Resistive RAM (ReRAM) is a type of non-volatile memory that stores data by changing the resistance of a special material within the memory cell. Unlike traditional memory types, ReRAM can retain information even when the power is switched off. For artificial intelligence (AI) applications, ReRAM is valued for its speed, energy efficiency, and ability to…
Category: Model Optimisation Techniques
In-Memory Computing
In-memory computing is a way of processing and storing data directly in a computer’s main memory (RAM) instead of using traditional disk storage. This approach allows data to be accessed and analysed much faster because RAM is significantly quicker than hard drives or SSDs. It is often used in situations where speed is essential, such…
AI Hardware Acceleration
AI hardware acceleration refers to the use of specialised computer chips and devices that are designed to make artificial intelligence tasks run much faster and more efficiently than with regular computer processors. These chips, such as graphics processing units (GPUs), tensor processing units (TPUs), or custom AI accelerators, handle the heavy mathematical calculations required by…
Edge AI Optimization
Edge AI optimisation refers to improving artificial intelligence models so they can run efficiently on devices like smartphones, cameras, or sensors, which are located close to where data is collected. This process involves making AI models smaller, faster, and less demanding on battery or hardware, without sacrificing too much accuracy. The goal is to allow…
Analog AI Accelerators
Analog AI accelerators are specialised hardware devices that use analogue circuits to perform artificial intelligence computations. Unlike traditional digital processors that rely on binary logic, these accelerators process information using continuous electrical signals, which can be more efficient for certain tasks. By leveraging properties of analogue electronics, they aim to deliver faster processing and lower…
Rollup Compression
Rollup compression is a technique used in blockchain systems to reduce the size of transaction data before it is sent to the main blockchain. By compressing the information, rollups can fit more transactions into a single batch, lowering costs and improving efficiency. This method helps blockchains handle more users and transactions without slowing down or…
Sparse Gaussian Processes
Sparse Gaussian Processes are a way to make a type of machine learning model called a Gaussian Process faster and more efficient, especially when dealing with large data sets. Normally, Gaussian Processes can be slow and require a lot of memory because they try to use all available data to make predictions. Sparse Gaussian Processes…
Variational Inference
Variational inference is a method used in statistics and machine learning to estimate complex probability distributions. Instead of calculating exact values, which can be too difficult or slow, it uses optimisation techniques to find an easier distribution that is close enough to the original. This helps to make predictions or understand data patterns when working…
Neural Network Pruning
Neural network pruning is a technique used to reduce the size and complexity of artificial neural networks by removing unnecessary or less important connections, neurons, or layers. This process helps make models smaller and faster without significantly affecting their accuracy. Pruning often follows the training of a large model, where the least useful parts are…
Weight-Agnostic Neural Networks
Weight-Agnostic Neural Networks are a type of artificial neural network designed so that their structure can perform meaningful tasks before the weights are even trained. Instead of focusing on finding the best set of weights, these networks are built to work well with a wide range of fixed weights, often using the same value for…