S-SETA: Selective Software-Only Error-Detection Technique Using Assertions

Software-based techniques offer several advantages to increase the reliability of processor-based systems at very low cost, but they cause performance degradation and an increase of the code size. To meet constraints in performance and memory, we propose SETA, a new control-flow software-only technique that uses assertions to detect errors affecting the program flow. SETA is an independent technique, but it was conceived to work together with previously proposed data-flow techniques that aim at reducing performance and memory overheads. Thus, SETA is combined with such data-flow techniques and submitted to a fault injection campaign. Simulation and neutron induced SEE tests show high fault coverage at performance and memory overheads inferior to the state-of-the-art.

Countermanding in rats as a practical model for investigation of adaptive control of behaviour, lifespan changes in behavioural control and neurotransmitter function

The countermanding paradigm was designed to investigate the ability to cancel a prepotent response when a stop signal is presented and allows estimation of the stop signal response time (SSRT), an otherwise unobservable behaviour. Humans exhibit adaptive control of behaviour in the countermanding task, proactively lengthening response time (RT) in expectation of stopping and reactively lengthening RT following stop trials or errors. Human performance changes throughout the lifespan, with longer RT, SSRT and greater emphasis on post-error slowing reported for older compared to younger adults. Inhibition in the task has generally been improved by drugs that increase extracellular norepinephrine. The current thesis examined a novel choice response countermanding task in rats to explore whether rodent countermanding performance is a suitable model for the study of adaptive control of behaviour, lifespan changes in behavioural control and the role of neurotransmitters in these behaviours. Rats reactively adjusted RT in the countermanding task, shortening RT after consecutive correct go trials and lengthening RT following non-cancelled, but not cancelled stop trials, in sessions with a 10 s, but not a 1 s post-error timeout interval. Rats proactively lengthened RT in countermanding task sessions compared to go trial-only sessions. Together, these findings suggest that rats strategically lengthened RT in the countermanding task to improve accuracy and avoid longer, unrewarded timeout intervals. Next, rats exhibited longer RT and relatively conserved post-error slowing, but no significant change in SSRT when tested at 12, compared to 7 months of age, suggesting that rats exhibit changes in countermanding task performance with aging similar to those observed in humans. Finally, acute administration of yohimbine (1.25, 2.5 mg/kg) and d-amphetamine (0.25, 0.5 mg/kg), which putatively increase extracellular norepinephrine and dopamine respectively, resulted in RT shortening, baseline-dependent effects on SSRT, and attenuated adaptive RT adjustments in rats in the case of d-amphetamine. These findings suggest that dopamine and norepinephrine encouraged motivated, reward-seeking behaviour and supported inhibitory control in an inverted-U-like fashion. Taken together, these observations validate the rat countermanding task for further study of the neural correlates and neurotransmitters mediating adaptive control of behaviour and lifespan changes in behavioural control.

Revisiting the synthetic control estimator

The synthetic control (SC) method has been recently proposed as an alternative to estimate treatment effects in comparative case studies. The SC relies on the assumption that there is a weighted average of the control units that reconstruct the potential outcome of the treated unit in the absence of treatment. If these weights were known, then one could estimate the counterfactual for the treated unit using this weighted average. With these weights, the SC would provide an unbiased estimator for the treatment effect even if selection into treatment is correlated with the unobserved heterogeneity. In this paper, we revisit the SC method in a linear factor model where the SC weights are considered nuisance parameters that are estimated to construct the SC estimator. We show that, when the number of control units is fixed, the estimated SC weights will generally not converge to the weights that reconstruct the factor loadings of the treated unit, even when the number of pre-intervention periods goes to infinity. As a consequence, the SC estimator will be asymptotically biased if treatment assignment is correlated with the unobserved heterogeneity. The asymptotic bias only vanishes when the variance of the idiosyncratic error goes to zero. We suggest a slight modification in the SC method that guarantees that the SC estimator is asymptotically unbiased and has a lower asymptotic variance than the difference-in-differences (DID) estimator when the DID identification assumption is satisfied. If the DID assumption is not satisfied, then both estimators would be asymptotically biased, and it would not be possible to rank them in terms of their asymptotic bias.