Mark Redekopp's research while affiliated with University of Southern California and other places

... Interdisciplinary team-based projects in engineering education are an approach to experiential learning which can provide students with a diverse learning opportunity to work closely with individuals from different disciplines [1,2,3]. Some of the benefits of participating on an interdisciplinary team include unique solutions to solving complex problems [3], improves integration of ideas from different disciplines [4], and allowing students the opportunity to connect with their learning on a deeper level [5]. ...

... Several studies examined how student learning changed between traditional lecture and flipped methods. The latter enabled students to better-achieve higher-order learning objectives [14,15,16,10,17] but not lower-order objectives. Some found better student performance on all questions at or above Bloom's level three [10] and other found improved performance, "specifically for the middle two quartiles of students (25-75% percentile)" [18]. ...

... And then students strengthen and solidifies their understanding through in-class learning activities that encompass group problem-solving, discussion, collaborative learning, and active learning [3]. Primary advantages of the flipped classroom pedagogy have been identified as increased flexibility that allows learning at an individual's own pace through pre-recorded videos, a positive classroom atmosphere, and more opportunities for active learning [4], [5]. On the other hand, the pedagogy involves challenges such as inadequate student preparation prior to class, lack of instant feedback for out-of-class assignments, and substantial instructor commitment requirements on producing videos [5]- [7]. ...

... Importance of conducting workload characterizations on large scale real-world non-GPU systems has been emphasized in multiple work. Redekopp et al. presented analysis and optimizations for complicated graph analysis algorithms [91]. They evaluated graph processing frameworks which follow Bulk Synchronous Parallel Processing (BSP) model on cloud. ...

... In addition, the overhead for installing and maintaining a local cluster is non-trivial unless strong system administration resources are available at the local institution. Availability of platform services such as storage and programming abstractions such as .NET or MapReduce reduces the overhead of installing, monitoring and managing such services locally [148]. "We used AWS for calculating similarity scores and large-scale data modeling. ...

... Study [12] observed that offloading low CPU utilizing (i.e., monitoring) tasks from the node's operating system to the BMC reduced the power consumption by a factor of 2.6. There is an increasing trend to perform system monitoring using OOB mechanisms [13,14]. ...