Transfer Learning vs Incremental Learning for training neural nets

Brainchip, the AI processor specialist, has looked whether transfer learning is more efficient than incremental learning in training neural nets to perform AI/ML tasks.. In transfer learning, applicable knowledge established in a previously trained AI model is “imported” and used as the basis of a new model. After taking this shortcut of using a pretrained ...

This story continues at Transfer Learning vs Incremental Learning for training neural nets

Or just read more coverage at Electronics Weekly

Rockley adds BEOL processing to Skywater foundry contract

Skywater, the Minnesota foundry, will undertake BEOL processing for Rockley Photonics’ spectrophotometer-on-a-chip health monitoring product. Rockley, founded by Bookham Technology founder Andrew Rickman, has been using Skywater for front-end processing. “SkyWater provides us with a domestic U.S. source for advanced wafer-level components which is a critical part of our multi-source supply chain strategy,” says Rickman, ...

This story continues at Rockley adds BEOL processing to Skywater foundry contract

Or just read more coverage at Electronics Weekly

iSpace announces its next-gen larger lunar lander

iSpace – the lunar space transportation company based in Tokyo – has unveiled its next generation lunar lander, dubbed Series 2. It will first be used for the company’s third lunar mission. The new design of the lander sees an increase in size and payload design capacity. Standing at approximately 9 ft tall and 14 ...

This story continues at iSpace announces its next-gen larger lunar lander

Or just read more coverage at Electronics Weekly

ST adds to GaN portfolio

Easing the transition to high-efficiency wide-bandgap technology, ST  has released the MasterGaN3 and MasterGaN5 integrated power packages for applications up to 45W and 150W, respectively. Joining the MasterGaN1, MasterGaN2, and MasterGaN4, which target applications from 65W to 400W, the additions give extra flexibility to choose the optimum GaN device and driver solution when designing switched-mode power ...

This story continues at ST adds to GaN portfolio

Or just read more coverage at Electronics Weekly

More on: Tesla’s AI supercomputer

Tesla revealed its plans for a 1.1Eflop artificial intelligence supercomputer late last week, built around an 7nm in-house-developed IC with 354 custom processors delivering 362Tflop of processing and 16Tbit/s of IO bandwidth. The computer, to be called Dojo, is where the company plans to train neural networks for use in its cars. Within it, the ...

This story continues at More on: Tesla’s AI supercomputer

Or just read more coverage at Electronics Weekly

Farnell signs Epishine

Epishine, the Swedish printed organic solar cell specialist,  has signed a franchise agreement with Farnell. Epishine’s printed organic solar cells are optimised for harvesting energy from indoor, low energy lighting enabling organic solar power to be used everywhere. Design engineers can use  Epishine’s Light Energy Harvesting Evaluation Kit. Farnell is the first high service distributor to ...

This story continues at Farnell signs Epishine

Or just read more coverage at Electronics Weekly

US DoD contracts for Intel 18A foundry process

Intel has picked up the US  Department of Defence as a customer for its new foundry service operation. The contract was awarded through the NSTXL consortium-based S2MARTS OTA in the first phase of its multi-phase Rapid Assured Microelectronics Prototypes – Commercial (RAMP-C) program.  The RAMP-C program was created to facilitate the use of a US-based ...

This story continues at US DoD contracts for Intel 18A foundry process

Or just read more coverage at Electronics Weekly

Samsung HBM-PIM AI accelerator converges logic and memory

At Hot Chips, Samsung  showed processing-in-memory (PIM) technology with High Bandwidth Memory (HBM) integrated into an accelerator. The development is claimed to accelerate the move towards the convergence of memory and logic. HBM-PIM incorporates the AI processing function to enhance high-speed data processing in supercomputers and AI applications. HBM-PIM has since been tested in the ...

This story continues at Samsung HBM-PIM AI accelerator converges logic and memory

Or just read more coverage at Electronics Weekly

BAE Systems furthers defence collaboration with Intel

BAE Systems has announced another strategic business agreement for its FAST Labs R&D organisation having early access to selected Intel technologies. The agreement, says the company, is intended to help accelerate development of military applications. The agreement will enable it “to develop and more quickly field next-generation defense applications based on Intel’s most advanced technology”. ...

This story continues at BAE Systems furthers defence collaboration with Intel

Or just read more coverage at Electronics Weekly