Blog Review: Nov. 28

Arm’s Bo Eyole contends that the next generation of machine learning algorithms will have to deal with a vast amount of messy, unlabeled data and takes a look at some of the techniques, such as reinforcement learning and evolutionary computing, now being explored.

Cadence’s Paul McLellan considers how IP systems are increasingly limited by memory bandwidth rather than compute power and where different memory architectures such as on-chip memory, HBM, and GDDR fit in.

Mentor’s Randy Allen argues that the separation between high performance computing and embedded software development is coming to an end as embedded compute power increases to meet the needs of machine learning and computer vision.

Synopsys’ Taylor Armerding explains the proposed Consumer Data Protection Act in the U.S., an effort to curb the sharing of consumer’s data with third parties and hold companies accountable for poor security if they are breached.

In a blog for SEMI, Alissa M. Fitzgerald of A.M. Fitzgerald & Associates points out some exciting new MEMS and sensor technologies, including MEMS that consume no power while standing by and biodegradable screen- and 3D-printed devices for soil monitoring.

A Rambus writer takes a look at what Differential Power Analysis-Resistant Software Libraries do and how they can be useful, particularly if designing side-channel resistant hardware is not a realistic option.

Nvidia’s Tony Kontzer points to a startup using convolutional neural networks to count, classify, and identify the gender of captured mosquitos in an effort to better understand disease transmission and prevent mosquito-borne epidemics.

Plus, check out the latest videos on topics ranging from directed self-assembly to AI training:

Smart Manufacturing: How to utilize manufacturing data and AI/ML to improve efficiency and yield.

ATE Lab To Fab: How to close the gap between the design and test worlds to improve coverage and shorten time to market.

Using DSA With EUV: Why directed self-assembly still has an important role to play at the most advanced nodes.

AI Training Chips: How to speed up algorithms and improve performance.

Chips In Space: Using eFPGAs in satellites and airplanes.