18543357. APPARATUSES, METHODS, AND SYSTEMS FOR NEURAL NETWORKS simplified abstract (Intel Corporation)

APPARATUSES, METHODS, AND SYSTEMS FOR NEURAL NETWORKS

Organization Name

Intel Corporation

Inventor(s)

Swagath Venkataramani of Tirupur (IN)

Dipankar Das of Pune (IN)

Ashish Ranjan of West Lafayette IN (US)

Subarno Banerjee of Kolkata (IN)

Sasikanth Avancha of Malur Taluk (IN)

Ashok Jagannathan of Bangalore (IN)

Ajaya V. Durg of Austin TX (US)

Dheemanth Nagaraj of Bangalore (IN)

Bharat Kaul of Bengaluru (IN)

Anand Raghunathan of West Lafayette IN (US)

APPARATUSES, METHODS, AND SYSTEMS FOR NEURAL NETWORKS - A simplified explanation of the abstract

This abstract first appeared for US patent application 18543357 titled 'APPARATUSES, METHODS, AND SYSTEMS FOR NEURAL NETWORKS

Simplified Explanation

The patent application describes methods and apparatuses for processing neural networks, including a system with fully connected layer chips and convolutional layer chips connected by an interconnect. The system includes tiles for forward propagation, back propagation, and weight gradient computation between memory intensive tiles.

• Fully connected layer chips and convolutional layer chips are coupled by an interconnect in the neural network processing apparatus.
• The system includes compute intensive tiles for forward propagation, back propagation, and weight gradient computation between memory intensive tiles.

Potential Applications

This technology could be applied in various fields such as image recognition, natural language processing, and autonomous vehicles.

Problems Solved

This technology helps in improving the efficiency and speed of neural network processing, enabling faster and more accurate computations.

Benefits

The benefits of this technology include faster processing speeds, improved accuracy in neural network computations, and enhanced performance in various applications.

Potential Commercial Applications

The potential commercial applications of this technology include in industries such as healthcare, finance, and manufacturing for tasks like medical image analysis, fraud detection, and quality control.

Possible Prior Art

One possible prior art could be the use of specialized hardware for neural network processing, such as GPUs and TPUs, but the specific architecture described in this patent application may be novel in its approach.

How does this technology compare to existing neural network processing systems in terms of speed and efficiency?

The article does not provide a direct comparison with existing systems in terms of speed and efficiency, leaving a gap in understanding the competitive advantages of this technology.

What are the potential limitations or challenges in implementing this neural network processing system in real-world applications?

The article does not address potential limitations or challenges in implementing this technology in real-world applications, leaving room for further exploration of practical considerations.

Original Abstract Submitted

Methods and apparatuses relating to processing neural networks are described. In one embodiment, an apparatus to process a neural network includes a plurality of fully connected layer chips coupled by an interconnect; a plurality of convolutional layer chips each coupled by an interconnect to a respective fully connected layer chip of the plurality of fully connected layer chips and each of the plurality of fully connected layer chips and the plurality of convolutional layer chips including an interconnect to couple each of a forward propagation compute intensive tile, a back propagation compute intensive tile, and a weight gradient compute intensive tile of a column of compute intensive tiles between a first memory intensive tile and a second memory intensive tile.