Investigate machine learning algorithms to detect anomalies incomputing infrastructures in real-time

Metafor is developing a new class of IT system management solution to monitor computer and application activities, and alert when anomalous behavior occurs. Current commercial tools for anomaly detection use simple statistical rules and thresholds to detect anomalies. These methods are failing for today’s dynamic cloud environment where change is constant. As a result, IT operators are flooded with false alerts; become overwhelmed with alert fatigue and learn to ignore the alerts.

Real-time fatigue detection by remote eye tracking

Fatigue is a significant risk factor for operating moving vehicles in work settings that involve long duty hours or shift work rotations. A critical objective in fatigue management is to develop technology to establish fitness for duty, monitor fatigue in real time and predict performance failure. Eye movements (eyelid closure, blink rate and duration) are sensitive indicators of fatigue. Six Safety Systems has developed an eye tracking system that can monitor eye movements remotely.

Visual History for Scheduling Behaviour

Actenum produces scheduling software (Actenum DSO) that computes project schedules based on user defined criteria, constraints and desired KPIs (Key Performance Indicators). The software represents schedules using Gantt charts. A user of Actenum DSO enters activities and the constraints between them and then runs an optimization algorithm to produce a schedule. Actenum has found that their customers have challenges understanding what their schedule optimization algorithm is doing.

Application of neuroimaging to optimize observational learning from a mobile application: influences of age and brain function

One side effect of stroke is damage to the left parietal lobe, important in motor control of the right hand. Many stroke patients experience loss of motor skills and have difficulty relearning how to perform daily tasks. Recent reports from stroke rehabilitation centres have shown increased improvements in relearning basic motor skills upon incorporation of observational learning techniques into their therapies.

Closing the Gap of Write- vs. Read-intensive Data Stores Using SSDs Year Two

Many big data challenges are characterized not only by a very large volume of data that has to be processed but also by a high data production rate. In this project, new storage approaches for big data will be explored. Key point is the efficient use of modern hardware, especially modern storage technology such as SSDs. These new technologies have highly improved performance in comparison to traditional hardware. However, classical data structures and algorithms can not directly be applied due to the different characteristics of these devices.

Disaster Recovery as a Green Cloud Service: Balancing Risk, Cost, and Carbon

Unlike centralized computing, which is typically performed in a single data-center, Cloud computing enables the computation to be spread across multiple geographically distributed data-centers which are abstracted as a single system by the Cloud management layer. This computational model enables disaster recovery (DR) by re-establishing the services provided by a data-center affected by the disaster in another healthy data-center capable of hosting the applications providing these services.

Employing Network Analysis and Mining for Effective Crowdsourcing Market Research and Reporting

The main aim of the project is to develop the characteristics of a Prosumer, i.e. a consumer that interacts on-line with a company providing useful and insightful comments on its products. Data from consumer interaction databases are will be analyzed using the tools of data mining in order to accomplish the goals of the project.

Learning representations of customer behavior to drive actionable insights in e-commerce

Rubikloud enables online businesses to turn their data into revenue by turning the focus from visualization and interpretation to driving smarter data-driven decisions. A key challenge to leveraging the advances in machine learning research and development is the nature of event-based data encountered in this domain: clicks, purchases, impressions, and conversions. Machine learning techniques typically operate on fixed-length vector representations of data, for example, collections of attributes, images, and word counts of text documents.

Plasticity and Fit Brains in Mild Cognitive Impairment

Worldwide, one new case of dementia is detected every four seconds. At this point in time, no effective drug therapy exists for cognitive impairment and dementia. As a result, there is much interest in lifestyle approaches. Specifically, complex mental activity, such as cognitive training, may be a promising method to combat cognitive decline in older adults.

Assessing team functional state in emergency response

The need for better measurement of emergency response teams has been recognized as one of the key challenges that characterize the field of team work studies. Ideally, when assessing team performance, one should combine information about the nature of the team (e.g., what is the team structure?) and about its members (e.g., is one of the member exhausted?). This challenge is particularly hard to address in the context of emergency response, due to the inherent complexity and dynamism of the domain. For instance, how do you assess team member’s level of fatigue when deployed in the field?