PHD - hybrid architecture for advanced signal processing and deep neural network

neues Angebot (05/03/2020)

stellenbeschreibung

Job Description Deep Neural Networks offer excellent performance for dealing with complex perception tasks, such as object detections, semantic scene segmentations, and behavior predictions. Inferring such networks in real-time demands high computation and memory bandwidth, along with the needed flexibility to cope with future network topologies. The most straightforward approach to solve this issue is by implementing dedicated hardware accelerator for inferring deep learning networks. In embedded devices, inferring with deep learning is not the only task. Processing incoming data to deliver robust input for deep learning networks, and handling advanced signal processing tasks that are nowadays cannot be efficiently solved with deep learning, such as time-frequency transformation, signal interference suppression of MIMO systems, are the most crucial parts of processing chain. The area and power constraint in embedded devices requires rethinking on hardware design approach of both above-mentioned domains. In this thesis, we evaluate the architectures of digital signal processors and deep learning accelerators by taking into account the underlying algorithms for specific target applications, and design hybrid architecture that fits both domain

* Create Something New

:
you profile and identify the algorithm regions of advanced signal processing and deep learning that warrant emerging hardware support in term of computational complexity. * reliable implementation:

you optimize the algorithm for hardware and specify the complexity of the hardware implementations. * think holistically:

you devise the hybrid architecture based on the identified algorithms. * integrated implementation:

you map and realize the processing chain into the hybrid architecture. * structured evaluation:

you supervise bachelor and master theses. qualifications:

education:

degree (master/diploma) in electrical engineering, informatics or computer engineering personality and working practice:

creative, assertiveness, abstract and strategic thinking, experience and knowledge:
practical experiences with fpgas and hardware description languages (vhdl, verilog, systemverilog, systemc), good skills in modern mathematics, e. g. machine learning, solver, neural networks, good knowledge in the area of advanced signal, image, video processing and vlsi design qualifications:

proficiency in programming languages (c/c++, python, matlab) languages:
fluent in english and german written and spoken

Bewerben Sie sich jetzt um

PHD - hybrid architecture for advanced signal processing and deep neural networks
Sie werden auf eine andere Website weitergeleitet.
jetzt anmelden
Seien Sie der Erste, der neue Stellenangebote erhält
Mein Stellen-Alarm
Ihr Alarm wurde Ihrer Seite
Mein Stellen-Alarm
hinzugefügt.
Ihre Bestätigung steht noch aus,
gehen Sie bitte zu Ihrem E-Mail-Posteingang
und klicken Sie auf den Bestätigungslink, den wir Ihnen gesendet haben
mehr...

zur Vakanz

Weitere Jobs von Robert Bosch Tool