Machine Learning for Signal Processing

The SAMPL Lab at the Weizmann Institute of Science offers a postdoctoral position within the project C'MON-QSENS! (Continuously Monitored Quantum Sensors: Smart Tools and Applications) funded by QuantERA EU program in Quantum Technologies. The appointment will be for a two years term, (possibly) renewable for a third year.

The successful candidate will work on theoretical and algorithmic aspects of Quantum Statistical Inference and Quantum Information, with particular focus on schemes that could be relevant for quantum sensors. Solid background on either real-time signal processing techniques, machine learning, quantum stochastic equations, sequential analysis and optimal control are very desirable.

The candidate will work in close collaboration with other members of the C’MON-QSENS! consortium:

  John Calsamiglia Costa (Universitat Autònoma de Barcelona, SP)

• Jan Kołodyński (University of Warsaw, PL)
• Klaus Mølmer (Aarhus University, DK)
• Witlef Wieczorek (Chalmers University of Technology, SE)
• Kasper Jensen (University of Nottingham, UK)

Interested applicants should send a CV along with a short statement of purpose or presentation letter and arrange for two or three letters of recommendation to be sent to yonina.eldar@weizmann.ac.il using "C'MON-postdoc" as subject.

Proposed start of appointment: Immediately.

PhD students and Postdocs in the areas of signal processing, machine learning, medical imaging, communications and radar processing:

Host Professor: Yonina Eldar, department of mathematics and computer science, Weizmann Institute

Position Description: The Signal Acquisition Measurement Processing and Learning (SAMPL) Lab invites applications for doctoral/postdoctoral positions in the areas of signal processing and machine learning with applications in communications, radar, medical imaging and optical imaging. In the area of medical imaging, the work will be performed in close collaboration with leading hospitals in Israel and abroad. Some of the topics will be in close collaboration with PIs at MIT, Stanford, Princeton and the Broad Institute.
We are looking for thought leaders in these fields who can develop new areas and applications. The balance of work between theory and practice will vary on project-basis, and a successful candidate should be proficient in both aspects. Excellent written and presentation skills in English are an advantage.

About SAMPL:  The lab focuses on sampling, modeling and processing of continuous-time and discrete-time signals with the aim of extracting as much information as possible using minimal resources to perform various tasks, including communication, radar, medical and optical imaging and biological inference. The lab also develops model-based methods for artificial intelligence (AI) that aid in obtaining increased information from fewer samples, and facilitates the transition from pure theoretical research to the development, design and implementation of prototype systems and clinical studies. Our approach can drastically reduce the sampling and processing rates well below the Nyquist rate, and greatly improve the resolution which can be obtained from a limited number of samples in time, space and frequency. It also paves the way to new technologies such as wireless ultrasound and joint radar and communication systems. The group works closely with several major hospitals in Israel and with research groups at MIT, Stanford, Duke, Princeton and more.

For more details please visit Prof. Eldar’s website: http://www.wisdom.weizmann.ac.il/~yonina/YoninaEldar/index.html

To submit your application, please send an updated CV with 3 letters of recommendation and a cover letter to yonina.eldar@weizmann.ac.il

Classification of SAR image data continues to be a big challenge. Major difficulties include the scarcity of available data, and the difficulty of semantically interpreting the SAR backscattered signal. There are no large-scale, SAR-derived image databases for Remote Sensing image analysis and knowledge discovery. Furthermore, while optical image classification has seen a breakthrough with the advent of Deep Learning methods that require Big Data, SAR-based systems have so far not experienced the same progress, likely because not enough data associated training labels is available.

The nature of the adversarial samples occurring spontaneously depends on the sensor type. In SAR, the effect of strong scattering or the model of image formation and the physical processes behind need very specific methods for dealing with adversarial samples. Given the particular nature of these samples, the solutions to avoid their insertion in the training sets or to alleviate their effects must be tailored accordingly.

The main objective of this project is to give solutions for deep learning with spontaneous adversarial samples in the case of SAR data.

The successful candidate will be employed for 3 years by University Politehnica of Bucharest and will be part of CEOSpaceTech, a very dynamic research laboratory oriented towards space applications. She or he receives a financial package, which is twice the average salary in the country and additional mobility and family allowance, granted according to the rules for Early Stage Researchers (ESRs) in an EU Marie Sklodowska-Curie Actions Innovative Training Networks (ITN). A career development plan will be prepared for her/him in accordance with the supervisor. The plan will include a choice of more than 20 streamed or registered courses, one stage in Germany, and various outreach activities. For more information please visit the Marie Sklodowska-Curie Actions Innovative Training Networks website.

YOUR TASKS

• Use transformation methods for a relevant SAR data representation, in order to avoid insertion of adversarial samples.
• Design of DNNs for SAR data classification in order to achieve a given invariance to spontaneous adversarial samples.
• Define projections of features when learning the semantic axes for 3D visualization such to contextually disambiguate the meaning and to ensure a consistent training.

PROFILE

• A Master of Science in Computer Science is required. It could comprise the full range of mathematical, physical, engineering and technology disciplines related to sensor data acquisition and programming.
• Proficient English level.
• Image analysis, neural networks, programming languages, basic knowledge on optics could be an advantage.

PLANNED SECONDMENTS

• DLR, Munich, Germany, Prof. M. Datcu,10 months, theoretical aspects of DNN for SAR images and related topics on the impact of adversarial samples.

ADDITIONAL INFORMATION

References

• Marmanis, Dimitrios, et al. “Artificial generation of big data for improving image classification: A generative adversarial network approach on SAR data.” arXiv preprint arXiv:1711.02010 (2017).
• Zhao, Juanping, et al. “Contrastive-Regulated CNN in the Complex Domain: A Method to Learn Physical Scattering Signatures From Flexible PolSAR Images.” IEEE Transactions on Geoscience and Remote Sensing (2019).
• Goodwin, Justin A., et al. “Learning Robust Representations for Automatic Target Recognition.” arXiv preprint arXiv:1811.10714 (2018).

We seek a highly motivated postdoctoral researcher for a cutting-edge research project sponsored by Total Exploration & Production Research & Technology, USA. The research is held at Braude College of Engineering, Israel, in close collaboration with Total and includes visiting periods at Total (Houston, USA). The postdoctoral researcher will develop novel deep learning algorithms for solving complex seismic inversion problems. Topics of interest include:

• Deep Learning for 3D seismic inversion and imaging

• Multi-task deep learning

• Variational Networks and regularization for deep network solutions to inverse problems

• Relations between deep learning and compressed sensing, for seismic inversion

• Hyper Networks: dynamically self-tuning deep networks

• Neural Architecture Search (NAS) algorithms 

Candidates should hold a Ph.D. in engineering, computer science, geoscience, geophysics or related fields, with excellent Python programming skills and publications in one or more of the following areas:

• Deep Learning or Machine Learning

• Signal and Image Processing

• Parallel Computing and High-Performance Computing (HPC) 

• Wave Propagation

Experience in Deep Learning is a significant advantage. Prior knowledge in seismic data is an advantage, however, not required. The Postdoctoral position carries no teaching obligations, although teaching opportunities are possible for those interested, as well as opportunities to supervise undergraduate and graduate students. The minimum commitment is for one year, with a possibility to extend for a second year. We offer a highly competitive salary, in a joint academia and industry research, with first class research environment and computing resources. Interested candidates are welcome to send an updated CV to the principal investigators:

Dr. Amir Adler

Senior Lecturer, Electrical Engineering

Braude College of Engineering, Israel

E-mail: adleram@braude.ac.il

Homepage: https://amiradler.mit.edu

Dr. Mauricio Araya-Polo

Senior R&D Project Manager

Computational Science and Engineering

Total E&P R&T, USA

E-mail: mauricio.araya@total.com

 

A full time one-year position as a postdoctoral Fellow is available at the AViReS Lab ( https://avires.dimi.uniud.it ) of the Department of Mathematics, Computer Science and Physics (DMIF), University of Udine.   

The postdoc fellow will work with an interdisciplinary team of experienced researchers in the fields of audio signal processing, computer vision, machine learning, robotics. Primary co-mentors will be Prof. Carlo Drioli and Prof. Gian Luca Foresti. The goals of his research will be to develop specific methods and algorithms allowing to improve the effectiveness of localization and tracking of a moving target through the dynamic reconfiguration of a sensor network, to study the integration possibilities of acoustic and optical sensors with respect to the problem under study, to address application scenarios related to the use of mobile robotic platforms, in particular autonomous aerial drones (UAVs/MAVs) equipped with acoustic and optical sensors.
The research activities will be normally conducted at the AViReS Lab facilities in Udine. However note that, due to the ongoing Covid-19 outbreak, it is possible that for the first few months the activities will have to be conducted remotely.  
Total grant payd by the financer amounts to 19.367,00 Euros.

Application deadline is May 12, 2020 at 02:00 p.m. (Italian time).

Please contact Carlo Drioli (carlo.drioli@uniud.it) or Gian Luca Foresti (gianluca.foresti@uniud.it) for any further questions.

Detailed information and directions to apply can be found at the following links: 

http://web.uniud.it/ateneo/normativa/albo_ufficiale/258-2020/DRN_188_2020_Notice%20of%20competition_.pdf
http://web.uniud.it/ateneo/normativa/albo_ufficiale/258-2020

Postdoctoral Position - Decision Making with Limited Data using Artificial Intelligence - An Active Inference Approach

Signals and Systems Division, Department of Electrical Engineering,  Uppsala University has a vacancy for a postdoctoral position. The position is a part of the strategic research area effort eSSENCE´s PostDoc-program towards new e-science methods and tools for artificial intelligence in research.

Project description: Success of machine learning (ML) and artificial intelligence (AI) methods typically rely on the availability of large amounts of data. This dependence on high amounts of data/interactions is an important handicap for applying the current AI approaches in data-limited scenarios, such as Internet-of-Things scenarios. This project will address this handicap of limited data using the active inference (ActInf) framework. Similar to the reinforcement learning, the ActInf framework generates actions/policies so that a specific desired outcome is obtained by interacting with the surroundings. ActInf is closely connected to probabilistic dynamical models, belief propagation, and model based reinforcement learning. ActInf can also be used to recover standard cost-based control solutions for the linear quadratic setup, a well-known scenario which is of central importance in the control community.

Duties: To conduct original research in the area of decision making under limited data  using the ActInf framework, in particular i) develop novel, general-purpose, active inference based adaptive data collection, decision making and control strategies to optimize the overall inference and control performance under limited data,  ii) reveal the trade-offs between data collection, decision making and control performance and provide guidelines for cost-efficient autonomous operation for various application scenarios.

The duties include theoretical analysis, algorithm design and implementation via software-based simulations, and reporting of the results in the form of technical papers. Participation in the undergraduate and/or graduate education and supervision of PhD students is also required.

Requirements: PhD degree or a foreign degree equivalent to a PhD degree in Electrical Engineering or Computer Science with a background in Automatic Control, Signal Processing, Machine Learning or Communications. The PhD degree must have been obtained no more than three years prior to the application deadline. The three year period can be extended due to circumstances such as sick leave, parental leave, duties in labour unions, etc.

A proven publication record in top-ranked journals or conferences is required. Emphasis will be placed on computer programming abilities together with a strong mathematical background where previous research in active inference or closely related areas such as probabilistic dynamical models, information theory, optimization theory or reinforcement learning will be beneficial.

Starting date: 2020-06-01 or as otherwise agreed.

Further Information: The complete announcement text can be found here: https://www.uu.se/en/about-uu/join-us/details/?positionId=315433

For further information do not hesitate to contact Ayca Özcelikkale, ayca.ozcelikkale@angstrom.uu.se or  Anders Ahlén, anders.ahlen@signal.uu.se.

Application Instructions: Please submit your application by April 1, 2020 through Uppsala University´s recruitment system: https://uu.varbi.com/en/what:login/type:job/jobID:315433

Note that applications by email cannot be considered.

 

 

Applicants are invited for two Early Stage Researchers (ESR) of EU Marie Sklodowska-Curie Actions (MSCA) to be hosted in the Istituto Italiano di Tecnologia (IIT). The positions are for a fixed-term of 3 years and the successful applicants are expected to register for PhD at the University of Ferrara, in Translational Neurosciences and Neurotechnologies. The two Early Stage Researchers (ESR) of COBRA H2020 project (G.A. N. 859588), an EU Innovative Training Network of MSCA involving 9 partners, which will train a group of 15 researchers that will be the next generation of researchers to accurately characterize and model the linguistic, cognitive and brain mechanisms deployed by human speakers in conversational interactions with human interlocutors as well as artificial dialog systems.

Contact: alessandro.dausilio@iit.it

ERS2: When people are engaged in meaningful social interaction, they automatically and implicitly adjust their speech, vocal patterns and gestures to accommodate to others. Although these processes have extensively been explored at the behavioral level, very little is known about their neural underpinnings. Prior investigations have shown that suppression of alpha oscillations, overlaying sensorimotor regions, are a possible marker of action-perception coupling during non-speech (Tognoli & Kelso, 2015) and speech based (Mukherjee et al., 2019) interactive tasks. The project, by running dual-EEG recordings, will investigate if behavioral speech alignment translates into identifiable brain oscillatory markers. Key objectives are (i) to develop and validate metrics to quantify phonetic accommodation during natural speech interactions and (ii) to identify electrophysiological markers of between-speaker convergence.

ERS10: For adults, mastering the segmental and supra-segmental aspects of a second language (L2) is particularly challenging. Although we know that such a capability is partially maintained during adulthood, we do not know yet how to facilitate effective and long-lasting L2 learning. This project is based on the hypothesis that when people engage in meaningful social interactions, they automatically and implicitly align at multiple levels (Pickering, Garrod, 2013), including the phonetic (Mukherjee et al., 2019) and the facial expression ones. ESR10 will tackle the fundamental scientific question of speech alignment in L2 and whether it drives long-lasting improvements in L2 skills. Key objectives are (i) to investigate the dynamics of alignment in L2 (English) and (ii) to quantify improvements when participants are engaged in a conversation with native speakers.

THE APPLICATION IS DONE VIA THE COBRA WEBPAGE: https://www.cobra-network.eu/

A one-year position as a postdoctoral fellow is available at the AVIRES Lab ( https://avires.dimi.uniud.it ) of the Department of Mathematics, Computer Science and Physics (DMIF), University of Udine.   

The postdoc fellow will work with an interdisciplinary team of experienced researchers in the fields of audio signal processing, computer vision, machine learning, robotics. The goals of his research will be to develop specific methods and algorithms allowing to improve the effectiveness of localization and tracking of a moving target through the dynamic reconfiguration of a sensor network, to study the integration possibilities of acoustic and optical sensors with respect to the problem under study, to address application scenarios related to the use of mobile robotic platforms, in particular aerial drones equipped with acoustic and optical sensors. 

Details here.

Application deadline is March 17, 2020 at 02:00 p.m. (Italian time).

Please refer to the links below or contact Carlo Drioli (carlo.drioli@uniud.it) for further questions.

Detailed information and directions to apply can be found at the following link: http://web.uniud.it/ateneo/normativa/albo_ufficiale/146-2020 

Project Title: Automatically detecting emotional expressions in dynamic group interactions from audio signals

The goal of the successful candidate is to design signal processing and machine learning algorithms to automatically detect emotional expressions (individual affect and group mood) from recorded audio data. Challenges include speaker localization and diarization (often with overlapping speech), identification of discrete behaviors (e. g. laughter), and expansion to higher, more abstract levels of socioemotional behaviors (e. g., verbal expressions of support or disagreement) in order to detect convergent affective phenomena and emergent group mood.

This project is part of the interdisciplinary research group "Mechanisms of Change in Dynamic Social Interactions", which integrates fundamental science, innovative methods, and applications in psychology and computer science.

Please find the full job announcement with all details here

https://www.inf.uni-hamburg.de/en/inst/ab/sp/job-offer.html

64888 - CSIRO Postdoctoral Fellowship in Machine Learning and Artificial Intelligence: Constraints. (2 Positions)

• Do you have experience in Machine Learning and Artificial Intelligence?
• Work with world class researchers to solve the world’s biggest challenges
• Join CSIRO’s Data61, the largest data innovation group in Australia

CSIRO Early Research Career (CERC) Postdoctoral Fellowships provide opportunities to scientists and engineers who have completed their doctorate and have less than three years relevant postdoctoral work experience. These fellowships aim to develop the next generation of future leaders of the innovation.   The Machine Learning and Artificial Intelligence Future Science Platform (MLAI FSP) will build an exciting new research portfolio to leverage CSIRO’s deep domain expertise and experience. As a member of the Platform team, you will work with top CSIRO scientists and engineers to develop new machine learning and artificial intelligence methods with a specific emphasis on solving significant science questions. Together we will build the next generation of science tools using high performance computing infrastructure and cloud technologies to underpin the next generation of Australian science.   You will sit within the Constraints activity area within the MLAI FSP, which investigates MLAI models with design constraints, for example scalability, uncertainty propagation and privacy. The role has two main potential implementations which are not necessarily exclusive: design of new ML algorithms that are privacy compliant, and analysis of the privacy leakage of ML algorithms.    Your duties will include:

• Investigate, design and evaluate formally new ML algorithms for secure learning and/or the potential leakage of the state of the art with fixes. 
• Implement these methods efficiently using programing tools such as R or Python. 
• Carrying out evaluation of the developed software to demonstrate its competitiveness and fitness for purpose. Taking responsibility for functionality, performance and robustness.
• Carry out high impact research of strategic importance to CSIRO, with the aim of achieving innovative and wide-reaching scientific outcomes and ideas for further research.

Location:      Eveleigh, Sydney  Salary:          AU$83k - AU$94k plus up to 15.4% superannuation   Tenure:         Specified term of 3 years  Reference:    64888   To be considered you will need: Under CSIRO policy only those who meet all essential criteria can be appointed.

1. A doctorate (or will shortly satisfy the requirements of a PhD) in a Platform-relevant discipline area, such as machine learning, artificial intelligence, computer science, statistics, privacy/crypto or applied mathematics.
   1. Please note: To be eligible for this role you must have no more than 3 years (or part time equivalent) of postdoctoral research experience.
2. Solid fundamental and applied knowledge of machine learning and statistics or privacy
3. Demonstrated ability to understand and develop mathematically-founded machine learning algorithms and their development in toolkits. 
4. High level computational and programming skills (in Python, R, or C++) to build machine learning models and conduct analyses. 
5. High level written and oral communication skills with the ability to effectively represent the research team internally and externally, including publishing in peer reviewed journals and/or authorship of scientific papers, reports, and presenting at national and/or international conferences.
6. A record of science innovation and creativity, including the ability & willingness to incorporate novel ideas and approaches into scientific investigations, preferably across diverse and inclusive teams.

The successful applicant may be required to obtain and provide a National Police Check or equivalent.   For more information about this role please view the Position Description   Flexible Working Arrangements We work flexibly at CSIRO, offering a range of options for how, when and where you work.  Talk to us about how this role could be flexible for you.  Balance   Diversity We are working hard to recruit diverse people and ensure that all our people feel supported to do their best work and feel empowered to let their ideas flourish.  Diversity and Inclusion Strategy   About CSIRO At CSIRO we solve the greatest challenges through innovative science and technology.  We pride ourselves on hiring the best talent – bold change-makers, imaginative problem solvers and people driven by impact, whose creativity and skills match their enthusiasm for science and innovation.   Join us and start creating tomorrow today!   Apply Online To apply online, please provide a CV or resume and cover letter outlining your suitability and motivation for the role.   Applications Close Sunday, 8th March 2020 at 11.59pm AEST