Neural Network Model Quantization On Mobile

The general definition of quantization states that it is the process of mapping continuous infinite values to a smaller set of discrete finite values. In this blog, we will talk about quantization in the context of neural network (NN) models, as the process of reducing the precision of the weights, biases, and activations. Moving from floating-point representations to low-precision fixed intege... » read more

Application-Optimized Processors

Executing a neural network on top of an NPU requires an understanding of application requirements, such as latency and throughput, as well as the potential partitioning challenges. Sharad Chole, chief scientist and co-founder of Expedera, talks about fine-grained dependencies, why processing packets out of order can help optimize performance and power, and when to use voltage and frequency scal... » read more

Thoughts On AI Consciousness

By Anda Ioana Enescu Buyruk and Catalin Tudor The rapid advancement of artificial intelligence (AI) has sparked profound discussions regarding the possibility of AI systems achieving consciousness. Such a development carries immense implications, forcing us to redirect our focus from studying the behavior of other organisms to scrutinizing ourselves. This article will delve into the concept ... » read more

CNN Hardware Architecture With Weights Generator Module That Alleviates Impact Of The Memory Wall

A technical paper titled “Mitigating Memory Wall Effects in CNN Engines with On-the-Fly Weights Generation” was published by researchers at Samsung AI Center and University of Cambridge. Abstract: "The unprecedented accuracy of convolutional neural networks (CNNs) across a broad range of AI tasks has led to their widespread deployment in mobile and embedded settings. In a pursuit for high... » read more

Optimizing Projected PCM for Analog Computing-In-Memory Inferencing (IBM)

A new technical paper titled "Optimization of Projected Phase Change Memory for Analog In-Memory Computing Inference" was published by researchers at IBM Research. "A systematic study of the electrical properties-including resistance values, memory window, resistance drift, read noise, and their impact on the accuracy of large neural networks of various types and with tens of millions of wei... » read more

Split Additive Manufacturing for Printed Neuromorphic Circuits (Karlsruhe Institute of Technology)

A new technical paper titled "Split Additive Manufacturing for Printed Neuromorphic Circuits" was published by researchers at Karlsruher Institut für Technologie (KIT). Abstract: "Printed and flexible electronics promises smart devices for application domains, such as smart fast moving consumer goods and medical wearables, which are generally untouchable by conventional rigid silicon tech... » read more

Issues And Challenges In Super-Resolution Object Detection And Recognition

If you want high performance AI inference, such as Super-Resolution Object Detection and Recognition, in your SoC the challenge is to find a solution that can meet your needs and constraints. You need inference IP that can run the model you want at high accuracy. You need inference IP that can run the model at the frame rate you want: higher frame rate = lower latency, more time for dec... » read more

Overview of Machine Learning Algorithms Used In Hardware Security (TU Delft)

A new technical paper titled "A Survey on Machine Learning in Hardware Security" was published by researchers at TU Delft. Abstract "Hardware security is currently a very influential domain, where each year countless works are published concerning attacks against hardware and countermeasures. A significant number of them use machine learning, which is proven to be very effective in ... » read more

Feasibility of Using Domain Wall-Magnetic Tunnel Junction for Magnetic Analog Addressable Memories

A new technical paper titled "Domain Wall-Magnetic Tunnel Junction Analog Content Addressable Memory Using Current and Projected Data" was published by researchers at UT Austin and Samsung Advanced Institute of Technology (SAIT). Abstract "With the rise in in-memory computing architectures to reduce the compute-memory bottleneck, a new bottleneck is present between analog and digital conver... » read more

Solving The Reliability Problem Of Memristor-Based Artificial Neural Networks

A technical paper titled "ReMeCo: Reliable Memristor-Based in-Memory Neuromorphic Computation" was published by researchers at Eindhoven University of Technology, University of Tehran, and USC. Abstract: "Memristor-based in-memory neuromorphic computing systems promise a highly efficient implementation of vector-matrix multiplications, commonly used in artificial neural networks (ANNs). H... » read more

← Older posts