Globally optimal learning rates in multilayer neural networks

A method for calculating the globally optimal learning rate in on-line gradient-descent training of multilayer neural networks is presented. The method is based on a variational approach which maximizes the decrease in generalization error over a given time frame. We demonstrate the method by computing optimal learning rates in typical learning scenarios. The method can also be employed when different learning rates are allowed for different parameter vectors as well as to determine the relevance of related training algorithms based on modifications to the basic gradient descent rule.

Cognitive behaviour therapy for mothers of children with food allergy:a case series

Background: Food allergy affects quality of life in patients and parents and mothers report high levels of anxiety and stress. Cognitive Behaviour Therapy (CBT) may be helpful in reducing the psychological impact of food allergy. The aim of this study was to examine the appropriateness and effectiveness of CBT to improve psychological outcomes in parents of children with food allergy. Methods: Five parents (all mothers) from a local allergy clinic requested to have CBT; six mothers acted as controls and completed questionnaires only. CBT was individual and face-to face and lasted 12 weeks. All participants completed measures of anxiety and depression, worry, stress, general mental health, generic and food allergy specific quality of life at baseline and at 12 weeks. Results: Anxiety, depression and worry in the CBT group significantly reduced and overall mental health and QoL significantly improved from baseline to 12 weeks (all p < 0.05) in mothers in the CBT group; control group scores remained stable. Conclusions: CBT appears to be appropriate and effective in mothers of children with food allergy and a larger randomised control trial now needs to be conducted. Ways in which aspects of CBT can be incorporated into allergy clinic visits need investigation.