We study the problem of distributed training neural networks (NNs) on devices with heterogeneous, limited, and time-varying availability computational resources. present an adaptive, resource-aware, on-device learning mechanism, DISTREAL, which is able to fully efficiently utilize available resources in a manner, increasing convergence speed. This achieved dropout mechanism that dynamically adj...

Compilers are an important part of today’s computational infrastructure because software is ever-increasingly written in high-level programming languages like C, C++, and Java. Early on, compilers were successful for desktop computing, and now they are also used for mission-critical software, trusted mobile code, and real-time embedded systems. The field of compiler design is driven by the adve...

❉❡♣❡♥❞❡♥❝2✲❛✇❛r❡ t❛s❦✲❜❛s❡❞ ♣❛r❛❧❧❡❧ ♣r♦❣r❛♠♠✐♥❣ ♠♦❞❡❧s ❤❛✈❡ ♣r♦✈❡♥ t♦ ❜❡ s✉❝❝❡ss❢✉❧ ❢♦r ❞❡✈❡❧♦♣✐♥❣ ❡✣✲ ❝✐❡♥t ❛♣♣❧✐❝❛t✐♦♥ s♦❢t✇❛r❡ ❢♦r ♠✉❧t✐❝♦r❡✲❜❛s❡❞ ❝♦♠♣✉t❡r ❛r❝❤✐t❡❝t✉r❡s✳ ❚❤❡ ♣r♦❣r❛♠♠✐♥❣ ♠♦❞❡❧ ✐s ❛♠❡♥❛❜❧❡ t♦ ♣r♦❣r❛♠♠❡rs✱ t❤❡r❡❜2 s✉♣♣♦rt✐♥❣ ♣r♦❞✉❝t✐✈✐t2✱ ✇❤✐❧❡ ❤❛r❞✇❛r❡ ♣❡r❢♦r♠❛♥❝❡ ✐s ❛❝❤✐❡✈❡❞ t❤r♦✉❣❤ ❛ r✉♥✲ t✐♠❡ s2st❡♠ t❤❛t ❞2♥❛♠✐❝❛❧❧2 s❝❤❡❞✉❧❡s t❛s❦s ♦♥t♦ ❝♦r❡s ✐♥ s✉❝❤ ❛ ✇❛2 t❤❛t ❛❧❧ ❞❡♣❡♥❞...

Resource Cost Aware Scheduling Problems Rodrigo A. Carrasco Managing the consumption of non-renewable and/or limited resources has become an important issue in many different settings. In this dissertation we explore the topic of resource cost aware scheduling. Unlike the purely scheduling problems, in the resource cost aware setting we are not only interested in a scheduling performance metric...

We motivate and introduce an extension of the well-known cumulative constraint which deals with time and volume dependent cost of resources. Our research is primarily interested in scheduling problems under time and volume variable electricity costs, but the constraint equally applies to manpower scheduling when hourly rates differ over time and/or extra personnel incur higher hourly rates. We ...

Exhaustive verification methods such as model-checking suffer from the well-known state-explosion problem: the set of states is too large to explore exhaustively in reasonable amounts of time and space (memory). But model-checkers are often plagued with another problem, which makes state-explosion even worse: the disk-swapping problem. The latter problem manifests itself when the model-checker ...

We present an architecture for remote, collaborative visualization of datasets over the Grid. This permits an implementation-agnostic approach, where different systems can be discovered, reserved and orchestrated without being concerned about specific hardware configurations. We illustrate the utility of our approach to deliver high-quality interactive visualizations of datasets (circa 1 millio...