Improving neural network training solutions using regularisation

This paper describes the application of regularisation to the training of feedforward neural networks, as a means of improving the quality of solutions obtained. The basic principles of regularisation theory are outlined for both linear and nonlinear training and then extended to cover a new hybrid training algorithm for feedforward neural networks recently proposed by the authors. The concept of functional regularisation is also introduced and discussed in relation to MLP and RBF networks. The tendency for the hybrid training algorithm and many linear optimisation strategies to generate large magnitude weight solutions when applied to ill-conditioned neural paradigms is illustrated graphically and reasoned analytically. While such weight solutions do not generally result in poor fits, it is argued that they could be subject to numerical instability and are therefore undesirable. Using an illustrative example it is shown that, as well as being beneficial from a generalisation perspective, regularisation also provides a means for controlling the magnitude of solutions. (C) 2001 Elsevier Science B.V. All rights reserved.

Training of Reaching in Stroke Survivors With Severe and Chronic Upper Limb Paresis Using a Novel Nonrobotic Device: A Randomized Clinical Trial

Background and Purpose—Severe upper limb paresis is a major contributor to disability after stroke. This study investigated the efficacy of a new nonrobotic training device, the Sensorimotor Active Rehabilitation Training (SMART) Arm, that was used with or without electromyography-triggered electrical stimulation of triceps brachii to augment elbow extension, permitting stroke survivors with severe paresis to practice a constrained reaching task.

Methods—A single-blind, randomized clinical trial was conducted with 42 stroke survivors with severe and chronic paresis. Thirty-three participants completed the study, of whom 10 received training using the SMART Arm with electromyography-triggered electrical stimulation, 13 received training using the SMART Arm alone, and 10 received no intervention (control). Training consisted of 12 1-hour sessions over 4 weeks. The primary outcome measure was “upper arm function,” item 6 of the Motor Assessment Scale. Secondary outcome measures included impairment measures; triceps muscle strength, reaching force, modified Ashworth scale; and activity measures: reaching distance and Motor Assessment Scale. Assessments were administered before (0 weeks) and after training (4 weeks) and at 2 months follow-up (12 weeks).

Results—Both SMART Arm groups demonstrated significant improvements in all impairment and activity measures after training and at follow-up. There was no significant difference between these 2 groups. There was no change in the control group.

Conclusions—Our findings indicate that training of reaching using the SMART Arm can reduce impairment and improve activity in stroke survivors with severe and chronic upper limb paresis, highlighting the benefits of intensive task-oriented practice, even in the context of severe paresis.

Age-related differences in rapid muscle activation after rate of force development training of the elbow flexors

In young adults, improvements in the rate of force development as a result of resistance training are accompanied by increases in neural drive in the very initial phase of muscle activation. The purpose of this experiment was to determine if older adults also exhibit similar adaptations in response to rate of force development (RFD) training. Eight young (21-35 years) and eight older (60-79 years) adults were assessed during the production of maximum rapid contractions, before and after four weeks of progressive resistance training for the elbow flexors. Young and older adults exhibited significant increases (P<0.01) in peak RFD, of 25.6% and 28.6% respectively. For both groups the increase in RFD was accompanied by an increase in the root mean square (RMS) amplitude and in the rate of rise (RER) in the electromyogram (EMG) throughout the initial 100 ms of activation. For older adults, however, this training response was only apparent in the brachialis and brachioradialis muscles. This response was not observed in surface EMG recorded from the biceps brachii muscle during either RFD testing or throughout training, nor was it observed in the pronator teres muscle. The minimal adaptations observed for older adults in the bifunctional muscles biceps brachii and pronator teres are considered to indicate a compromise of the neural adaptations older adults might experience in response to resistance training.

Transfer of resistance training to enhance rapid coordinated force production by older adults

The purpose of this study was to examine the capacity of resistance training to enhance the rapid and coordinated production of force by older people. Thirty adults (greater than or equal to 60 years) completed a visually guided aiming task that required the generation of isometric torque in 2 df about the elbow prior to and following a 4-week training period. Groups of six participants were allocated to two progressive ( 40 - 100% maximal voluntary contraction (MVC)) resistance-training (PRT) groups, to two constant low-load (10% MVC) training groups (CLO) and to one no-training control group. Training movements required the generation of either combined flexion and supination (FLESUP), or combined extension and supination (EXTSUP). In response to training, target acquisition times in the aiming task decreased for all groups; however, both the nature of the training load and the training movement influenced the pattern and magnitude of improvements (EXTSUP_ CLO: 36%, FLESUP_ PRT 26%, EXTSUP_ PRT 22%, FLESUP_ CLO 20%, CONTROL 15%). For one group that trained with progressively increasing loads, there arose a subsequent decrease in performance in one condition of the transfer task. For each group, these adaptations were accompanied by systematic changes in the coordination of muscles about the elbow joint, particularly the biceps brachii.

The consequences of resistance training for movement control in older adults

Older adults who undertake resistance training are typically seeking to maintain or increase their muscular strength with the goal of preserving or improving their functional capabilities. The extent to which resistance training adaptations lead to improved performance on tasks of everyday living is not particularly well understood. Indeed, studies examining changes in functional task performance experienced by older adults following periods of resistance training have produced equivocal findings. A clear understanding of the principles governing the transfer of resistance training adaptations is therefore critical in seeking to optimize the prescription of training regimes that have as their aim the maintenance and improvement of functional movement capacities in older adults. The degenerative processes that occur in the aging motor system are likely to influence heavily any adaptations to resistance training and the subsequent transfer to functional task performance. The resulting characteristics of motor behavior, such as the substantial decline in the rate of force development and the decreased steadiness of force production, may entail that specialized resistance training strategies are necessary to maximize the benefits for older adults. In this review, we summarize the alterations in the neuromuscular system that are responsible for the declines in strength, power, and force control, and the subsequent deterioration in the everyday movement capabilities of older adults. We examine the literature concerning the neural adaptations that older adults experience in response to resistance training, and consider the readiness with which these adaptations will improve the functional movement capabilities of older adults.

The sites of neural adaptation induced by resistance training in humans

Although it has long been supposed that resistance training causes adaptive changes in the CNS, the sites and nature of these adaptations have not previously been identified. In order to determine whether the neural adaptations to resistance training occur to a greater extent at cortical or subcortical sites in the CNS, we compared the effects of resistance training on the electromyographic (EMG) responses to transcranial magnetic (TMS) and electrical (TES) stimulation. Motor evoked potentials (MEPs) were recorded from the first dorsal interosseous muscle of 16 individuals before and after 4 weeks of resistance training for the index finger abductors (n=8), or training involving finger abduction-adduction without external resistance (n=8). TMS was delivered at rest at intensities from 5% below the passive threshold to the maximal output of the stimulator. TMS and TES were also delivered at the active threshold intensity while the participants exerted torques ranging from 5 to 60% of their maximum voluntary contraction (MVC) torque. The average latency of MEPs elicited by TES was significantly shorter than that of TMS MEPs (TES latency=21.5+/-1.4 ms; TMS latency=23.4+/-1.4 ms; P

Neural adaptations to resistance training - Implications for movement control

It has long been believed that resistance training is accompanied by changes within the nervous system that play an important role in the development of strength. Many elements of the nervous system exhibit the potential for adaptation in response to resistance training, including supraspinal centres, descending neural tracts, spinal circuitry and the motor end plate connections between motoneurons and muscle fibres. Yet the specific sites of adaptation along the neuraxis have seldom been identified experimentally, and much of the evidence for neural adaptations following resistance training remains indirect. As a consequence of this current lack of knowledge, there exists uncertainty regarding the manner in which resistance training impacts upon the control and execution of functional movements. We aim to demonstrate that resistance training is likely to cause adaptations to many neural elements that are involved in the control of movement, and is therefore likely to affect movement execution during a wide range of tasks.

Resistance training enhances the stability of sensorimotor coordination

Strategics for the control of human movement are constrained by the neuroanatomical characteristics of the motor system. In particular, there is evidence that the capacity of muscles for producing force has a strong influence on the stability of coordination in certain movement tasks. In the present experiment, our aim was to determine whether physiological adaptations that cause relatively long-lasting changes in the ability of muscles to produce force can influence the stability of coordination in a systematic manner. We assessed the effects of resistance training on the performance of a difficult coordination task that required participants to synchronize or syncopate movements of their index finger with an auditory metronome. Our results revealed that training that increased isometric finger strength also enhanced the stability of movement coordination. These changes were accompanied by alterations in muscle recruitment patterns. In Particular, the trained muscles were recruited in a more consistent fashion following the programme of resistance training. These results indicate that resistance training produces functional adaptations of the neuroanatomical constraints that underlie the control of voluntary movement.

Manual asymmetries in the preparation and control of goal-directed movements

The primary purpose of this experiment was to determine if left hand reaction time advantages in manual aiming result from a right hemisphere attentional advantage or an early right hemisphere role in movement preparation. Right-handed participants were required to either make rapid goal-directed movements to small targets or simply lift their hand upon target illumination. The amount of advance information about the target for a particular trial was manipulated by precuing a subset of potential targets prior to the reaction time interval. When participants were required to make aiming movements to targets in left space, the left hand enjoyed a reaction advantage that was not present for aiming in right space: or simple finger lifts. This advantage was independent of the amount or type of advance information provided by the precue. This finding supports the movement planning hypothesis. With respect to movement execution, participants completed their aiming movements more quickly when aiming with their right hand, particularly in right space. This right hand advantage in right space was due to the time required to decelerate the movement and to make feedback-based adjustments late in the movement trajectory. (C) 2001 Academic Press.