AlpVision SA, a pioneering information and communication technology company and a global leader in advanced signal processing applications, announces at the AI summit in London the release plan of bioserver.net, a new cloud computing service based on Biological Neural Networks (BNN).
By integrating neural cell culture wetware, CMOS MEAs hardware, signal processing software and internet frontend technologies, bioserver.net will offer artificial intelligence researchers a remote access to living, low-energy consumption computing units of real neural networks as opposed to the silicon-based artificial neural networks. “BNN cultures have long been used by the pharmaceutical industry in clinical pharmacology during drug development using dedicated cell culture protocols to grow neural networks and maintain them for several months, possibly several years. The recent development of high-density CMOS MEAs interfacing now enables high throughput of both stimulation and read-out signals in direct connection to the living BNNs at increased resolution, both spatial through thousands of electrodes and temporal through signal sampling at several kHz. “This represents an unprecedented opportunity to engineer biological processing systems integrating wetware components. We envision that BNNs can bring a inherent neuron processing capability at a much lower power consumption cost than their silicon equivalents. Keep in mind the brain only consumes about 20W to perform its high level complex cognitive tasks,” said Fred D. Jordan, PhD, AlpVision’s co-founder and Chief Executive Officer.
In addition to facilitating remote access to monitoring BNN cultures for AI researchers without the need to build and maintain a wetlab facility, the bioserver.net frontend API will also offer various remote programming options for the backend computers in charge with controlling the BNNs in real time, similar to a remote desktop application. “In order to conduct their research, AI experts need to exploit the computing and learning capabilities of the BNNs with more advanced signal processing. Thanks to the bioserver.net programming tools, the high throughput MEA signals (of up to several gigabytes per minute) may thus be locally pre-computed, post-processed and filtered according to the individual needs of the master application. It is even possible to operate deep learning algorithms in the end-to-end system architecture over the BNN layer,” said Martin Kutter, PhD, AlpVision co-founder and President.