Accelerating the convergence of the back-propagation method

By T. Vogl, J. K. Mangis, J. K. Rigler, W. T. Zink, D. Alkon. 1988

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Deep learning in neural networks: An overview (2014, background): “Some algorithms for controlling BP step size adapt a global learning rate (Lapedes and Farber, 1986; Vogl et al., 1988; Battiti, 1989; LeCun et al., 1993; Yu et al., 1995), others compute individual learning rates for each weight (Jacobs, 1988; Silva and Almeida, 1990).”
Neural Networks: Tricks of the Trade (2012, methodology): “1 Called “stabilizers” by Tikhonov [21].”
Sentic patterns: Dependency-based rules for concept-level sentiment analysis (2014, unknown): “The ELM approach was introduced to overcome some issues in back-propagation network [33] training, specifically, potentially slow convergence rates, critical tuning of optimization parameters [34], and the presence of local minima that call for multi-start and re-training strategies [35].”
Reinforcement Learning for Primary care e Appointment Scheduling (2017, background): “• Learning rate – The learning rate considered by optimizer RMSProp [74]; • Standard deviations for value and policy network initializations: • Gamma – Controls the weight of reward discounting; • Experience buffer size – The size of the batch used to train the RL agent;”
Empirical evaluation of the improved Rprop learning algorithms (2003, methodology): “update for all weights is reversed if E(t) ?� E (t−1) (� =1 :0 in [22]; 1:0 i� 6 1:05 in [ 25 ]), with the ideas in [18,24].”
End-to-End Speech Recognition: A Survey (2023, background): “Many approaches exist for learning rate control [161], [162]: NewBob [163], [164] and enhancements [162]; global versus parameter-wise learning rate control (exponential decay, power decay, etc.) [165]; learning rate warm-up [44]; warm restarts/cosine annealing [166]; weight decay versus gradually…”
Let a biogeography-based optimizer train your Multi-Layer Perceptron (2014, background): “Back Propagation (BP) is a gradient-based algorithm that has some drawbacks such as slow convergence [13,64,49] and a tendency to get trapped in local minima [17,29,32], making it unreliable for practical applications.”