Deep convolutional neural networks (CNNs) are rapidly becoming the dominant approach to computer vision and a major component of many other pervasive machine learning tasks, such as speech recognition, natural language processing, and fraud detection.  As a result, accelerators for efficiently evaluating DNNs are rapidly growing in popularity.  Our work in this area focuses on two key challenges: minimizing the off-chip data transfer and maximizing the utilization of the computation units.  In this talk, I will present an overview of my research work on understanding and impr

Enabled through the rapid advancement of real-time sensing and communication technologies, we are seeing a more flexible resource consumption behavior in many of our societal infrastructure systems that previously faced "inelastic" demand. This transformation can provide us with new operational flexibility and potentially enable us to make our urban world safer and more efficient. However, dispatching this new form of flexibility requires modeling and influencing human behavior.

Join us as we discuss the Carbon Neutrality Initiative (CNI) at UCSB! To begin the discussion we will have Dr. David Lea, the Global Climate Leadership Council (GCLC) representative for UCSB, discuss his role. This will be followed be a brief presentation by Jordan Sager, our Campus Energy Manager and Elizabeth Szulc, a Carbon Neutrality Fellow at UCSB. We will conclude with a Q&A and will leave plenty of room for discussion.

With the decline and eventual end of historical rates of lithographic scaling, we arrive at a crossroad where synergistic and holistic decisions are required to preserve performance over power scaling in digital computing. Numerous technologies are emerging with this exact aim, from devices (transistors), memories, 3D integration, specialization, photonics, and others.

High power, high efficiency semiconductor diode lasers are used as the primary pump source in virtually all new solid state and fiber laser systems.  In many cases, the output power scalability of the full system is limited by the pump diode performance (power, spatial brightness, and/or spectral brightness).  Over the past decade, research and development in the commercial sector has resulted in significant improvements increase in each of these areas.  Both diode laser power and spatial brightness are thermally-limited, and addressing each requires addressing the waste heat

LED lighting is already saving massive amounts of energy as a more efficient replacement for simple lighting applications. Unlike previous efficient lighting technologies, LED offers higher efficiency without any compromise to lighting performance or quality. Beyond replacement of old technologies for simple illumination, LED technology can offer even further energy savings and additional benefits enabled by the new technology platform.

Spend an enjoyable evening with Rob Bernard, Chief Environmental Strategist and Microsoft, and learn why sustainability is no longer just a moral argument made by tree huggers and environmentalist, but one that now has the full-throated support and resources of major businesses aligned to solve this problem. The enormous consequences of climate change and human activity are no longer theoretical, giving rise to an entire ecosystem of established companies and startups that collectively seek to solve or mitigate ramifications associated with water, drought, food, energy, and more.

The IEEE Photonics Society's second Spring Workshop on Light Science at UCSB is aimed at increasing awareness of optics and photonics with speakers from both university and industry settings. 

The event will include a career fair, in which attendees will have one-on-one time with representatives of local companies as well as a poster session to display research being done at UCSB. 

The brain can perform massively parallel information processing while consuming only ~1- 100 fJ per synaptic event.  Two-terminal memristors based on filament forming metal oxides (FFMOs) or phase change memory (PCM) materials have recently been demonstrated to function as non-volatile memory that can emulate neuronal and synaptic functions. Despite recent progress in the fabrication of device arrays however, to date no architecture has been shown to operate with the projected energy efficiency while maintaining high accuracy.

The industrial sector is a major contributor to global energy and resource use, climate change, and pollutant emissions. To accelerate the much-needed transition to cleaner industrial pathways, manufacturers, investors, and policy makers need rigorous analyses of emerging technologies that encompass not only environmental, but also technical and economic performance criteria.