Visual question answering requires high-order reasoning about an image, which is a fundamental capability needed by machine systems to follow complex directives. Recently, modular networks have been shown to be an effective framework for performing visual reasoning tasks. While modular networks were initially designed with a degree of model transparency, their performance on complex visual reasoning benchmarks was lacking. Current stateof-the-art approaches do not provide an effective mechanism for understanding the reasoning process. In this paper, we close the performance gap between interpretable models and state-of-the-art visual reasoning methods. We propose a set of visual-reasoning primitives which, when composed, manifest as a model capable of performing complex reasoning tasks in an explicitly-interpretable manner. The fidelity and interpretability of the primitives' outputs enable an unparalleled ability to diagnose the strengths and weaknesses of the resulting model. Critically, we show that these primitives are highly performant, achieving state-of-the-art accuracy of 99.1% on the CLEVR dataset. We also show that our model is able to effectively learn generalized representations when provided a small amount of data containing novel object attributes. Using the CoGenT generalization task, we show more than a 20 percentage point improvement over the current state of the art.

Non-technical summary In mammalian skeletal muscle, the coupling between action potential activation and contraction is supposed to be ultimately mediated by the interaction of two ion channels, the L-type calcium channel (so-called dihydropyridine receptor channel) at the transverse tubular system, and the sarcoplasmic reticulum (SR) calcium release channel (so-called ryanodine receptor channel). This paper demonstrates that adult skeletal muscle fibres transfected in vivo with DNA plasmids are able to express functional transgenic dihydropyridine receptor channels. More importantly, the data suggest that transgenic dihydropyridine receptor channels replace native channels in their interaction with SR calcium release channels. Our findings open new avenues for structural and functional studies of the molecular interactions underlying excitation-contraction coupling within the physiologically relevant cellular context of adult mammalian skeletal muscle fibres. AbstractWe investigated the effects of the overexpression of two enhanced green fluorescent protein (EGFP)-tagged α1sDHPR variants on Ca 2+ currents (I Ca ), charge movements (Q) and SR Ca 2+ release of muscle fibres isolated from adult mice. Flexor digitorum brevis (FDB) muscles were transfected by in vivo electroporation with plasmids encoding for EGFP-α1sDHPR-wt and EGFP-α1sDHPR-T935Y (an isradipine-insensitive mutant). Two-photon laser scanning microscopy (TPLSM) was used to study the subcellular localization of transgenic proteins, while I Ca , Q and Ca 2+ release were studied electrophysiologically and optically under voltage-clamp conditions. TPLSM images demonstrated that most of the transgenic α1sDHPR was correctly targeted to the transverse tubular system (TTS). Immunoblotting analysis of crude extracts of transfected fibres demonstrated the synthesis of bona fide transgenic EGFP-α1sDHPR-wt in quantities comparable to that of native α1sDHPR. Though expression of both transgenic variants of the alpha subunit of the dihydropyridine receptor (α1sDHPR) resulted in ∼50% increase in Q, they surprisingly had no effect on the maximal Ca 2+ conductance (g Ca ) nor the SR Ca 2+ release. Nonetheless, fibres expressing EGFP-α1sDHPR-T935Y exhibited up to 70% isradipine-insensitive I Ca (I Ca -ins) with a right-shifted voltage dependence compared to that in control fibres. Interestingly, Q and SR Ca 2+ release also displayed right-shifted voltage dependence in fibres expressing EGFP-α1sDHPR-T935Y. In contrast, the midpoints of the voltage dependence of g Ca , Q and Ca 2+ release were not different from those in control fibres and in fibres expressing EGFP-α1sDHPR-wt. Overall, our results suggest that transgenic α1sDHPRs are correctly trafficked and inserted in the TTS membrane, and that a substantial fraction of them works as conductive Ca 2+ channels capable of physiologically controlling the release of Ca 2+ from the SR. A plausible corollary of this work is that the expression of transgenic variants of the α1sDHPR leads to the replacement of native channels...

Central Core Disease (CCD) and Malignant Hyperthermia (MH) are linked to single amino acid substitutions in the skeletal muscle Ca 2þ release channel, the type 1 ryanodine receptor (RyR1). We focus on two autosomal dominant (AD) 712a

No abstract