Vijayalakshmi Srinivasan

This inventor holds 89 USPTO granted patents and 8 published patent applications and 2 EPO patents, primarily in Deep Learning Optimization (CPC class G06F12-0862). Top assignee: International Business Machines Corporation. Active years: 2003-2026.

IDiyas Innovation Intelligence. (2026). Inventor Profile: Vijayalakshmi Srinivasan. Retrieved from https://idiyas.com/inventor/vijayalakshmi-srinivasan

Data source: USPTO (United States Patent and Trademark Office) public patent records. Weekly synchronization. How IDiyas builds this profile

USPTO Granted Patents = 89 (Patent Applications = 8)

% Patents Active = 48.3

EPO Patents = 2 (Patent Applications = 4)

Average Co-Inventor Count = 3.8

ph-index = 12

Forward Citations = 580(Granted Patents)

Forward Citations (Not Self Cited) = 559(Dec 10, 2025)

Inventors with similar research interests:

Location History:

Ann Arbor, MI (US) (2003)
New York City, NY (US) (2016 - 2023)
New York, NY (US) (2006 - 2024)
Yorktown Heights, NY (US) (2011 - 2024)

Company Filing History:

89 out of 166,409

International Business Machines Corporation

patents

Years Active: 2003-2026

Loading Chart...

Areas of Expertise:

Data Manipulation

Artificial Intelligence

Matrix Multiplication

Deep Neural Networks

Systolic Arrays

Cache Preloading

Memory Architecture

Quantization

Performance Analysis

Data Delivery

Multi-Processor Systems

Thread-Based Management

89 patents (USPTO):Sort by Forward CitationsExplore Patents

1. 12689537 - Single-producer-multiple consumers synchronization and multicast data transfer (2026-07-21),

2. 12657439 - Fused convolutions for fast deep neural network (2026-06-16),

3. 12639396 - Reformatting of tensors to provide sub-tensors (2026-05-26),

4. 12632233 - Integrating loop unrolling and loop splitting to reduce control overheads (2026-05-19),

5. 12468947 - Stickification using anywhere padding to accelerate data manipulation (2025-11-11),

6. 12399743 - Padding input data for artificial intelligence accelerators (2025-08-26),

7. 12236338 - Single function to perform combined matrix multiplication and bias add operations (2025-02-25),

8. 12141513 - Method to map convolutional layers of deep neural network on a plurality of processing elements with SIMD execution units, private memories, and connected as a 2D systolic processor array (2024-11-12),

9. 12094525 - Multichannel memory to augment local memory (2024-09-17),

10. 12056594 - Low precision deep neural network enabled by compensation instructions (2024-08-06),

11. 11941111 - Exploiting fine-grained structured weight sparsity in systolic arrays (2024-03-26),

12. 11831467 - Programmable multicast protocol for ring-topology based artificial intelligence systems (2023-11-28),

13. 11669489 - Sparse systolic array design (2023-06-06),

14. 11620132 - Reusing an operand received from a first-in-first-out (FIFO) buffer according to an operand specifier value specified in a predefined field of an instruction (2023-04-04),

15. 11556450 - Hybrid data-model parallelism for efficient deep learning (2023-01-17),

Title: **Vijayalakshmi Srinivasan: Innovating in Deep Learning Technologies**

Introduction

Vijayalakshmi Srinivasan is a distinguished inventor based in New York, NY, renowned for his contributions to the field of computing and deep learning. With a remarkable portfolio of 83 patents, Srinivasan has made significant strides in advancing technology, particularly in enhancing neural network efficiency and performance.

Latest Patents

Among his latest patents are innovations that demonstrate his expertise in manipulating tensor computations for machine learning applications. One of his notable inventions is the method for a single function that performs combined matrix multiplication and bias add operations. This invention streamlines the computational process by integrating several operations into a single invocation, optimizing performance in data processing.

Another groundbreaking patent involves a method to map convolutional layers of deep neural networks on multiple processing elements utilizing SIMD execution units. This method improves the performance of predefined convolution processing by formalizing design space exploration based on a computer’s architecture. The processor calculates performance metrics based on predefined parameters, thereby guiding optimal design choices for enhancing deep learning efficiency.

Career Highlights

Vijayalakshmi Srinivasan is currently associated with International Business Machines Corporation (IBM), where he continues to push the boundaries of innovation in technology. His work has not only contributed to theoretical advancements but also has practical implications in real-world applications, particularly in artificial intelligence and machine learning.

Collaborations

Throughout his career, Srinivasan has collaborated with esteemed colleagues such as Alper Buyuktosunoglu and Brian Robert Prasky. Their combined expertise in various areas of computing has led to the development of cutting-edge technologies that impact the industry significantly.

Conclusion

Vijayalakshmi Srinivasan's work exemplifies the spirit of innovation and dedication in the field of deep learning and computational technologies. With numerous patents to his name and ongoing contributions at IBM, he continues to shape the future of technology through his inventive pursuits. His achievements serve as an inspiration for aspiring inventors and professionals in the tech industry.

Profile summary based on public USPTO records.