Learning Image Descriptors with Boosting Trzcinski, Tomasz; Christoudias, Mario; Lepetit, Vincent
IEEE transactions on pattern analysis and machine intelligence,
2015-March-1, 2015-Mar, 2015-3-1, 20150301, Letnik:
37, Številka:
3
Journal Article
Recenzirano
Odprti dostop
We propose a novel and general framework to learn compact but highly discriminative floating-point and binary local feature descriptors. By leveraging the boosting-trick we first show how to ...efficiently train a compact floating-point descriptor that is very robust to illumination and viewpoint changes. We then present the main contribution of this paper-a binary extension of the framework that demonstrates the real advantage of our approach and allows us to compress the descriptor even further. Each bit of the resulting binary descriptor, which we call BinBoost, is computed with a boosted binary hash function, and we show how to efficiently optimize the hash functions so that they are complementary, which is key to compactness and robustness. As we do not put any constraints on the weak learner configuration underlying each hash function, our general framework allows us to optimize the sampling patterns of recently proposed hand-crafted descriptors and significantly improve their performance. Moreover, our boosting scheme can easily adapt to new applications and generalize to other types of image data, such as faces, while providing state-of-the-art results at a fraction of the matching time and memory footprint.
The study of atmospheric chemistry-climate interactions is one of today's great computational challenges. Advances in the architecture of Graphics Processing Units (GPUs) in both raw computational ...power and memory bandwidth sparked the interest for General-Purpose computing on graphics accelerators in scientific applications. However, the introduction of GPUs in the High Performance Computing (HPC) landscape increased the complexity of software development, due to the inherent heterogeneity requirements of programming models and design approaches, creating a gap in uptake and attainable performance in the presently available scientific community codes. This paper provides an overview of the challenges encountered when using GPU accelerators to achieve optimal performance to calculate the kinetics of chemical tracers in climate models, the techniques used to address them and the insights gained from the process. The paper presents the development of a chemical kinetics code-to-code parser to automatically generate chemical kinetics calculations on three different generations of GPU accelerators (M2070, K80, and P100). The accelerated portion of the application achieves a speedup of up to 22×, equivalent to performance gains of +19 percent up to +90 percent compared with the processor-only version, when using a cluster of 8 Nodes with dual Intel E5-2680 v3 processor and a Kepler architecture (K80), allowing faster completion of the simulations. The paper also provides practical insights and relevant considerations for the development and acceleration of complex applications.
Domain Adaptation for Microscopy Imaging Becker, Carlos; Christoudias, C. Mario; Fua, Pascal
IEEE transactions on medical imaging
34, Številka:
5
Journal Article
Odprti dostop
Electron and light microscopy imaging can now deliver high-quality image stacks of neural structures. However, the amount of human annotation effort required to analyze them remains a major ...bottleneck. While machine learning algorithms can be used to help automate this process, they require training data, which is time-consuming to obtain manually, especially in image stacks. Furthermore, due to changing experimental conditions, successive stacks often exhibit differences that are severe enough to make it difficult to use a classifier trained for a specific one on another. This means that this tedious annotation process has to be repeated for each new stack. In this paper, we present a domain adaptation algorithm that addresses this issue by effectively leveraging labeled examples across different acquisitions and significantly reducing the annotation requirements. Our approach can handle complex, nonlinear image feature transformations and scales to large microscopy datasets that often involve high-dimensional feature spaces and large 3D data volumes. We evaluate our approach on four challenging electron and light microscopy applications that exhibit very different image modalities and where annotation is very costly. Across all applications we achieve a significant improvement over the state-of-the-art machine learning methods and demonstrate our ability to greatly reduce human annotation effort.
The ongoing energy transition from conventional fuels to renewable energy sources (RES) has given nations the potential to achieve levels of energy self-sufficiency previously thought unattainable. ...RES in the form of utility-scale solar and wind energy are currently the leading alternatives to fossil-fuel generation. Precise location siting that factors in efficiency limitations related to current and future climate variables is essential for enabling the green energy transition envisioned for 2050. In this context, understanding and mapping the intermittency of RES provides insights to energy system operators for their seamless integration into the grid. The Eastern Mediterranean and Middle East (EMME) region has the potential to harness vast amounts of RES. The scarcity of observations from weather station networks and the lack of private sector incentives for transitioning to RES mean that relevant, supporting weather and climate studies have been limited. This study employs the Weather Research and Forecasting model with Chemistry (WRF-CHEM) to estimate the RES technical potential of EMME countries and map the hourly generation profiles per source and country, simulated for the reference year 2015 and considering future conditions. The findings indicate that by 2050, seven countries within the region could transform into net energy exporters, while the remaining nine might remain reliant on energy imports or fossil fuels. Egypt emerges as a “powerhouse”, potentially enjoying a potential surplus energy generation of 76 GW per hour, whereas the United Arab Emirates may face an annual deficit of 955 TWh. Further, we derived the hourly generation profiles for wind and solar during different seasons. Four dominant patterns were identified. We find a complementary relationship for six countries, and for four countries, a substitute relationship between solar and wind energy generation. Greece stands out with a near-constant wind energy source, which would facilitate its integration into the national grid.
Display omitted
•89% of projected energy demand in the Eastern Mediterranean and Middle East in 2050 can be met through utility-scale renewable energy Sources (RES).•Egypt has the remarkably high potential of achieve 76 GW per hour surplus energy production.•Greece may require minor energy services to accommodate RES in the grid.•The United Arab Emirates and Iran may have relatively high electricity deficits if relying solely on utility-scale RES potential.
Modelling atmospheric composition and climate change on the global scale remains a great scientific challenge. Earth system models spend up to 85% of their total required computational resources on ...the integration of atmospheric chemical kinetics. We refactored a general atmospheric chemical kinetics solver system to maintain accuracy in single precision to alleviate the bottleneck in memory-limited climate-chemistry simulations and file input/output (I/O) and introduced vectorisation by intrinsic functions to increase data-level parallelism exposure. The application was validated using seven standard chemical mechanisms and evaluated against high-precision implicit methods. We reduced required integration steps by ×1.5–3-fold, in line with double precision, while maintaining numerical stability under the same conditions, accuracy to within 1%, and benefiting from halving the required memory and reducing overall simulation time by up to a factor two. Our results suggest single-precision chemical kinetics can allow significant reduction of computational requirements and/or increase of complexity in climate-chemistry simulations.