18526603. DEVICE AND METHOD WITH BATCH NORMALIZATION simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)
Contents
- 1 DEVICE AND METHOD WITH BATCH NORMALIZATION
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 DEVICE AND METHOD WITH BATCH NORMALIZATION - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Key Features and Innovation
- 1.6 Potential Applications
- 1.7 Problems Solved
- 1.8 Benefits
- 1.9 Commercial Applications
- 1.10 Prior Art
- 1.11 Frequently Updated Research
- 1.12 Questions about Batch Normalization in Accelerators
- 1.13 Original Abstract Submitted
DEVICE AND METHOD WITH BATCH NORMALIZATION
Organization Name
Inventor(s)
Seung Hwan Hwang of Seongnam-si (KR)
DEVICE AND METHOD WITH BATCH NORMALIZATION - A simplified explanation of the abstract
This abstract first appeared for US patent application 18526603 titled 'DEVICE AND METHOD WITH BATCH NORMALIZATION
Simplified Explanation
The patent application describes a device and method utilizing batch normalization in an accelerator for convolution operations.
- Core modules with multiple cores perform convolution operations.
- Local reduction operation modules calculate local statistical values for each core.
- Global reduction operation module generates global statistical values based on local values.
- Normalization operation module normalizes feature map data using global statistical values.
Key Features and Innovation
- Accelerator with batch normalization for convolution operations.
- Efficient computation of statistical values for normalization.
- Improved performance and accuracy in processing feature map data.
Potential Applications
- Image and video processing applications.
- Machine learning and deep learning tasks.
- Signal processing and pattern recognition systems.
Problems Solved
- Addressing the need for efficient normalization in convolution operations.
- Enhancing the accuracy and reliability of statistical calculations in accelerators.
Benefits
- Improved performance in processing feature map data.
- Enhanced accuracy in convolution operations.
- Streamlined computation of statistical values for normalization.
Commercial Applications
Optimized Accelerators for Image Processing
This technology can be utilized in developing accelerators for image processing applications, enhancing speed and accuracy in tasks such as object recognition and image classification.
Prior Art
The patent application builds upon existing techniques in batch normalization and accelerator design to optimize convolution operations.
Frequently Updated Research
There may be ongoing research in optimizing batch normalization techniques for various types of accelerators and convolution operations.
Questions about Batch Normalization in Accelerators
How does batch normalization improve the efficiency of convolution operations in accelerators?
Batch normalization helps in normalizing feature map data, leading to improved performance and accuracy in convolution tasks.
What are the potential challenges in implementing batch normalization in accelerator designs?
Implementing batch normalization in accelerators may require careful consideration of hardware constraints and computational efficiency to ensure optimal performance.
Original Abstract Submitted
A device and method with batch normalization are provided. An accelerator includes: core modules, each core module including a respective plurality of cores configured to perform a first convolution operation using feature map data and a weight; local reduction operation modules adjacent to the respective core modules, each including a respective plurality of local reduction operators configured to perform a first local operation that obtains first local statistical values of the corresponding core module; a global reduction operation module configured to perform a first global operation that generates first global statistical values of the core module based on the first local statistical values of the core modules; and a normalization operation module configured to perform a first normalization operation on the feature map data based on the first global statistical values.
