, Melin
- 2001
(Show Context)
Citation Context ...posed. They include artificial neural networks [26] [29], decision trees [4], rule induction [7], Bayesian belief networks [30], evolutionary algorithms [14] [18], classifier systems [25], fuzzy sets =-=[3]-=- [28], and association rules [27]. Hybrid models combining a few approaches are also popular [10] [24]. Kwon and Moon [19] proposed neuro-genetic hybrids for the stock prediction and showed notable su... |

6 |
Neuron reordering for better neuro-genetic hybrids
- Kim, Moon
- 2002
(Show Context)
Citation Context ...dden, output units, respectively. In this work, the matrix size is 20×(75+20+1). We should note that most GAs for ANN optimization used linear encodings [9] [21]. We take the 2D encoding suggested in =-=[17]-=-. Figure 5 shows an example neural network and the corresponding chromosome. – Selection, crossover, and mutation: Roulette-wheel selection is used for parent selection. The offspring is produced by g... |

6 |
Daily stock prediction using neuro-genetic hybrids
- Kwon, Moon
- 2003
(Show Context)
Citation Context ...nly the buy-and-hold strategy but also other traditional ensemble approaches. 1 Introduction Stock prediction is a historically hot topic. There were a variety of studies on this topic [10] [14] [18] =-=[19]-=- [24] [26] [28] [29]. Early studies were mostly about deterministic measures to help the prediction [1] [11] [16] [20]. Since about the early nineties many approaches based on stochastic or heuristic ... |

6 |
Residential property price time series forecasting with neural networks
- Wilson, Paris, et al.
- 2002
(Show Context)
Citation Context ... strategy but also other traditional ensemble approaches. 1 Introduction Stock prediction is a historically hot topic. There were a variety of studies on this topic [10] [14] [18] [19] [24] [26] [28] =-=[29]-=-. Early studies were mostly about deterministic measures to help the prediction [1] [11] [16] [20]. Since about the early nineties many approaches based on stochastic or heuristic models have been pro... |

4 |
Genetic programming prediction of stock prices
- Kanoudan
(Show Context)
Citation Context ...over not only the buy-and-hold strategy but also other traditional ensemble approaches. 1 Introduction Stock prediction is a historically hot topic. There were a variety of studies on this topic [10] =-=[14]-=- [18] [19] [24] [26] [28] [29]. Early studies were mostly about deterministic measures to help the prediction [1] [11] [16] [20]. Since about the early nineties many approaches based on stochastic or ... |

3 |
Turning point identification and Bayesian forecasting of a volatile time series
- Wolfe
- 1988
(Show Context)
Citation Context ...ineties many approaches based on stochastic or heuristic models have been proposed. They include artificial neural networks [26] [29], decision trees [4], rule induction [7], Bayesian belief networks =-=[30]-=-, evolutionary algorithms [14] [18], classifier systems [25], fuzzy sets [3] [28], and association rules [27]. Hybrid models combining a few approaches are also popular [10] [24]. Kwon and Moon [19] p... |

2 |
Using inductive learning to predict bankruptcy
- Gentry, Shaw, et al.
(Show Context)
Citation Context ... [20]. Since about the early nineties many approaches based on stochastic or heuristic models have been proposed. They include artificial neural networks [26] [29], decision trees [4], rule induction =-=[7]-=-, Bayesian belief networks [30], evolutionary algorithms [14] [18], classifier systems [25], fuzzy sets [3] [28], and association rules [27]. Hybrid models combining a few approaches are also popular ... |

2 |
The use of an association rules matrix for economic modelling
- Veliev, Rubinov, et al.
- 1999
(Show Context)
Citation Context ...eural networks [26] [29], decision trees [4], rule induction [7], Bayesian belief networks [30], evolutionary algorithms [14] [18], classifier systems [25], fuzzy sets [3] [28], and association rules =-=[27]-=-. Hybrid models combining a few approaches are also popular [10] [24]. Kwon and Moon [19] proposed neuro-genetic hybrids for the stock prediction and showed notable success. However, since they did no... |

1 |
Computerized Trading Techniques
- Hochheimer
(Show Context)
Citation Context ...n is a historically hot topic. There were a variety of studies on this topic [10] [14] [18] [19] [24] [26] [28] [29]. Early studies were mostly about deterministic measures to help the prediction [1] =-=[11]-=- [16] [20]. Since about the early nineties many approaches based on stochastic or heuristic models have been proposed. They include artificial neural networks [26] [29], decision trees [4], rule induc... |

1 |
Trading with ARIMA forecasts
- Kepka
- 1985
(Show Context)
Citation Context ...a historically hot topic. There were a variety of studies on this topic [10] [14] [18] [19] [24] [26] [28] [29]. Early studies were mostly about deterministic measures to help the prediction [1] [11] =-=[16]-=- [20]. Since about the early nineties many approaches based on stochastic or heuristic models have been proposed. They include artificial neural networks [26] [29], decision trees [4], rule induction ... |

1 | The end point moving average - Lafferty - 1995 |

1 |
Predicting stock price using fuzzy gray prediction system. Expert Systems with Applications
- Wang
(Show Context)
Citation Context ...-hold strategy but also other traditional ensemble approaches. 1 Introduction Stock prediction is a historically hot topic. There were a variety of studies on this topic [10] [14] [18] [19] [24] [26] =-=[28]-=- [29]. Early studies were mostly about deterministic measures to help the prediction [1] [11] [16] [20]. Since about the early nineties many approaches based on stochastic or heuristic models have bee... |