Time filter

Source Type

Babenko A.,National Research University Higher School of Economics | Lempitsky V.,Skolkovo Institute of Science and Technology Skoltech
Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition

We propose a new vector encoding scheme (tree quantization) that obtains lossy compact codes for high-dimensional vectors via tree-based dynamic programming. Similarly to several previous schemes such as product quantization, these codes correspond to codeword numbers within multiple codebooks. We propose an integer programming-based optimization that jointly recovers the coding tree structure and the codebooks by minimizing the compression error on a training dataset. In the experiments with diverse visual descriptors (SIFT, neural codes, Fisher vectors), tree quantization is shown to combine fast encoding and state-of-the-art accuracy in terms of the compression error, the retrieval performance, and the image classification error. © 2015 IEEE. Source

Sarkar S.,Whitehead Institute For Biomedical Research | Maetzel D.,Whitehead Institute For Biomedical Research | Korolchuk V.I.,Vitality | Jaenisch R.,Whitehead Institute For Biomedical Research | Jaenisch R.,Skolkovo Institute of Science and Technology Skoltech

Autophagy is essential for cellular homeostasis and its dysfunction in human diseases has been implicated in the accumulation of misfolded protein and in cellular toxicity. We have recently shown impairment in autophagic flux in the lipid storage disorder, Niemann-Pick type C1 (NPC1) disease associated with abnormal cholesterol sequestration, where maturation of autophagosomes is impaired due to defective amphisome formation caused by failure in SNARE machinery. Abrogation of autophagy also causes cholesterol accumulation, suggesting that defective autophagic flux in NPC1 disease may act as a primary causative factor not only by imparting its deleterious effects, but also by increasing cholesterol load. However, cholesterol depletion treatment with HP-β-cyclodextrin impedes autophagy, whereas pharmacologically stimulating autophagy restores its function independent of amphisome formation. Of potential therapeutic relevance is that a low dose of HP-β-cyclodextrin that does not perturb autophagy, coupled with an autophagy inducer, may rescue both the cholesterol and autophagy defects in NPC1 disease. © 2014 Landes Bioscience. Source

Babenko A.,Moscow Institute of Physics and Technology | Lempitsky V.,Skolkovo Institute of Science and Technology Skoltech
Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition

We introduce a new compression scheme for high-dimensional vectors that approximates the vectors using sums of M codewords coming from M different codebooks. We show that the proposed scheme permits efficient distance and scalar product computations between compressed and uncompressed vectors. We further suggest vector encoding and codebook learning algorithms that can minimize the coding error within the proposed scheme. In the experiments, we demonstrate that the proposed compression can be used instead of or together with product quantization. Compared to product quantization and its optimized versions, the proposed compression approach leads to lower coding approximation errors, higher accuracy of approximate nearest neighbor search in the datasets of visual descriptors, and lower image classification error, whenever the classifiers are learned on or applied to compressed vectors. © 2014 IEEE. Source

Chai Y.,University of Oxford | Lempitsky V.,Skolkovo Institute of Science and Technology Skoltech | Zisserman A.,University of Oxford
Proceedings of the IEEE International Conference on Computer Vision

We propose a new method for the task of fine-grained visual categorization. The method builds a model of the base-level category that can be fitted to images, producing high-quality foreground segmentation and mid-level part localizations. The model can be learnt from the typical datasets available for fine-grained categorization, where the only annotation provided is a loose bounding box around the instance (e.g. bird) in each image. Both segmentation and part localizations are then used to encode the image content into a highly-discriminative visual signature. The model is symbiotic in that part discovery/localization is helped by segmentation and, conversely, the segmentation is helped by the detection (e.g. part layout). Our model builds on top of the part-based object category detector of Felzenszwalb et al., and also on the powerful Grab Cut segmentation algorithm of Rother et al., and adds a simple spatial saliency coupling between them. In our evaluation, the model improves the categorization accuracy over the state-of-the-art. It also improves over what can be achieved with an analogous system that runs segmentation and part-localization independently. © 2013 IEEE. Source

Steinkellner T.,Medical University of Vienna | Mus L.,Italian Institute of Technology | Mus L.,Pavlov Medical University | Eisenrauch B.,Medical University of Vienna | And 16 more authors.

Addiction to psychostimulants (ie, amphetamines and cocaine) imposes a major socioeconomic burden. Prevention and treatment represent unmet medical needs, which may be addressed, if the mechanisms underlying psychostimulant action are understood. Cocaine acts as a blocker at the transporters for dopamine (DAT), serotonin (SERT), and norepinephrine (NET), but amphetamines are substrates that do not only block the uptake of monoamines but also induce substrate efflux by promoting reverse transport. Reverse transport has been a focus of research for decades but its mechanistic basis still remains enigmatic. Recently, transporter-interacting proteins were found to regulate amphetamine-triggered reverse transport: calmodulin kinase IIα (αCaMKII) is a prominent example, because it binds the carboxyl terminus of DAT, phosphorylates its amino terminus, and supports amphetamine-induced substrate efflux in vitro. Here, we investigated whether, in vivo, the action of amphetamine was contingent on the presence of αCaMKII by recording the behavioral and neurochemical effects of amphetamine. Measurement of dopamine efflux in the dorsal striatum by microdialysis revealed that amphetamine induced less dopamine efflux in mice lacking αCaMKII. Consistent with this observation, the acute locomotor responses to amphetamine were also significantly blunted in αCaMKII-deficient mice. In addition, while the rewarding properties of amphetamine were preserved in αCaMKII-deficient mice, their behavioral sensitization to amphetamine was markedly reduced. Our findings demonstrate that amphetamine requires the presence of αCaMKII to elicit a full-fledged effect on DAT in vivo: αCaMKII does not only support acute amphetamine-induced dopamine efflux but is also important in shaping the chronic response to amphetamine. © 2014 American College of Neuropsychopharmacology. Source

Discover hidden collaborations