Exploratory Research Projects (ERPs)
A Call for Exploratory
Research Projects is currently OPEN.
For additional questions, email us at research@cmuportugal.org
The CMU Portugal Program supports regularly the launch of Exploratory Research Projects (ERPs) with the main objective of promoting Portugal’s international competitiveness and innovation capacity in Science and Technology (S&T) in Information and Communication Technologies (ICT).
ERPs are selected through Exploratory Research project Calls, funded by the Fundação para a Ciência e a Tecnologia (FCT) for the period of 12 (twelve) months.
ERPs 2022
The CMU Portugal Program supports under the 2022 call, 6 Exploratory Research Projects (ERPs) designed to assist teams of researchers from Portuguese institutions, Carnegie Mellon University and industry partners, to bootstrap high-impact potential research activities of strategic relevance for the Carnegie Mellon Portugal Program.
Documents of the 2022 Call for ERPs:
– Announcement of the Call by the Portuguese Fundação para a Ciência e a Tecnologia
– Terms of Reference for the Exploratory Research Projects 2021 Call for Proposals
Principal Investigator in Portugal: Evgheni Polisciuc
Principal Investigator at CMU: Dominik Moritz
CMU Department: Human-Computer Interaction Institute
Principal Contractor in Portugal: Universidade de Coimbra
Keywords: Evaluation of Visualizations; Machine Learning; Automatic Visualization; Information Visualization
Summary: In the modern digital world, making information accessible is one of the key factors for success. In any sector, the quality and quantity of valuable data are differentiating aspects for the competitiveness of the companies and institutions. In the private sector, this presents an opportunity to gain an advantage, while in the public sector, the data is employed to make the quality of life better. Enhanced data analysis can empower users to make informed decisions, derive insights, and discover unknowns.
This project proposes to (i) devise novel methods for the automatic evaluation of graphics based on machine learning; (ii) apply these methods to compare and assess the quality of visualizations; (iii) evaluate the effectiveness of the developed models. It is planned to develop and study ML models that are designed specifically for unique characteristics of visualizations, including the aspects that are not related to data representation. This will provide the basis for further generalization of automatic visualizations, expanding the spectrum of visualization designs. Ultimately, it will push boundaries towards an automatic visualization that goes beyond a traditional expression of underlying data but considers the aspects of graphics that are relevant for humans, such as comprehensively, aesthetics, or creativity.
Principal Investigator in Portugal: Cristina Mendes Santos
Principal Investigator at CMU: Mayank Goel
CMU Department: Software and Societal Systems Department
Principal Contractor in Portugal: Associação Fraunhofer Portugal Research
Partner Institutions: Centro Hospitalar Universitário de S. João, Instituto Português de Oncologia do Porto Francisco Gentil.
Keywords: Artificial Intelligence;Digital Phenotyping;Internet Interventions;Breast Cancer Survivors
Summary: Despite the efficacy of psychosocial interventions in minimizing psychosocial morbidity in breast cancer survivors (BCS), the delivery of Interventions is limited by physical, organizational, and individual barriers, which contribute to a mental healthcare treatment gap in cancer settings. Digital Phenotyping enhanced Internet Interventions may provide remarkable opportunities to overcome these limitations. By enabling the dynamic collection and assessment of multimodal data, these interventions may be used to refine diagnostic processes, improve condition monitoring for actionable outcomes, such as early signs of relapse, and tailor interventions, configuring a disruptive healthcare delivery model.
However, limited research has been conducted on translating digital phenotyping signals into clinically actionable digital phenotypes or prediction models capable of better explaining, assessing, and tailoring internet interventions to BCS. Moreover, little is known about their acceptability, feasibility, and efficacy in BCS. This proposal intends to bridge these research gaps by applying a Machine Learning approach leveraging data from smartphones and fitness trackers in BCS to predict their health outcomes while undergoing an internet intervention named iNNOV Breast Cancer (iNNOVBC).
Principal Investigator in Portugal: Maria Rute André
Principal Investigator at CMU: Maysam Chamanzar
CMU Department: Department of Electrical and Computer Engineering
Principal Contractor in Portugal: Universidade de Aveiro
Partner Institution: Instituto de Telecomunicações
Keywords: Neural probes; Photonics; Hydrogel; Neuromorphic computing
Summary: Optical methods have been widely used for stimulation and functional imaging of neural circuits, namely, optogenetics which is a powerful tool for selective excitation or inhibition of specific cell types using light of different wavelengths desirable to stimulate or inhibit a subset of neurons that express the same light-sensitive proteins. However, optical stimulation and imaging of neurons deep in the brain require implantable optical neural probes for simultaneous stimulation and recording of neural circuits. These neural probe platforms, monolithically integrating the optical signal sources, such as LEDs, and the recording electrodes on a flexible polymer substrate have been already shown. However, these solutions have a disadvantage related to the heat dissipation occurring at the electrical circuit level, which must be carefully controlled to avoid brain damage. This requires delivering patterns of light into brain tissue with high spatial resolution, using biocompatible waveguides, that will displace the light source from the vicinity of the brain, yet the conventional methods based on external optics for light delivery are limited to superficial layers of the tissue because of absorption and scattering.
The PHEASANT project proposes to develop flexible neural probes integrated with monolithic sensors to study brain activity, at the level of neurons. Moreover, this project will be able to optimize procedures and devices, without losing track of relevant issues, creating economical value with useful products that might be exploited.
Principal Investigator in Portugal: Jorge Fernandes
Principal Investigator at CMU: Marc Dandin
CMU Department: Electrical and Computer Engineering
Principal Contractor in Portugal: INESC ID
Keywords: mm-size implant; Deep implant; Bioelectronic implant; Electroceuthicals
Summary: Bioelectronic Medicine (BEM) is at the intersection of scientific disciplines and has experienced a huge development in the past 5 years. BEM can revolutionize how medicine is practice by favoring the use of electroceuticals to interface with the nervous system instead of drugs. The advances in genomics, allowing better knowledge of cellular behavior; and in microelectronics, allowing lower size and lower power consumption; create the conditions for advanced electroceuticals development. Subdermal and deep mm-size implants are the object of this project which is expected to investigate state-of-the-art integrated circuits (ICs) suitable for electroceuticals, comprising power management, analog and/or digital signal processing, wireless data transfer, and nerve interfaces.
The goal is to design a microelectronic implant circuit using state-of-the-art techniques at circuit level to obtain a simple implant with extremely low-power consumption (<1mW), capable of harvesting its’ own energy, having enough communication capabilities to receive commands and send status, and finally, being capable of stimulating nerves.
Principal Investigator in Portugal: Luís Pedrosa
Principal Investigator at CMU: Srinivasan Seshan
CMU Department: Computer Science Department
Principal Contractor in Portugal: INESC ID
Keywords: Programmable Networks;Network Synthesis;Network Optimization
Summary: Network Operators such as Altice, MEO, and NOS deploy a wide range of network functions (NFs) to benefit their customers. For example, firewalls and anomaly detectors monitor network traffic to prevent attacks. Traffic monitors account for network usage for billing; web caches keep copies of frequently accessed images and files to allow clients to view them faster. Network functions are a core component of the 4G Evolved Packet Core (EPC) and are a core component of 5G as well. They are widely deployed throughout today’s cloud and edge data-centers. In short, NFs are crucial for the security, performance, and evolution of networks.
The goal of this project, called SALAD-Nets, is to allow developers, operators, and even cloud customers to design a virtual network of software NFs, and for a synthesizer to map this large-scale distributed network onto the underlying network substrate, as deployed in the data-center. Taking into account each device’s capabilities and limitations, SALAD-Nets partitions and optimizes the network functionality, generating the code that powers not only the myriad devices in the network, but also the controller that orchestrates it all to realize on the physical network fabric the same functionality that was specified in the virtual network topology.
Principal Investigator in Portugal: Ricardo Melo
Investigator at CMU: Sarah E. Fox
CMU Department: Software and Societal Systems Department
Principal Contractor in Portugal: Associação Fraunhofer Portugal Research
Partner Institutions: Universidade do Porto – Faculdade de Letras da Universidade do Porto.
Keywords: Values; Health; Participatory Design; Human-Centred Design
Summary: Underlying all technology design are values (e.g., efficiency, empathy, accountability, transparency), which may be conspicuous in the designed artefact. In healthcare, there is a growing number of digital technologies which collect and display patients’ data, such as health monitoring and self-tracking. The design of visualizations for these data also holds a set of values. When these values—those of the design and knowledge, and those of patients—are not aligned, this may lead to a lack of representation, inclusiveness, or trust in the data. In chronic diseases, digital technology and data visualization are growingly available and may play a critical role in supporting self-care and decision making.
Following the principles of Human-Centered Design, project Signo proposes to bring researchers in design, human-computer interaction, and philosophy together with patients and healthcare professionals in the context of ophthalmology to, collectively in co-creation, explore, experiment, prototype, and test technology and data visualizations which are built on a shared set of values, and understand how these values are communicated.
By applying methods of co-creation and participatory exploration, the project will generate a set of recommendations and a set of experimental prototypes, exploring the issue of values, the purpose of which is to support the design of more useful and effective technology in healthcare. As a contribution to society, such usefulness should translate into more appropriate health technology design, support clinical decision, better-informed patients, and, ultimately,better health for all.
ERPs 2021
The CMU Portugal Program supported under the 2021 call, 6 Exploratory Research Projects (ERPs) designed to assist teams of researchers from Portuguese institutions, Carnegie Mellon University and industry partners, to bootstrap high-impact potential research activities of strategic relevance for the Carnegie Mellon Portugal Program.
Documents of the 2021 Call for ERPs:
– Announcement of the Call by the Portuguese Fundação para a Ciência e a Tecnologia
– Terms of Reference for the Exploratory Research Projects 2021 Call for Proposals
Principal Investigator in Portugal: Nuno Santos, INESC-ID / IST
Principal Investigator at CMU:Nicolas Christin
CMU Department: Software and Social Systems Department
Principal Contractor in Portugal: INESC-ID/INESC/IST/ULisboa
Participant Institutions in Portugal: INESC TEC; NOVA.ID.FCT ; FCiências.ID
Keywords: Online Anonymity Networks – Network Traffic Analysis – Machine Learning – Privacy and Multiparty Computation
Summary: Cybercrime is escalating to unprecedented levels. Perpetrators often communicate on the Internet using highly sophisticated anonymization systems that allow them to thrive without being tracked by Law Enforcement Agencies (LEAs). Tor is by far the most popular of such systems. What makes Tor communications so hard to trace is that it relies on a large-scale network of servers – called relays – that employ advanced encryption and complex traffic obfuscation techniques. For this reason, although anonymous networks play a vital role on the Web for protecting user privacy and allowing for censorship-free access to information, they have also been used as the backbone of the so-called Dark Web, providing a key technological pillar sustaining the flourishing ecosystem of cybercrime.
Driven by our ultimate goal of building a practical cybercrime investigation tool for analyzing Dark Web traffic, this work will advance the state of the art on cutting-edge topics in privacy-preserving computation, machine learning, and “ethical-by-design” systems. By extending our early work, we will deliver a new prototype of our tool that will be able to efficiently process deanonymization queries in a privacy-preserving manner.
Principal Investigator in Portugal: José dos Santos, INESC-ID/INESC/IST/ULisboa
Principal Investigator at CMU: Limin Jia
CMU Department: Electrical and Computer Engineering
Principal Contractor in Portugal: INESC-ID/INESC/IST/ULisboa
Participant Institutions in Portugal: Instituto de Telecomunicações (IT)
Keywords: code injection vulnerabilities – dynamic taint tracking – symbolic execution – concolic execution
Summary: JavaScript is the de facto language for client-side programming and, with the advent of Node.js, has rapidly become one of the most popular languages for implementing server-side applications. Node.js code is not sandboxed, making it open to a broad range of security attacks. Among them, one of the most serious is injection attacks, which allows attackers to run arbitrary code on the targeted execution platform. Node.js has been used to build high-profile applications, such as Skype, Slack and WhatsApp, and thus, injection attacks on Node.js code can have serious consequences as they can lead to breaches of user data or be used as building blocks for more sophisticated attacks on a company’s network and servers. In this project, we plan to develop DIVINA: a new analysis tool for detecting injection vulnerabilities in Node.js applications. Our goal is for DIVINA to be both effective—with low false negative and false positive rates—and efficient—with low overheads—so that it can be integrated in standard code review pipelines. We will leverage the combination of dynamic taint tracking and dynamic symbolic execution. We aim to deliver a prototype implementation of the analysis tool and results on applying our tool to a set of curated Node.js packages.
Principal Investigator in Portugal: Mahmoud Tavakoli, Inst. for Systems and Robotics, Universidade de Coimbra
Principal Investigators at CMU: Lining Yao; Carmel Majid
CMU Department: Human-Computer Interaction Institute, Mechanical Engineering
Principal Contractor in PT: Instituto de Sistemas e Robótica (ISR)
Keywords: Computational design – Electrónica e Robótica Impressa – Additive Manufacturing – Soft Robotics
Summary: Exoskins – broadly defined as exoskeletons, exosuits, orthotics, and assistive wearable technologes – have already been proposed for medical treatments and healthcare, helping patients experiencing motor disorders to receive treatments or rehabilitate. However, since each wearer’s health condition and body shape ais different, it is challenging to custom fit to each patient. To this end, a novel medical exoskin solution is needed: one ideal medical exoskin may provide a light-weight actuation system, and on-demand tunable stiffnesses to meet the patients’ evolving medical and comfort requirements while being compact and affordable.
We propose to develop an AI-enabled computational design framework to develop patient-specific exoskins with variable stiffness robotic material in compliant mechanisms. Such exoskins will have tunable resistance (for physical rehabilitation or exercise), reconfigurable degree-of-freedoms and active actuation (for prosthesis). We plan to leverage data-driven methods to develop the material simulator and integrate it into an AI-enabled computational platform to guide the design and manufacturing of such devices. Such efforts will advance the social and economic impact of such functional systems in healthcare and medical devices and foster industry-science-engineering relationship
Principal Investigator in Portugal: Cláudia Soares, NOVA LINCS, Computer Science Department Nova School of Science and Technology
Principal Investigators at CMU: Haiyi Zhu; Yuejie Chi
CMU Department: Human-Computer Interaction Institute
Principal Contractor in Portugal: NOVA.ID.FCT
Participant Institutions in Portugal: Faculdade de Economia da UNL – Nova SBE (FE/UNL) ; IST-ID
Keywords: Health recommender systems – Medical referral – Interpretable Machine learning – Value-based care
Summary: MD2TRUST aims to develop a recommender system for referral of specialist care doctors to be used by primary care doctors, that is compatible with current referral practice, and that can transparently encourage organizational change, towards a more effective patient-centric healthcare management.
Typically, for a given patient with clinical needs, the primary care physicians can make a choice of several specialists to whom they may refer. As such, primary-specialty referral may affect many aspects of patient care, such as quality of care, patient satisfaction and healthcare costs, etc. Researchers recently leveraged the patient consultation history extracted from insurance claims data to construct the patient sharing network between physicians based on the shared patients. Essentially, the patient sharing network operationalizes an informal information-sharing network in which physicians provide care to shared patients. These current metrics derived from network science can serve as informative features to boost predictive model performance and optimize the health system for improved medical outcomes.
This project will analyze the current state of medical referral in CUF Health and compare it with another reality in the USA by implementing novel recommendation systems that are: 1) trustworthy, 2) advised by policy, 3) not disruptive of the usual referral process used today, 4)fostering online communities.
Principal Investigator in Portugal: Alexandre Ferreira da Silva, Universidade do Minho
Principal Investigator at CMU: Zachary Manchester
CMU Department: Robotics Institute
Principal Contractor in PT: Universidade do Minho (UM)
Participant Institutions in PT: Instituto Superior Técnico (IST/ULisboa)
Keywords: PocketQube – Research and Education – Open Access – Low Cost
Summary: The PROMETHEUS project aims at providing easy access to space for the research and education (R&E) community. Learning or researching about space can be more fruitful via hands-on projects. However, Space can seem like a pitfall environment with high risks and therefore high costs to develop projects at the student level. This barrier is critical and hinders R&E opportunities given the lack of experience in this environment. The team has identified that there is no current tool that enables easy and low-cost access to such space experiences of learning or manufacturing a simple satellite device.
The PROMETHEUS project aims at being a contributor for this momentum, by serving as an open-source PocketQube platform, which is a small satellite in a 5 cm cube form factor, an ideal size for R&E. From the foundation, this proposed platform has all the minimum required conditions to be easily integrated as a tool that facilitates hands-on activities and access to space. However, the PROMETHEUS goes beyond that and shares the entire pipeline for the satellite deployment as it will be licensed, certified, and launched. PROMETHEUS aims at giving access to space for everyone.
Principal Investigator in Portugal: Tiago Guerreiro, LASIGE / FCUL
Principal Investigator at CMU: Jodi Forlizzi
CMU Department: Human- Computer Interaction Institute
Principal Contractor in Portugal: FCiências.ID
Participant Institutions in Portugal: IST-ID
Keywords: HRI – Agency – Older Adults – Co-design
Summary: Aging of the population carries several concerns and challenges. There will be fewer young people to take care of the elder and waiting lists for nursing homes and other care facilities will grow. Most importantly, their wellbeing and sense of self-worth is at risk.
Previous work has looked at assistive robots as a way to allow people to live more independently at home, or keep active in a care home. However, they do not support people in achieving their own goals. This project wants to empower them by transferring agency to older adults in human-robot interaction. One speculative example would be to place a robot in a grandson’s home controlled by his grandmother while sitting in her chair in a care home: she would look at the toddler, have a chat about school, and offer milk and cookies through this skillful proxy. How can robots support older people in achieving their wishes and how can they control or program a robot to perform tasks? The project’s vision is to advance research on human-robot interaction to facilitate a new type of relationship between elders and assistive robots.
However, shifting agency to older adults brings several challenges to the control of those robots. The project will combine collaborative research & design with elders and caregivers, iterative system development, and iterative evaluation through controlled field studies by leveraging existing relationships with Portuguese elder care facilities.
ERPs 2019
The CMU Portugal Program supported under the 2019call, 7 Exploratory Research Projects (ERPs) designed to assist teams of researchers from Portuguese institutions, Carnegie Mellon University and industry partners, to bootstrap high-impact potential research activities of strategic relevance for the Carnegie Mellon Portugal Program.
Documents of the 2019 Call for ERPs:
– Announcement of the Call by the Portuguese Fundação para a Ciência e a Tecnologia
– Terms of Reference for the Exploratory Research Projects 2019 Call for Proposals
Principal Investigator in Portugal: Ana Paiva
Principal Investigator at CMU: Louis-Phillippe Morency
Participant institutions: Instituto de Engenharia de Sistemas e Computadores, Investigação e Desenvolvimento em Lisboa, Instituto Superior Técnico, Universidade de Lisboa; ISCTE – Instituto Universitário de Lisboa; Language Technologies Institute – Carnegie Mellon University.
Summary: Humor has been linked to numerous positive outcomes and laughter is thought to decrease anxiety. From a psychological standpoint, the employment of humor can lead to strategies that help individuals deal with stressful or traumatic situations. Humor has also been used in therapy, as it has been linked to positive outcomes. The AGENTS project is leveraging on the power of humor to create more naturalistic and lifelike interactions with social embodied agents, in particular, social robots. To achieve this goal, a top-down approach of humor will be employed that can be modeled to match each user’s preferences.
The project uses a 2×2 conceptualization of humor that involves its social function (humor used to enhance oneself or used to enhance others) and the valence of the humoristic content (positive, negative). By using such conceptualization, the goal is to create a dataset of humorous jokes or stories and through the application of supervised machine learning techniques, to extract and automatize multimodal humor delivery.
The end-goal of this process will be the implementation of user personalized humoristic interactions in the context of a group card game involving more than one human and more than one social agent. This is expected to lead to better interaction outcomes and increase the value perception of the agent, by contributing to greater task enjoyment, improved perception of the agent and greater intention to interact again with these social agents in the future.
Project Website: https://www.agents-humor.com/
Principal Investigator in Portugal: Stanislav Maslovski
Principal Investigator at CMU: Sheng Shen
Participant institutions: Universidade de Aveiro; Instituto de Telecomunicações Aveiro; Department of Mechanical Engineering at Carnegie Mellon University.
Summary: Metasurfaces are versatile tools for controlling wave fronts and performing nearly-instantaneous operations on the angular spectrum of propagating electromagnetic waves. In this project, we investigate the previously unexplored possibility of realizing hybrid intelligent beamforming systems comprising the programmable metasurfaces (the quasi-optical hardware) and the algorithmic (the silicon/software) layers. In order to allow for fully flexible and dynamic control over the antenna radiation pattern, this new architecture borrows a few key ideas from the areas of AI and neural networks, such as the AI network training methods.
This architecture has a potential to realize real-time dynamic channel propagation estimations and adaptive beamforming using a trained AI network that incorporates the programmable metasurfaces as an integral part of such network. As a novel and exploratory concept, this project is centered on the feasibility studies and numerical simulations associated with the proposed architecture and its possible applications under various beamforming scenarios. We aim to identify the potential of this new technology for future telecommunication systems, estimate the performance gains, and make recommendations on the use of the programmable metasurfaces for the mentioned applications.
Principal Investigator in Portugal: Kai Li
Principal Investigator at CMU: Pei Zhang
Participant institutions: Instituto Superior de Engenharia do Porto, Instituto Politécnico do Porto; Faculdade de Engenharia da Universidade do Porto; Department of Electrical and Computer Engineering – Carnegie Mellon University.
Summary: The CRUAV project investigates cooperative Unmanned Aerial Vehicles (UAVs) for real-time surveillance of traffic violations, accidents, or other road emergencies in intelligent transportation systems. Specifically, many Internet-of-Things (IoT) nodes (e.g., lightweight cameras and portable tachymeters) are deployed on roadside to take videos or images and collect environmental data. Every IoT node is equipped with solar panels or wind power generators to harvest energy to power its operations. In particular, the battery energy of the IoT nodes can be drastically different from each other, depending on the ambient environmental conditions of the individual nodes. Each IoT node generates its data at an application-specific sampling rate and buffers the data awaiting transmission. The UAV equipped with a wireless transceiver and an onboard processor is instructed to fly over an unpopulated area with little to no 5G service but a need for bursty transmissions of high-bandwidth data, e.g., high-resolution images or videos.
The real-time and secure data aggregation achieved by the proposed frameworks in this project enables a variety of innovative applications spanning both the commercial and official sectors: providing road users with the latest information on traffic and monitoring possible traffic violations, for example. The outcome of the project will be systematically evaluated for airborne surveillance of intelligent transportation.
Principal Investigator in Portugal: João Fernando Ferreira
Principal Investigator at CMU: Nicolas Christin
Participant institutions: Instituto de Engenharia de Sistemas e Computadores, Investigação e Desenvolvimento em Lisboa, Instituto Superior Técnico, Universidade de Lisboa; Inesc Tec – Instituto de Engenharia de Sistemas e Computadores, Tecnologia e Ciência; Department of Computer Science- Carnegie Mellon University.
Summary: PassCert’s short-term vision is to build an open-source, proof-of-concept password manager that through the use of formal verification, is guaranteed to satisfy properties related to data storage and password generation, since these were identified as barriers to adoption and effective use of PMs . While text passwords are one of the most used security mechanisms, users fail to use them effectively and safely. To combat this, experts recommend the use of Password Managers (PMs) to help users generate and manage their passwords. However, their adoption is low as users do not trust PMs. Formal verification can provide strong assurances, making software more reliable. Therefore, the expectation is that a PM with formally verified features will increase users’ trust and, consequently, their adoption of PMs. This can help address many of the existing security problems regarding password authentication.
Project website: https://passcert-project.github.io
Principal Investigator in Portugal: Isabel Trancoso
Principal Investigator at CMU: Bhiksha Raj Ramakrishnan
Participant institutions: Instituto de Engenharia de Sistemas e Computadores, Investigação e Desenvolvimento em Lisboa, Instituto Superior Técnico, Universidade de Lisboa; Language Technologies Institute – Carnegie Mellon University.
Summary: The growing number of Machine Learning as a Service applications has caused an increasing awareness of their potential to compromise users’ privacy, as shown by the intense debate around the GDPR. Among other data types, a large amount of information may be extracted from speech going far beyond linguistic contents. This extra information includes demographic traits like gender, age range, level of education, emotional status, personality traits, levels of stress, intoxication, sleepiness, etc. and even cues about diseases that affect speech. This implies that one should regard speech as “Personally Identifiable Information”.
Current machine learning models can remotely transcribe speech recordings, identify speakers, and perform “diarization”, often referred to as the problem of determining “who spoke when” in a conversation. Privacy in speech processing is the overarching topic of the PRIVADIA project, with a particular focus in diarization. The main challenges are in combining state-of-the-art speaker representations or embeddings with cryptographic techniques. The project also explores alternative approaches to privacy based on deep learning speech manipulation techniques.
Project Website: https://privadia.hlt.inesc-id.pt/
Principal Investigator in Portugal: Rui Maranhão
Principal Investigator at CMU: Hakan Erdogmus
Participant institutions: Instituto de Engenharia de Sistemas e Computadores, Investigação e Desenvolvimento em Lisboa, Instituto Superior Técnico, Universidade de Lisboa; Department of Electrical and Computer Engineering – Carnegie Mellon University.
Summary: Software vulnerabilities lead to massive financial losses for software companies because of business disruption, loss of privacy, reputational damage, legal implications, and life-threatening situations. For instance, in 2014, an Apple bug (‘goto fail’) in a widely used SSL implementation caused applications to accept invalid certificates. Although several success stories exist, there are also several concerning limitations hindering wide adoption: viz. high number of false positives and warnings that are innocuous and difficult to act upon. Continuous Integration (CI) is an increasingly popular practice among modern development teams however, due to the overwhelming amount of information generated by all of these phases and tools, software engineers feel that some of the production phases are frustrating and tend to ignore valuable output. Following the CodeAware [1] vision (CodeAware is sensor-based fine-grained monitoring and management of software that can easily be integrated into the CI pipeline), the project proposes the development of a novel framework for automatically and efficiently detecting security issues that can be integrated with confidence on the CI pipelines through the implementation of more fine-grained, more unified and faster approaches to CI static analysis.
Principal Investigator in Portugal: Luís Pedrosa
Principal Investigator at CMU: Justine Sherry
Participant institutions: Instituto de Engenharia de Sistemas e Computadores, Investigação e Desenvolvimento em Lisboa, Instituto Superior Técnico, University of Lisbon; FCiências.ID – Associação para a Investigação e Desenvolvimento de Ciências; Department of Computer Science- Carnegie Mellon University.
Summary: When implementing network functions (NFs), developers are often confronted with a choice: implement the NF in software and face the challenge of performance or use one of a wide variety of programmable networking devices, such as programmable switches and SmartNICs, to trade-off some flexibility for the ability to process packets at full line rate. With SyNAPSE we ask the question “Why not have both?”. The project offers a synthesis based approach to automatically generate accelerated implementations of Software NFs, using smart network devices whenever possible to increase performance. The SyNAPSE project aim is to advance the state of the art in building programmable networks using synthesis. Towards this end, a prototype synthesis engine will be built that takes as input an NF implementation mapped onto a network and synthesizes a network controller and any requisite switch programs that combined produce equivalent behavior. The project will further build the infrastructure needed to design, develop, and evaluate the prototype, while also laying the groundwork for longer-term collaboration.
Project Website: https://synapse.inesc-id.pt/
ERPs 2017
The CMU Portugal Program supported under the 2017 call, 8 Exploratory Research Projects (ERPs) with the main objective of promoting Information and Communication Technologies projects in strategic emerging areas.
Documents of the 2017 Call for ERPs:
– Announcement of the Call by the Portuguese Fundação para a Ciência e a Tecnologia
– Terms of Reference for the Exploratory Research Projects 2017 Call for Proposals
The following projects were selected for their their international competitiveness potential and for enabling technologies in different application sectors but also for contributing in some measure to the Atlantic International Research Center (AIR Center) initiative:
Principal Investigator in Portugal: Fernando José da Silva Velez
Principal Investigator at CMU: Jon Peha
Proponent Institution: Instituto de Telecomunicações (IT) (UBI)
Partner Institution: Universidade de Coimbra (UC)
Department at CMU: Electrical and Computer Engineering
Summary: CONQUEST addresses 5G technology and is an exploratory project that investigates aspects of spectrum sharing in both terrestrial cellular scenarios and heterogeneous networks (HetNets) with drone small cells.
It is a research collaboration between Prof. Jon Peha team in CMU, and two main Portuguese research teams, one from Instituto de Telecomunicações (IT), led by Prof. Fernando Velez and another from University of Coimbra (CISUC), led by Prof. Edmundo Monteiro. The CMU team will mainly contribute with knowledge on spectrum sharing and public policy in the management of spectrum. The contributions from the Portuguese teams will include optimizing the economic trade-off of HetNets by considering shared spectrum in carrier aggregation, business aspects, opportunistic spectrum access, simulation, security issues and hardware testing of real HetNets in the ORCIP platform (http://www.orcip.pt).
Principal Investigator in Portugal: Maria Inês Camarate de Campos Lynce de Faria
Principal Investigator at CMU: Ruben Martins
Proponent Institution: Instituto de Engenharia de Sistemas e Computadores, Investigação e Desenvolvimento em Lisboa (INESC ID/INESC/IST/ULisboa)
Partner Institution: IMAR – Instituto do Mar (IMAR)
Department at CMU: Computer Science
Summary: Knowing the link between species and their habitats is essential for the sustainable management of the Exclusive Economic Zone of the Azores. The DeepData project will build a computational infrastructure whose purpose is to serve the scientific community and provide information to policy-makers and the general public. MARE Açores researchers will provide deep sea data of the Azores sea and predictive models for species distribution. The INESC-ID Lisboa researchers will build a database that will incorporate data from the Azores sea from different sources. Researchers at Carnegie Mellon University (CMU) will apply automatic learning techniques to available data to make predictive models more accurate. INESC-ID and CMU will collaborate in the application of formal methods to automate data analysis.
Principal Investigator in Portugal: Pedro Sanches Amorim
Principal Investigator at CMU: Nicolas Christian
Proponent Institution: Faculdade de Engenharia da Universidade do Porto (FE/UP)
Partner Institution: INESC Tecnologia e Ciência (INESC TEC)
Department at CMU: Computer Science
Summary: The blockchain in itself is a very simple apparatus: an encrypted linked list, often referred to as a distributed ledger. Yet, recent developments turned it into a safe, trustable, decentralized, and immutable chain of encrypted transactions, opening up a whole new realm of applications once deemed unfeasible. Notwithstanding its success in digital (or crypto) currencies, its usefulness is underlined by its potential of application in diverse fields. One such problem is coordinating a supply chain. This project aims to be a proof of concept of a blockchain tailored to address supply chain coordination, specifically worldwide procurement supported by a decentralized network. Consider an (optionally) anonymous and distributed eBay with automatic contract enforcement rules that can be used by any supplier to signal their product availability or bid for an order.
Principal Investigator in Portugal: Hugo Alexandre Paredes Guedes da Silva
Principal Investigator at CMU: Jeffrey Bigham
Proponent Institution: Universidade de Trás-os-Montes e Alto Douro
Partner Institutions: FCiências.ID – Associação para a Investigação e Desenvolvimento de Ciências (Fciências.ID)
Department at CMU: Computer Science
Summary: In current climate research of extreme phenomena algorithms often fail to identify and track consecutive occurrences, requiring the development of specific algorithms to handle these events. However, for scientists, separating consecutive occurrences of phenomena is a relative simple task. This task can be outsourced to an expert crowd and algorithms could still be used, with clear benefits for the computational power required. The eCSAAP – expert Crowdsourcing for Semantic Annotation of Atmospheric Phenomena – project aims to explore the usage of expert crowdsourcing for annotating this kind of systems, so automated methods and computational resources can be optimized in a future hybrid approach. The project will allow information sharing and the development an interdisciplinary research ground, according to the strategic guidelines of the recently created AIR Centre.
Principal Investigator in Portugal: Manuel Ricardo de Almeida Rodrigues Marques
Principal Investigator at CMU: Manuela Veloso
Proponent Institution: Associação do Instituto Superior Técnico para a Investigação e o Desenvolvimento (IST-ID)
Partner Institutions: MITI – Madeira Interactive Technologies Institute; Associação (Madeira-ITI)
Department at CMU: Computer Science
Summary: Nowadays, many people with Cerebral Palsy, Alzheimer’s disease and other degenerative diseases do not have motor coordination to feed themselves autonomously. The goal of the Feedbot project is to develop a portable robot arm so that people with severe motor disabilities can eat independently. Existing solutions have preprogramed movements, requiring users to adapt to them. The proposed robot arm learns the behaviours of each user and adapts the movements during each meal. Its portability is also a very important feature since enables its daily use, at home or in the office, as well as in any restaurant, whether the user is alone or with others. This device will thus contribute to a great increase in the autonomy of people with motor disabilities.
Project website: http://users.isr.ist.utl.pt/~manuel/FeedBot/
Principal Investigator in Portugal: Paulo Sérgio Duque de Brito
Principal Investigator at CMU: Inês Azevedo
Proponent Institution: Instituto Politécnico de Portalegre (IPPortalegre)
Partner Institution: Universidade de Aveiro (UA)
Department at CMU: SEES
Summary: This project guides our team towards the investigation of optimized gasification solutions in order to accomplish the following strategic opportunities:
– Enhance bioenergy´s value proposition by identifying suitable biomass/municipal waste blends that can ensure a sustainable demand for gasification facilities.
– Determine an optimal decision-making process across multiple scenarios and future constraints deploying a risk analysis based on technical and economic parameters.
Principal Investigator in Portugal: Luis Filipe Coelho Antunes
Principal Investigator at CMU: Lujo Bauer
Proponent Institution: Faculdade de Ciências da Universidade do Porto (FCUP)
Partner Institutions: Universidade de Aveiro (UA) ; Associação Porto Digital (APD)
Department at CMU: Electrical and Computer Engineering
Summary: Given the exponential growth of Internet of Things (IoT), its diversity, and the seamless and heterogeneous nature of communications, we are faced with many prominent challenges in terms of management, interoperability, security, and privacy. Also, when the IoT matures, it is likely that most connected devices will be invisible to us so we should tackle these problems before it is too late. We plan to develop a middleware platform to autonomously combine different privacy-preserving techniques in order to secure and properly manage the identity of the things/sensors, secure its communications and empower the user regarding the data generated by these objects. By doing so, we will contribute to solve the existing tension between legislation and technology.
Principal Investigator in Portugal: Kazi Mohammed Saidul Huq
Principal Investigator at CMU: Douglas Sicker
Proponent Institution: Instituto de Telecomunicações (IT)
Partner Institution: Universidade de Aveiro (UA)
Department at CMU: Engineering and Public Policy
Summary: What will a beyond 5G network look like? It is too early to define this with any certainty. However, it is widely agreed that in contrast to 5G networks, beyond 5G network should achieve greater system capacity (>1000 times) than the 5G network. It is widely accepted that there are three major ways to obtain several orders of this magnitude in throughput gain, those being: extreme densification of infrastructure, large quantities of new bandwidth, and many more antennas, allowing a throughput gain in the “spatial” dimension. In reality, these processes are complementary in many respects. For example, in the search for more bandwidth, we are moving toward higher frequencies, especially in the promising THz wavelength spectrum. These high frequencies require the support of more antenna and smaller cells to overcome propagation loss effect. Toward making THz a reality for broadband communication, this proposal will address two important issues (i) 3D channel models for broadband THz, and (ii) novel techniques for modeling the effects of mobility in broadband THz. To check the accuracy of these models, we will first propose to develop new mathematical tools, which we will then validate via an already available system level simulator under realistic scenarios.
