Efficient Deep Learning for Stereo Matching Wenjie Luo; Schwing, Alexander G.; Urtasun, Raquel
2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR),
2016-June
Conference Proceeding
In the past year, convolutional neural networks have been shown to perform extremely well for stereo estimation. However, current architectures rely on siamese networks which exploit concatenation ...followed by further processing layers, requiring a minute of GPU computation per image pair. In contrast, in this paper we propose a matching network which is able to produce very accurate results in less than a second of GPU computation. Towards this goal, we exploit a product layer which simply computes the inner product between the two representations of a siamese architecture. We train our network by treating the problem as multi-class classification, where the classes are all possible disparities. This allows us to get calibrated scores, which result in much better matching performance when compared to existing approaches.
Creating road maps is essential for applications such as autonomous driving and city planning. Most approaches in industry focus on leveraging expensive sensors mounted on top of a fleet of cars. ...This results in very accurate estimates when exploiting a user in the loop. However, these solutions are very expensive and have small coverage. In contrast, in this paper we propose an approach that directly estimates road topology from aerial images. This provides us with an affordable solution with large coverage. Towards this goal, we take advantage of the latest developments in deep learning to have an initial segmentation of the aerial images. We then propose an algorithm that reasons about missing connections in the extracted road topology as a shortest path problem that can be solved efficiently. We demonstrate the effectiveness of our approach in the challenging TorontoCity dataset 23 and show very significant improvements over the state-of-the-art.
Long non-coding RNAs (lncRNAs) act as important regulators of tumorigenesis and development in bladder cancer. However, the underlying molecular mechanisms remain elusive. We previously identified a ...novel lncRNA signature related to immunity and progression in bladder cancer. Here we further explored the function of RP11-89, a lncRNA discovered in the previous signature. Loss- and gain-of function experiments were performed using CCK-8 assay, flow cytometry, Transwell assays, scratch tests and subcutaneous nude mouse models. High-throughput RNA sequencing was conducted to identify dysregulated genes in bladder cancer cells with RP11-89 knockdown or overexpression. Regulation of RP11-89 on miR-129-5p and PROM2 was explored through luciferase reporter assay, RIP assay and RNA pull-down assay. RP11-89 promoted cell proliferation, migration and tumorigenesis and inhibited cell cycle arrest via the miR-129-5p/PROM2 axis. We found that RP11-89 "sponges" miR-129-5p and upregulates PROM2. Elevated PROM2 in cells was associated with attenuated ferroptosis through iron export, formation of multivesicular bodies and less mitochondrial abnormalities. We demonstrated that RP11-89 is a novel tumorigenic regulator that inhibits ferroptosis via PROM2-activated iron export. RP11-89 may serve as a potential biomarker for targeted therapy in bladder cancer.
Activation of microglia is a prominent pathological feature in tauopathies, including Alzheimer's disease. How microglia activation contributes to tau toxicity remains largely unknown. Here we show ...that nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) signaling, activated by tau, drives microglial-mediated tau propagation and toxicity. Constitutive activation of microglial NF-κB exacerbated, while inactivation diminished, tau seeding and spreading in young PS19 mice. Inhibition of NF-κB activation enhanced the retention while reduced the release of internalized pathogenic tau fibrils from primary microglia and rescued microglial autophagy deficits. Inhibition of microglial NF-κB in aged PS19 mice rescued tau-mediated learning and memory deficits, restored overall transcriptomic changes while increasing neuronal tau inclusions. Single cell RNA-seq revealed that tau-associated disease states in microglia were diminished by NF-κB inactivation and further transformed by constitutive NF-κB activation. Our study establishes a role for microglial NF-κB signaling in mediating tau spreading and toxicity in tauopathy.
End-To-End Interpretable Neural Motion Planner Zeng, Wenyuan; Luo, Wenjie; Suo, Simon ...
2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR),
06/2019
Conference Proceeding
Open access
In this paper, we propose a neural motion planner for learning to drive autonomously in complex urban scenarios that include traffic-light handling, yielding, and interactions with multiple ...road-users. Towards this goal, we design a holistic model that takes as input raw LIDAR data and a HD map and produces interpretable intermediate representations in the form of 3D detections and their future trajectories, as well as a cost volume defining the goodness of each position that the self-driving car can take within the planning horizon. We then sample a set of diverse physically possible trajectories and choose the one with the minimum learned cost. Importantly, our cost volume is able to naturally capture multi-modality. We demonstrate the effectiveness of our approach in real-world driving data captured in several cities in North America. Our experiments show that the learned cost volume can generate safer planning than all the baselines.
Haplodeficiency of the microglia gene TREM2 increases risk for late-onset Alzheimer’s disease (AD) but the mechanisms remain uncertain. To investigate this, we used high-resolution confocal and ...super-resolution (STORM) microscopy in AD-like mice and human AD tissue. We found that microglia processes, rich in TREM2, tightly surround early amyloid fibrils and plaques promoting their compaction and insulation. In Trem2- or DAP12-haplodeficient mice and in humans with R47H TREM2 mutations, microglia had a markedly reduced ability to envelop amyloid deposits. This led to an increase in less compact plaques with longer and branched amyloid fibrils resulting in greater surface exposure to adjacent neurites. This was associated with more severe neuritic tau hyperphosphorylation and axonal dystrophy around amyloid deposits. Thus, TREM2 deficiency may disrupt the formation of a neuroprotective microglia barrier that regulates amyloid compaction and insulation. Pharmacological modulation of this barrier could be a novel therapeutic strategy for AD.
•TREM2/DAP12 signaling regulates microglia process envelopment of amyloid plaques•Loss of microglia envelopment in TREM2/DAP12 deficiency reduces plaque compaction•STORM microscopy shows greater fibril branching and surface area in TREM2 deficiency•Human R47H TREM2 variant impairs the microglia barrier and worsens axonal dystrophy
Yuan, Condello, et al. demonstrate that haplodeficiency of the microglia-specific gene TREM2 markedly impairs the ability of microglia to compact and insulate amyloid deposits. Loss of this neuroprotective microglia function leads to marked axonal dystrophy, potentially contributing to the increased risk of dementia for carriers of TREM2 loss-of-function mutations.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Accumulation of neurotoxic amyloid- is a major hallmark of Alzheimer's disease. Formation of amyloid- is catalysed by -secretase, a protease with numerous substrates. Little is known about the ...molecular mechanisms that confer substrate specificity on this potentially promiscuous enzyme. Knowledge of the mechanisms underlying its selectivity is critical for the development of clinically effective -secretase inhibitors that can reduce amyloid- formation without impairing cleavage of other -secretase substrates, especially Notch, which is essential for normal biological functions. Here we report the discovery of a novel -secretase activating protein (GSAP) that drastically and selectively increases amyloid- production through a mechanism involving its interactions with both -secretase and its substrate, the amyloid precursor protein carboxy-terminal fragment (APP-CTF). GSAP does not interact with Notch, nor does it affect its cleavage. Recombinant GSAP stimulates amyloid- production in vitro. Reducing GSAP concentrations in cell lines decreases amyloid- concentrations. Knockdown of GSAP in a mouse model of Alzheimer's disease reduces levels of amyloid- and plaque development. GSAP represents a type of -secretase regulator that directs enzyme specificity by interacting with a specific substrate. We demonstrate that imatinib, an anticancer drug previously found to inhibit amyloid- formation without affecting Notch cleavage, achieves its amyloid- -lowering effect by preventing GSAP interaction with the -secretase substrate, APP-CTF. Thus, GSAP can serve as an amyloid- -lowering therapeutic target without affecting other key functions of -secretase.
Full text
Available for:
DOBA, IJS, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
Although two-stage methods of instance segmentation achieve better performance than one-stage counterparts, the segmentation results on overlapping objects are unsatisfactory. We found that occlusion ...significantly impacts the location of adjacent objects and produces coarse masks without adequate refinements. To circumvent the issue, we propose a hybrid model for instance segmentation called HTCIS, which iteratively forms the detection and segmentation. The main idea is to improve overall performance by optimizing every component based on a two-stage cascade structure. Compared with existing models, our approach decreases the loss of feature information, including semantic and detailed features. The detection branch prioritizes location accuracy when ranking bounding boxes, while the segmentation branch explores more contextual information and segments pixels in a multi-view fashion with the guide of an attention mechanism. Experimental results demonstrate that HTCIS is capable of processing occlusion. We conclude that multi-refinement of two-stage cascade is essential for accurate segmentation of overlapping objects, and our optimization is efficient in achieving this goal.
Full text
Available for:
EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, KILJ, KISLJ, MFDPS, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, UKNU, UL, UM, UPUK, VKSCE, ZAGLJ
Full-length RNA sequencing (RNA-Seq) has been applied to bulk tissue, cell lines and sorted cells to characterize transcriptomes, but applying this technology to single cells has proven to be ...difficult, with less than ten single-cell transcriptomes having been analyzed thus far. Although single splicing events have been described for ≤200 single cells with statistical confidence, full-length mRNA analyses for hundreds of cells have not been reported. Single-cell short-read 3' sequencing enables the identification of cellular subtypes, but full-length mRNA isoforms for these cell types cannot be profiled. We developed a method that starts with bulk tissue and identifies single-cell types and their full-length RNA isoforms without fluorescence-activated cell sorting. Using single-cell isoform RNA-Seq (ScISOr-Seq), we identified RNA isoforms in neurons, astrocytes, microglia, and cell subtypes such as Purkinje and Granule cells, and cell-type-specific combination patterns of distant splice sites. We used ScISOr-Seq to improve genome annotation in mouse Gencode version 10 by determining the cell-type-specific expression of 18,173 known and 16,872 novel isoforms.
The design of highly active, Earth-abundant and stable electrocatalysts is important for efficient water splitting. In this work, we report the fabrication of RuP and Ru2P nanoparticles supported on ...ordered macroporous N-doped carbon hollow spheres (RuP/H-NC and Ru2P/H-NC) through a facile and scalable space-confined pyrolysis process. The RuP/H-NC catalyst exhibits Pt-like activity in alkaline electrolyte, by means of the macroporous structure with a larger specific area and more exposed active sites, as well as the synergistic effect between the RuP nanoparticles and N-doped carbon. Specifically, the RuP/H-NC catalyst yields superior hydrogen evolution reaction activity in terms of low overpotential of 19 mV in 1 M KOH to achieve a current density of 10 mA cm−2 and excellent durability, outperforming Ru2P/H-NC and most of the reported non-Pt catalysts. Further density function theory calculation reveals that RuP is more intrinsically active with favorable hydrogen adsorption Gibbs free energy than that of Ru2P.