International business machines corporation (20240127009). IN-MEMORY COMPUTING FOR APPROXIMATING KERNEL FUNCTIONS simplified abstract
Contents
- 1 IN-MEMORY COMPUTING FOR APPROXIMATING KERNEL FUNCTIONS
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 IN-MEMORY COMPUTING FOR APPROXIMATING KERNEL FUNCTIONS - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Potential Applications
- 1.6 Problems Solved
- 1.7 Benefits
- 1.8 Potential Commercial Applications
- 1.9 Possible Prior Art
- 1.10 Original Abstract Submitted
IN-MEMORY COMPUTING FOR APPROXIMATING KERNEL FUNCTIONS
Organization Name
international business machines corporation
Inventor(s)
[[:Category:Julian Röttger B�chel of Zurich (CH)|Julian Röttger B�chel of Zurich (CH)]][[Category:Julian Röttger B�chel of Zurich (CH)]]
Manuel Le Gallo-bourdeau of Horgen (CH)
Irem Boybat Kara of Adliswil (CH)
Abu Sebastian of Adliswil (CH)
IN-MEMORY COMPUTING FOR APPROXIMATING KERNEL FUNCTIONS - A simplified explanation of the abstract
This abstract first appeared for US patent application 20240127009 titled 'IN-MEMORY COMPUTING FOR APPROXIMATING KERNEL FUNCTIONS
Simplified Explanation
The patent application describes a method for determining a probability distribution corresponding to a kernel function, sampling weights from this distribution, programming memristive devices in an analog crossbar based on the sampled weights, and performing matrix-vector multiplication operations using the programmed crossbar.
- Memristive devices in an analog crossbar are programmed with weights sampled from a determined probability distribution.
- Each memristive device represents a weight in the analog crossbar.
- Matrix-vector multiplication operations are performed on analog inputs x and y using the programmed crossbar.
- A dot product is computed on the results of the matrix-vector multiplication operations.
Potential Applications
This technology could be applied in:
- Artificial intelligence
- Machine learning
- Signal processing
Problems Solved
This technology helps solve:
- Efficient computation of matrix-vector multiplication
- Implementation of neural networks in hardware
Benefits
The benefits of this technology include:
- Faster computation
- Lower power consumption
- Scalability for large-scale applications
Potential Commercial Applications
This technology could be used in:
- Edge computing devices
- IoT devices
- Neuromorphic computing systems
Possible Prior Art
One possible prior art for this technology is:
- Memristive crossbar arrays for neural network implementations
What is the impact of this technology on current computing systems?
This technology can significantly improve the efficiency and speed of matrix-vector multiplication operations, which are fundamental in many computational tasks. By implementing these operations in hardware using memristive devices, the overall performance of computing systems can be enhanced.
How does this technology compare to traditional software-based approaches for matrix-vector multiplication?
This technology offers the advantage of parallel processing and lower power consumption compared to traditional software-based approaches. By utilizing analog crossbars and memristive devices, the computation can be performed in a more efficient and scalable manner.
Original Abstract Submitted
a probability distribution corresponding to the kernel function is determined and weights are sampled from the determined probability distribution corresponding to the given kernel function. memristive devices of an analog crossbar are programmed based on the sampled weights, where each memristive device of the analog crossbar is configured to represent a corresponding weight. two matrix-vector multiplication operations are performed on an analog input x and an analog input y using the programmed crossbar and a dot product is computed on results of the matrix-vector multiplication operations.
