Research Assistant, Project: 'Deep, differentiable networks for robust real-world robot behavior'

(salary grade E13 TV-L, under the reserve that funds are granted, starting no later than 1.10.2021 / for 3 years / closing date for applications 20.11.2020, Ref SCIoI-C3-35)

The Tech­nis­che Uni­versität Ber­lin invites applic­a­tions for a pos­i­tion for the Cluster of Excel­lence “Sci­ence of Intel­li­gence”.
What are the prin­ciples of intel­li­gence, shared by all forms of intel­li­gence, no mat­ter whether arti­fi­cial or bio­lo­gical, whether robot, com­puter pro­gram, human, or animal? And how can we apply these prin­ciples to cre­ate intel­li­gent tech­no­logy?
Answer­ing these ques­tions – in an eth­ic­ally respons­ible way – is the cent­ral sci­entific object­ive of the new Cluster of Excel­lence Sci­ence of Intel­li­gence (www.scienceofintelligence.de), where research­ers from a large num­ber of ana­lytic and syn­thetic dis­cip­lines – arti­fi­cial intel­li­gence, machine learn­ing, con­trol, robot­ics, com­puter vis­ion, beha­vi­oural bio­logy, psy­cho­logy, edu­ca­tional sci­ence, neur­os­cience, and philo­sophy – join forces to cre­ate a multi-dis­cip­lin­ary research pro­gram across uni­versit­ies and research insti­tutes in Ber­lin. Inter­dis­cip­lin­ary research pro­jects have been defined (https://www.scienceofintelligence.de/research/projects), which com­bine ana­lytic and syn­thetic research and which address key aspects of indi­vidual, social, and col­lect­ive intel­li­gence.

Working field

“Deep, dif­fer­en­ti­able net­works for robust real-world robot beha­vior”

Arti­fi­cial neural net­works are inspired by their bio­lo­gical coun­ter­parts. Pro­gress in AI and robot­ics indic­ates that indeed this inspir­a­tion has trans­ferred import­ant aspects of com­pu­ta­tion from bio­logy to com­puter sci­ence. Dif­fer­en­ti­ab­il­ity and recur­rence are two recur­ring com­pu­ta­tional pat­terns that one can find in bio­logy and in arti­fi­cial neural com­pu­ta­tion, e.g. in recurs­ive Bayes fil­ters and in recur­rent neural net­works (RNN). In this pro­ject, we will identify spe­cific com­pu­ta­tional pat­terns cap­able of gen­er­at­ing intel­li­gent beha­vior in the real world based on sens­ory feed­back. Good can­did­ates for such pat­terns are net­works of dif­fer­en­ti­able Bayes fil­ters. In col­lab­or­a­tion with beha­vi­oral bio­lo­gists, psy­cho­lo­gists, and cog­nit­ive sci­ent­ists, we will take inspir­a­tion from known aspects of human and animal cog­ni­tion. Our goal is to identify com­pu­ta­tional pat­terns for intel­li­gent beha­vior and to demon­strate their effect­ive­ness in real-world robotic sys­tems.

Requirements

  • MS degree in com­puter sci­ence or sim­ilar field; PhD is desired
  • Research exper­i­ence in robot­ics, machine learn­ing, com­puter vis­ion, and/or con­trol
  • Exper­i­ence in apply­ing (deep) learn­ing to robotic con­trol prob­lems
  • Interest in inter­dis­cip­lin­ary research in the con­text of the Cen­ter of Excel­lence “Sci­ence of Intel­li­gence”
  • Excel­lent soft­ware engin­eer­ing and pro­gram­ming skills in C++ and/or Python
  • Excel­lent Eng­lish writ­ing and com­mu­nic­a­tion skills; good Ger­man lan­guage skills or the will­ing­ness to learn them

Application procedure

Can­did­ates should upload their applic­a­tion prefer­ably via the portal www.scienceofintelligence.de/jobs in order to receive full con­sid­er­a­tion.

Applic­a­tions should include: motiv­a­tion let­ter, cur­riculum vitae, tran­scripts of records (for both BSc and MSc), cop­ies of degree cer­ti­fic­ates (BSc, MSc, PhD), abstracts of Bach­elor-, Mas­ter, and PhD-thesis, list of pub­lic­a­tions and one selec­ted manuscript (if applic­able), two names of qual­i­fied per­sons who are will­ing to provide ref­er­ences, and any doc­u­ments can­did­ates feel may help us assess their com­pet­ence.

By sub­mit­ting your applic­a­tion via email you con­sent to hav­ing your data elec­tron­ic­ally pro­cessed and saved. Please note that we do not provide a guar­anty for the pro­tec­tion of your per­sonal data when sub­mit­ted as unpro­tec­ted file. Please find our data pro­tec­tion notice acc. DSGVO (Gen­eral Data Pro­tec­tion Reg­u­la­tion) at TU web­site quick access 214041.

To ensure equal oppor­tun­it­ies between women and men, applic­a­tions by women with the required qual­i­fic­a­tions are expli­citly desired. Qual­i­fied indi­vidu­als with dis­ab­il­it­ies will be favored. The TU Ber­lin val­ues the diversity of its mem­bers and is com­mit­ted to the goals of equal oppor­tun­it­ies.

Tech­nis­che Uni­versität Ber­lin – Der Präsid­ent – Cluster SCIoI, Sekr. SCIoI, March­str. 23, 10587 Ber­lin

 

 

User registration

You don't have permission to register

Reset Password