A data-centric weak supervised learning for highway traffic incident detection.

Using the data from loop detector sensors for near-real-time detection of traffic incidents on highways is crucial to averting major traffic congestion. While recent supervised machine learning methods offer solutions to incident detection by leveraging human-labeled incident data, the false alarm rate is often too high to be used in practice. Specifically, the inconsistency in the human labeling of the incidents significantly affects the performance of supervised learning models. To that end, we focus on a data-centric approach to improve the accuracy and reduce the false alarm rate of traffic incident detection on highways. We develop a weak supervised learning workflow to generate high-quality training labels for the incident data without the ground truth labels, and we use those generated labels in the supervised learning setup for final detection. This approach comprises three stages. First, we introduce a data preprocessing and curation pipeline that processes traffic sensor data to generate high-quality training data through leveraging labeling functions, which can be domain knowledge-related or simple heuristic rules. Second, we evaluate the training data generated by weak supervision using three supervised learning models-random forest, k-nearest neighbors, and a support vector machine ensemble-and long short-term memory classifiers. The results show that the accuracy of all of the models improves significantly after using the training data generated by weak supervision. Third, we develop an online real-time incident detection approach that leverages the model ensemble and the uncertainty quantification while detecting incidents. Overall, we show that our proposed weak supervised learning workflow achieves a high incident detection rate (0.90) and low false alarm rate (0.08).

Loss of Miro1-directed mitochondrial movement results in a novel murine model for neuron disease.

Defective mitochondrial distribution in neurons is proposed to cause ATP depletion and calcium-buffering deficiencies that compromise cell function. However, it is unclear whether aberrant mitochondrial motility and distribution alone are sufficient to cause neurological disease. Calcium-binding mitochondrial Rho (Miro) GTPases attach mitochondria to motor proteins for anterograde and retrograde transport in neurons. Using two new KO mouse models, we demonstrate that Miro1 is essential for development of cranial motor nuclei required for respiratory control and maintenance of upper motor neurons required for ambulation. Neuron-specific loss of Miro1 causes depletion of mitochondria from corticospinal tract axons and progressive neurological deficits mirroring human upper motor neuron disease. Although Miro1-deficient neurons exhibit defects in retrograde axonal mitochondrial transport, mitochondrial respiratory function continues. Moreover, Miro1 is not essential for calcium-mediated inhibition of mitochondrial movement or mitochondrial calcium buffering. Our findings indicate that defects in mitochondrial motility and distribution are sufficient to cause neurological disease.

Mitochondrial association, protein phosphorylation, and degradation regulate the availability of the active Rab GTPase Ypt11 for mitochondrial inheritance.

The Rab GTPase Ypt11 is a Myo2-binding protein implicated in mother-to-bud transport of the cortical endoplasmic reticulum (ER), late Golgi, and mitochondria during yeast division. However, its reported subcellular localization does not reflect all of these functions. Here we show that Ypt11 is normally a low-abundance protein whose ER localization is only detected when the protein is highly overexpressed. Although it has been suggested that ER-localized Ypt11 and ER-mitochondrial contact sites might mediate passive transport of mitochondria into the bud, we found that mitochondrial, but not ER, association is essential for Ypt11 function in mitochondrial inheritance. Our studies also reveal that Ypt11 function is regulated at multiple levels. In addition to membrane targeting and GTPase domain-dependent effector interactions, the abundance of active Ypt11 forms is controlled by phosphorylation status and degradation. We present a model that synthesizes these new features of Ypt11 function and regulation in mitochondrial inheritance.

Structure-function analysis of the yeast mitochondrial Rho GTPase, Gem1p: implications for mitochondrial inheritance.

Mitochondria undergo continuous cycles of homotypic fusion and fission, which play an important role in controlling organelle morphology, copy number, and mitochondrial DNA maintenance. Because mitochondria cannot be generated de novo, the motility and distribution of these organelles are essential for their inheritance by daughter cells during division. Mitochondrial Rho (Miro) GTPases are outer mitochondrial membrane proteins with two GTPase domains and two EF-hand motifs, which act as receptors to regulate mitochondrial motility and inheritance. Here we report that although all of these domains are biochemically active, only the GTPase domains are required for the mitochondrial inheritance function of Gem1p (the yeast Miro ortholog). Mutations in either of the Gem1p GTPase domains completely abrogated mitochondrial inheritance, although the mutant proteins retained half the GTPase activity of the wild-type protein. Although mitochondrial inheritance was not dependent upon Ca(2+) binding by the two EF-hands of Gem1p, a functional N-terminal EF-hand I motif was critical for stable expression of Gem1p in vivo. Our results suggest that basic features of Miro protein function are conserved from yeast to humans, despite differences in the cellular machinery mediating mitochondrial distribution in these organisms.

Identification of genes expressed in the Arabidopsis female gametophyte.

The angiosperm female gametophyte typically consists of one egg cell, two synergid cells, one central cell, and three antipodal cells. Each of these four cell types has unique structural features and performs unique functions that are essential for the reproductive process. The gene regulatory networks conferring these four phenotypic states are largely uncharacterized. As a first step towards dissecting the gene regulatory networks of the female gametophyte, we have identified a large collection of genes expressed in specific cells of the Arabidopsis thaliana female gametophyte. We identified these genes using a differential expression screen based on reduced expression in determinant infertile1 (dif1) ovules, which lack female gametophytes. We hybridized ovule RNA probes with Affymetrix ATH1 genome arrays and validated the identified genes using real-time RT-PCR. These assays identified 71 genes exhibiting reduced expression in dif1 ovules. We further validated 45 of these genes using promoter::GFP fusions and 43 were expressed in the female gametophyte. In the context of the ovule, 11 genes were expressed exclusively in the antipodal cells, 11 genes were expressed exclusively or predominantly in the central cell, 17 genes were expressed exclusively or predominantly in the synergid cells, one gene was expressed exclusively in the egg cell, and three genes were expressed strongly in multiple cells of the female gametophyte. These genes provide insights into the molecular processes functioning in the female gametophyte and can be used as starting points to dissect the gene regulatory networks functioning during differentiation of the four female gametophyte cell types.