The University of Washington is at the forefront of an international effort to innovate the semiconductor industry while building a skilled U.S.-based workforce to design and manufacture chip technology. UW ECE and Physics Professor Mo Li is leading the UW's contribution to this effort.
UW joins $110M cross-Pacific effort to advance artificial intelligence
The UW joins a landmark $110 million cross-Pacific effort and will partner with Amazon, NVIDIA and the University of Tsukuba, Japan, to advance artificial intelligence.
UW ECE alumnus and Affiliate Professor Jai Jaisimha (Ph.D. ‘96) has co-founded a new nonprofit organization, the Transparency Coalition.ai, which is advocating for transparency and regulation of the data used to train artificial intelligence.
Q&A: How to train AI when you don’t have enough data
Sarah McQuate from UW News recently interviewed UW ECE Professor Jenq-Neng Hwang about his research and how his team trains machine learning algorithms for artificial intelligence using limited data sets.
Bingzhao Li — bringing light and sound into computer chips
2023 Yang Award recipient Bingzhao Li is a postdoctoral research fellow in the UW Laboratory of Photonic Systems, which researches integrated photonic devices, optoelectronic materials and quantum photonics.
The University of Washington is at the forefront of an international effort to innovate the semiconductor industry while building a skilled U.S.-based workforce to design and manufacture chip technology. UW ECE and Physics Professor Mo Li is leading the UW's contribution to this effort.
The UW joins a landmark $110 million cross-Pacific effort and will partner with Amazon, NVIDIA and the University of Tsukuba, Japan, to advance artificial intelligence.
UW ECE alumnus and Affiliate Professor Jai Jaisimha (Ph.D. ‘96) has co-founded a new nonprofit organization, the Transparency Coalition.ai, which is advocating for transparency and regulation of the data used to train artificial intelligence.
Sarah McQuate from UW News recently interviewed UW ECE Professor Jenq-Neng Hwang about his research and how his team trains machine learning algorithms for artificial intelligence using limited data sets.
Ultra-thin meta-optics have the potential to make imaging systems lighter and thinner than ever. Using a new inverse design framework, a UW ECE-led research team has demonstrated broadband thermal imaging with meta-optics for a wide range of applications.
882uweeViewNews Object
(
[_showAnnouncements:protected] =>
[_showTitle:protected] =>
[showMore] =>
[_type:protected] => spotlight
[_from:protected] => newsawards_landing
[_args:protected] => Array
(
[post_type] => spotlight
[meta_query] => Array
(
[0] => Array
(
[key] => type
[value] => news
[compare] => LIKE
)
)
[posts_per_page] => 6
[post_status] => publish
)
[_jids:protected] =>
[_taxa:protected] => Array
(
)
[_meta:protected] => Array
(
[0] => Array
(
[key] => type
[value] => news
[compare] => LIKE
)
)
[_metarelation:protected] => AND
[_results:protected] => Array
(
[0] => WP_Post Object
(
[ID] => 34205
[post_author] => 27
[post_date] => 2024-04-25 14:30:37
[post_date_gmt] => 2024-04-25 21:30:37
[post_content] =>
[post_title] => UPWARDS for the Future
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => upwards-for-the-future
[to_ping] =>
[pinged] =>
[post_modified] => 2024-04-25 14:57:27
[post_modified_gmt] => 2024-04-25 21:57:27
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34205
[menu_order] => 1
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[1] => WP_Post Object
(
[ID] => 34173
[post_author] => 27
[post_date] => 2024-04-19 15:16:20
[post_date_gmt] => 2024-04-19 22:16:20
[post_content] =>
[post_title] => UW joins $110M cross-Pacific effort to advance artificial intelligence
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => uw-joins-110m-cross-pacific-effort-to-advance-artificial-intelligence
[to_ping] =>
[pinged] =>
[post_modified] => 2024-04-19 15:16:20
[post_modified_gmt] => 2024-04-19 22:16:20
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34173
[menu_order] => 3
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[2] => WP_Post Object
(
[ID] => 34124
[post_author] => 27
[post_date] => 2024-04-11 13:27:46
[post_date_gmt] => 2024-04-11 20:27:46
[post_content] => By Wayne Gillam | UW ECE News
[caption id="attachment_34126" align="alignright" width="575"] UW ECE alumnus and Affiliate Professor Jai Jaisimha (Ph.D. ‘96) has co-founded a new nonprofit organization, the Transparency Coalition.ai, which is advocating for transparency and regulation of the data used to train artificial intelligence. / Photo by Igor Omilaev, courtesy of the Transparency Coalition.ai[/caption]
Artificial intelligence holds great promise as well as possible peril for society. This rapidly evolving technology stands to accelerate advances in science, engineering, and healthcare as well as improve efficiency and productivity in a vast range of industries. But AI also has serious downsides, which include the potential to spread misinformation and disinformation online, exacerbate algorithmic biases, compromise individual privacy, and even obliterate copyright protections for intellectual property.
It is the peril of AI that drew the attention of UW ECE alumnus and Affiliate Professor Jai Jaisimha (Ph.D. ‘96), who has worked as a technology entrepreneur for decades and has held leadership positions in major tech companies that leverage AI, such as Amazon and Microsoft. Now, along with his co-founder Rob Eleveld, Jaisimha has created a new nonprofit organization, the Transparency Coalition.ai, which seeks to address these concerns by advocating for greater transparency and regulation of the data used to train and inform AI and generative AI models, such as ChatGPT.
“Rob and I were concerned about the rollout of generative AI and how it was going. We were already seeing evidence of societal harm. And we knew that we had the expertise needed to help address some of these issues,” Jaisimha said. “Because of my work in computer vision and imaging when I was at the UW, I spent years studying and understanding pattern recognition, how it worked, and the mathematics behind it. I built early versions of these AI models that are similar to the ones that are out there now.”
[caption id="attachment_34130" align="alignleft" width="250"] UW ECE alumnus and Affiliate Professor Jai Jaisimha[/caption]
The roots of Jaisimha’s knowledge about AI models and the data used to train them goes back to his time as a doctoral student working with his adviser, UW ECE Professor Emeritus Eve Riskin, who is now the dean of undergraduate education in the electrical and computer engineering department at the Stevens Institute of Technology. With oversight by Riskin, Jaisimha applied statistical pattern recognition techniques to better understand datasets and how to use them to make online browsing a more interactive and personalized experience. During this time, he also had an internship at a startup that did research for government agencies, which taught him the importance of data curation as well as algorithmic model construction, testing and validation.
These experiences as a student carried over into his professional life, which has been defined by his entrepreneurial ventures and corporate leadership.
“When you’re building a company, it’s not just about making a cool demo,” Jaisimha said. “It’s about making sure you have the entire pipeline of AI model building in place: data collection, data cleansing, being rigorous about validating and testing your results, and automating performance monitoring. That way, you end up building a robust system. Those are all things I learned at UW ECE.”
The Transparency Coalition.ai
According to Jaisimha, there is a large amount of research available about ethical and responsible AI development, and many companies have even formed entire departments focused on the ethical implementation of AI in their products. However, profit incentives have interfered with these good intentions. Teams focused on ethical AI in companies are often shrunk or even shut down. And thoughtful research in this area is too often ignored by industry leaders in favor of getting products to market faster and increasing profits.
Jaisimha and his colleagues at the Transparency Coalition.ai believe that the government has an important role to play in regulating AI and providing needed oversight in a highly competitive marketplace. They have chosen to focus their efforts on advocating for greater transparency in AI training data, which Jaisimha believes is key to promoting more ethical and responsible AI development.
Jaisimha said that focusing on AI training data is important because today’s generative AI models are ingesting large amounts of uncurated data. This includes copyrighted material and content behind paywalls, in social media and on personal websites, and even illicit and illegal content, such as child pornography. This indiscriminate ingestion of large amounts of uncurated data makes generative AI systems prone to frequent and well-documented “hallucinations,” where the system provides warped images or wrong and misleading answers when prompted.
"Because I’m an affiliate professor in UW ECE, I’ve been able to tap into the wider UW ecosystem, whether it’s the RAISE group, or computer science and engineering, or linguistics, or law. There’s a lot of good thinking happening across the university, and I’m looking for ways to bring it forward to legislators and to the public." — UW ECE alumnus and Affiliate Professor Jai Jaisimha
In contrast, a generative AI model with a transparent, curated dataset is less prone to hallucinations, greatly reduces the risk for societal harm, and is customized for solving specific problems. In many cases, the curation of training data also optimizes the AI model to accomplish tasks more efficiently. Because of these facts, Jaisimha believes that transparent, regulated, and curated AI training data will not only be good for society, in the long-term, it will be good for business too.
“If you want to solve real-world problems, you have to embrace the idea of either taking these big, generative AI models and refining them for a specific application or building the model in a way that acknowledges that need for customization from the beginning,” Jaisimha said. “We know it’s possible to implement some degree of constraint on AI training data. And I believe the result will be a thriving ecosystem of companies building technologies that will be more practical, useful, and focused on solving real-world problems.”
The Transparency Coalition.ai has set its sights on state-level advocacy, and it is reaching out to legislators in Washington and California. According to Jaisimha, state governments can often move to implement policy more rapidly than the federal government. Washington and California are also headquarters for a high number of leading technology companies working with artificial intelligence, which widens the impact of policies and legislation enacted in these states nationally and internationally.
Since its founding in October 2023, the Transparency Coalition.ai has already scored a significant victory. A bill establishing a new AI Task Force for the state of Washington was signed into law by Governor Jay Inslee last month. Among its other mandates, the Task Force will be considering appropriate regulation for AI training data, which is something the Transparency Coalition.ai advocated for among legislators involved in establishing the group.
At the UW, Jaisimha is collaborating with Professor Chirag Shah in the UW Information School, who runs the Responsibility in AI Systems and Experiences, or RAISE, team at the University. RAISE conducts high-quality research on ethical and responsible AI development, and Jaisimha is connecting the organization with Washington state legislators through the Transparency Coalition.ai to help better inform AI policy and regulation.
UW connections
Jaisimha became a UW ECE affiliate professor in September 2023, and going forward, he plans to retain and continue to grow his connections at the UW. In addition to being affiliated with RAISE and advising student teams in the Department’s Engineering Innovation and Entrepreneurship capstone program, known as ENGINE, he is involved in CoMotion at the UW. There, he is a mentor for students and faculty and is actively involved in the CoMotion Innovation Gap Fund, which helps to move UW inventions and innovations into commercial investment and development. Jaisimha also has several family connections to the University. His wife received her master’s degree in business administration from the Foster School of Business, and their eldest son will be graduating this year from UW ECE with a bachelor’s degree. Their youngest son is a sophomore studying informatics at the UW iSchool.
“We say in our family that we ‘bleed purple.’ The four of us are deeply connected to the UW, and I have a strong sense of needing to pay it forward,” Jaisimha said. “I see both my affiliate work at UW ECE and my mentorship work at CoMotion as a form of giving back. The Transparency Coalition.ai is also a form of giving back to the community.”
Jaisimha encourages students pursuing careers in AI development to not worry too much about whether what they are studying now will remain relevant in this fast-changing world. Instead, he notes there is a specific way of breaking down problems that engineering students learn, which is a useful skill to apply in many situations. He also recommends that students interested in AI explore edge technology, such as TinyML, which brings machine learning networks into resource-constrained devices. And he suggests that students consider taking statistics classes, which can help an ECE graduate stand out in fields related to AI and machine learning.
In regard to his work with the Transparency Coalition.ai, Jaisimha and his co-founder are continuing their outreach to state legislators and seeking donor support for their efforts. He encourages everyone who can to get involved and contact their state legislators to let them know about the need for AI transparency and oversight. To this end, the Transparency Coalition.ai recently installed a bill tracker on their website, which helps make it easy for people to see what bills related to AI are active in their state. Jaisimha also plans to continue to bring pertinent information about AI development to government officials going forward.
“Because I’m an affiliate professor in UW ECE, I’ve been able to tap into the wider UW ecosystem, whether it’s the RAISE group, or computer science and engineering, or linguistics, or law. There’s a lot of good thinking happening across the university, and I’m looking for ways to bring it forward to legislators and to the public,” Jaisimha said. “I’ve never stopped educating myself. Now that I’m back at the UW, I’m continuing my education and benefiting from being at one of the greatest institutions in the world.”
For more information about UW ECE Affiliate Professor Jai Jaisimha and his advocacy work, visit the Transparency Coalition.ai website.
[post_title] => A call for AI data transparency
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => ai-transparency
[to_ping] =>
[pinged] =>
[post_modified] => 2024-04-11 16:13:50
[post_modified_gmt] => 2024-04-11 23:13:50
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34124
[menu_order] => 4
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[3] => WP_Post Object
(
[ID] => 34045
[post_author] => 36
[post_date] => 2024-04-05 11:33:54
[post_date_gmt] => 2024-04-05 18:33:54
[post_content] => [caption id="attachment_34048" align="alignright" width="523"] UW ECE Assistant Professor Jungwon Choi's research focuses on designing power circuits that can receive electrical currents at high frequencies from a charging source and transfer the energy to the battery. Shown above, Choi with UW ECE graduate student Manas Palmal.[/caption]
Adapted from an article by Brooke Fisher, Photos by Dennis Wise | UW College of Engineering
Imagine rolling into a parking spot and your electric vehicle (EV) automatically begins to charge, quickly and without cables, thanks to a compact charging station on the ground. UW ECE Assistant Professor Jungwon Choi can do more than envision it — she’s developing the technology.
“Charging time is a barrier for people buying EVs,” Choi says. “I’m interested in how we can make more efficient power circuits to charge the battery in electric vehicles.”
To enhance EV charging, Choi is involved in research on many levels. In addition to advancing the design of spiral coils for high-frequency wireless charging — in which power is transmitted electromagnetically between coils located in a vehicle and charging station — her primary research focuses on designing power circuits that can receive electrical currents at high frequencies from a charging source and transfer the energy to the battery.
[caption id="attachment_34073" align="alignleft" width="418"] UW ECE graduate student Ghovindo Surya turns the knob on an oscilloscope to test the electrical currents received by the power circuit.[/caption]
“We want to have high efficiency,” Choi explains. “When we have 100% power at input and the battery receives only 80% power, then it’s lost as heat. It’s harmful for the system and energy is lost.”
A unique feature of power converters that Choi’s team is working to advance is the two-way flow of energy, which would enable EV batteries to store energy that could be utilized as backup power.
“In an emergency situation, or in case of a blackout, we could draw power from a vehicle into a house,” Choi says.
Learn more about how UW engineering research is driven to advance vehicle electrification on the UW College of Engineering website.
[post_title] => Leading the charge to enhance power transfer
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => leading-the-charge-to-enhance-power-transfer
[to_ping] =>
[pinged] =>
[post_modified] => 2024-04-05 11:33:54
[post_modified_gmt] => 2024-04-05 18:33:54
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34045
[menu_order] => 5
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[4] => WP_Post Object
(
[ID] => 34037
[post_author] => 27
[post_date] => 2024-03-29 10:34:16
[post_date_gmt] => 2024-03-29 17:34:16
[post_content] =>
[post_title] => Q&A: How to train AI when you don’t have enough data
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => jenq-neng-hwang-ai-training
[to_ping] =>
[pinged] =>
[post_modified] => 2024-03-29 10:46:24
[post_modified_gmt] => 2024-03-29 17:46:24
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34037
[menu_order] => 6
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[5] => WP_Post Object
(
[ID] => 33933
[post_author] => 27
[post_date] => 2024-03-14 09:11:19
[post_date_gmt] => 2024-03-14 16:11:19
[post_content] => By Anna Wirth-Singh | UW Department of Physics
[caption id="attachment_33938" align="alignright" width="550"] Ultra-thin meta-optics have the potential to make imaging systems lighter and thinner than ever. Using a new inverse design framework, a UW ECE-led research team has demonstrated broadband thermal imaging with meta-optics for applications ranging from consumer electronics to thermal sensing and night vision. Shown above, a side view of a fabricated wafer containing meta-optics held above a conventional refractive lens. The meta-optics were developed by a multi-institutional research team led by UW ECE and Physics Associate Professor Arka Majumdar. Photo by Anna Wirth-Singh.[/caption]
Long-wavelength infrared (LWIR) imaging holds critical significance across many applications, from consumer electronics to defense and national security. It finds applications in night vision, remote sensing, and long-range imaging. However, the conventional refractive lenses employed in these imaging systems are bulky and heavy, which is undesirable for almost all applications. Compounding this issue is the fact that many LWIR refractive lenses are crafted from expensive and limited-supply materials, such as germanium.
The next generation of optical systems demands lenses that are not only lighter and thinner than ever before, but also uphold uncompromising image quality. This demand has fueled a surge of efforts to develop ultra-thin sub-wavelength diffractive optics, known as meta-optics. Meta-optics, in their simplest form, consist of arrays of sub-wavelength scale nanopillars on a flat surface, with each pillar introducing a local phase shift to light passing through. By strategically arranging these pillars, the light can be controlled to produce steering and lensing. While conventional refractive lenses are close to a centimeter thick, meta-optics are about 500 microns thick, which dramatically reduces the overall thickness of the optics.
However, one challenge with meta-optics is strong chromatic aberrations. That is, light of different wavelengths interacts with the structure in different ways, and the result is typically a lens that cannot simultaneously focus light of different wavelengths in the same focal plane. Largely because of this issue, meta-optics have not yet fully replaced their refractive counterparts despite the benefits in size and weight reduction. In particular, the area of LWIR meta-optics is relatively unexplored compared to visible wavelength meta-optics, and the potential advantages of meta-optics over conventional refractive lenses are significant given the unique and extensive applications of this wavelength range.
[caption id="attachment_33941" align="alignleft" width="300"] UW ECE and Physics Associate Professor Arka Majumdar. Photo by Ryan Hoover.[/caption]
Now, in a new paper published in Nature Communications, a multi-institutional team of researchers, led by UW ECE and Physics Associate Professor Arka Majumdar, has introduced a new design framework termed “MTF-engineering.” The modulation transfer function, or MTF, describes how well a lens maintains image contrast as a function of spatial frequency. This framework addresses the challenges associated with broadband meta-optics to design and experimentally demonstrate thermal imaging with meta-optics in laboratory and real-world settings. The team built upon already successful inverse design techniques by developing a framework which optimizes both the pillar shape and the global arrangement simultaneously. UW ECE-affiliated team members included recent alumni Luocheng Huang (the paper’s lead author) and Zheyi Han, postdoctoral researchers Saswata Mukherjee, Johannes Fröch, and Quentin Tanguy as well as UW ECE Professor Karl Böhringer, who is the director of the Institute for Nano-Engineered Systems at the UW.
Leveraging artificial intelligence and a new inverse design framework
[caption id="attachment_33951" align="alignright" width="550"] Meta-optics, in their simplest form, consist of arrays of sub-wavelength scale pillars on a flat surface, with each pillar introducing a local phase shift to light passing through. By strategically arranging these pillars, the light can be controlled to produce steering and lensing. (Above) A full view of a fabricated wafer containing meta-optics. (Below) Scanning electron microscope images of the nanopillars contained within the team’s meta-optics. These meta-optics contain both complex light scatterers (left) and simple scatterers (right). Photos provided by Anna Wirth-Singh.[/caption]
One key innovation in the research team’s approach is the use of artificial intelligence — a deep neural network (DNN) model — to map between pillar shape and phase. In an inverse design process for large area optics, it is not computationally feasible to simulate how the light interacts with each pillar at each iteration. To solve this problem, the authors simulated a large library of nanopillars (also called “meta-atoms”) and used the simulated data to train a DNN. The DNN enabled a quick mapping between scatterer and phase in the optimization loop, allowing the inverse design of large-area optics containing millions of micron-scale pillars.
Another key innovation in this work is the figure of merit (FoM), leading to the framework being termed “MTF-engineering.” In inverse design, one defines an FoM and computationally optimizes the structure or arrangement to maximize the FoM. However, it is often not intuitive why the produced result is optimal. For this work, the authors leveraged their expertise in meta-optics to define an FoM that is intuitive.
Majumdar explained, “The figure of merit is related to the area under the MTF curve. The idea here is to pass as much information as possible through the lens, which is captured in the MTF. Then, combined with a light computational backend, we can achieve a high-quality image.” He continued, “The figure of merit reflects what we intuitively know about optics. This particular FoM is optimized when all the wavelengths perform equally well, thus constraining our optics to have uniform performance over the specified wavelengths without explicitly defining uniformity as an optimization criterion.”
This approach, combining intuition from meta-optics and a light computational backend, significantly improves performance compared to simple metalenses.
The authors fabricated their designed optics from a single silicon wafer, which is promising for future applications involving germanium-free LWIR imaging systems. While acknowledging that there is still room for improvement to achieve imaging quality comparable to commercial refractive lens systems, this work represents a significant step toward that goal.
The researchers have generously made their MTF-engineering framework, named “metabox,” available online via GitHub, inviting others to use it for designing their own meta-optics. The research team expressed excitement about the potential works that may emerge from the utilization of metabox in the broader scientific community.
This article is an adaptation of a blog post available at Springer Nature Research Communities. For more information about the research described above read, “Broadband thermal imaging using meta-optics” in Nature Communications or contact UW ECE and Physics Associate Professor Arka Majumdar.
[post_title] => Ultra-flat optics for broadband thermal imaging
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => ultra-flat-optics
[to_ping] =>
[pinged] =>
[post_modified] => 2024-03-14 09:14:23
[post_modified_gmt] => 2024-03-14 16:14:23
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=33933
[menu_order] => 7
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
)
[_numposts:protected] => 6
[_rendered:protected] => 1
[_classes:protected] => Array
(
[0] => view-block
[1] => block--spotlight-robust-news
)
[_finalHTML:protected] =>
The University of Washington is at the forefront of an international effort to innovate the semiconductor industry while building a skilled U.S.-based workforce to design and manufacture chip technology. UW ECE and Physics Professor Mo Li is leading the UW's contribution to this effort.
The UW joins a landmark $110 million cross-Pacific effort and will partner with Amazon, NVIDIA and the University of Tsukuba, Japan, to advance artificial intelligence.
UW ECE alumnus and Affiliate Professor Jai Jaisimha (Ph.D. ‘96) has co-founded a new nonprofit organization, the Transparency Coalition.ai, which is advocating for transparency and regulation of the data used to train artificial intelligence.
Sarah McQuate from UW News recently interviewed UW ECE Professor Jenq-Neng Hwang about his research and how his team trains machine learning algorithms for artificial intelligence using limited data sets.
Ultra-thin meta-optics have the potential to make imaging systems lighter and thinner than ever. Using a new inverse design framework, a UW ECE-led research team has demonstrated broadband thermal imaging with meta-optics for a wide range of applications.
[_postID:protected] => 184
[_errors:protected] => Array
(
)
[_block:protected] =>
[_db:protected] => WP_Query Object
(
[query] => Array
(
[post_type] => spotlight
[meta_query] => Array
(
[0] => Array
(
[key] => type
[value] => news
[compare] => LIKE
)
)
[posts_per_page] => 6
[post_status] => publish
)
[query_vars] => Array
(
[post_type] => spotlight
[meta_query] => Array
(
[0] => Array
(
[key] => type
[value] => news
[compare] => LIKE
)
)
[posts_per_page] => 6
[post_status] => publish
[error] =>
[m] =>
[p] => 0
[post_parent] =>
[subpost] =>
[subpost_id] =>
[attachment] =>
[attachment_id] => 0
[name] =>
[pagename] =>
[page_id] => 0
[second] =>
[minute] =>
[hour] =>
[day] => 0
[monthnum] => 0
[year] => 0
[w] => 0
[category_name] =>
[tag] =>
[cat] =>
[tag_id] =>
[author] =>
[author_name] =>
[feed] =>
[tb] =>
[paged] => 0
[meta_key] =>
[meta_value] =>
[preview] =>
[s] =>
[sentence] =>
[title] =>
[fields] =>
[menu_order] =>
[embed] =>
[category__in] => Array
(
)
[category__not_in] => Array
(
)
[category__and] => Array
(
)
[post__in] => Array
(
)
[post__not_in] => Array
(
)
[post_name__in] => Array
(
)
[tag__in] => Array
(
)
[tag__not_in] => Array
(
)
[tag__and] => Array
(
)
[tag_slug__in] => Array
(
)
[tag_slug__and] => Array
(
)
[post_parent__in] => Array
(
)
[post_parent__not_in] => Array
(
)
[author__in] => Array
(
)
[author__not_in] => Array
(
)
[orderby] => menu_order
[order] => ASC
[ignore_sticky_posts] =>
[suppress_filters] =>
[cache_results] => 1
[update_post_term_cache] => 1
[lazy_load_term_meta] => 1
[update_post_meta_cache] => 1
[nopaging] =>
[comments_per_page] => 50
[no_found_rows] =>
)
[tax_query] => WP_Tax_Query Object
(
[queries] => Array
(
)
[relation] => AND
[table_aliases:protected] => Array
(
)
[queried_terms] => Array
(
)
[primary_table] => wp_posts
[primary_id_column] => ID
)
[meta_query] => WP_Meta_Query Object
(
[queries] => Array
(
[0] => Array
(
[key] => type
[value] => news
[compare] => LIKE
)
[relation] => OR
)
[relation] => AND
[meta_table] => wp_postmeta
[meta_id_column] => post_id
[primary_table] => wp_posts
[primary_id_column] => ID
[table_aliases:protected] => Array
(
[0] => wp_postmeta
)
[clauses:protected] => Array
(
[wp_postmeta] => Array
(
[key] => type
[value] => news
[compare] => LIKE
[compare_key] => =
[alias] => wp_postmeta
[cast] => CHAR
)
)
[has_or_relation:protected] =>
)
[date_query] =>
[request] => SELECT SQL_CALC_FOUND_ROWS wp_posts.ID FROM wp_posts INNER JOIN wp_postmeta ON ( wp_posts.ID = wp_postmeta.post_id ) WHERE 1=1 AND (
( wp_postmeta.meta_key = 'type' AND wp_postmeta.meta_value LIKE '{f06064d1d8ac7b9c33ea5cf96159f9b52ca33ccec5d25c6cb4dbf54bdef87652}news{f06064d1d8ac7b9c33ea5cf96159f9b52ca33ccec5d25c6cb4dbf54bdef87652}' )
) AND wp_posts.post_type = 'spotlight' AND ((wp_posts.post_status = 'publish')) GROUP BY wp_posts.ID ORDER BY wp_posts.menu_order ASC LIMIT 0, 6
[posts] => Array
(
[0] => WP_Post Object
(
[ID] => 34205
[post_author] => 27
[post_date] => 2024-04-25 14:30:37
[post_date_gmt] => 2024-04-25 21:30:37
[post_content] =>
[post_title] => UPWARDS for the Future
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => upwards-for-the-future
[to_ping] =>
[pinged] =>
[post_modified] => 2024-04-25 14:57:27
[post_modified_gmt] => 2024-04-25 21:57:27
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34205
[menu_order] => 1
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[1] => WP_Post Object
(
[ID] => 34173
[post_author] => 27
[post_date] => 2024-04-19 15:16:20
[post_date_gmt] => 2024-04-19 22:16:20
[post_content] =>
[post_title] => UW joins $110M cross-Pacific effort to advance artificial intelligence
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => uw-joins-110m-cross-pacific-effort-to-advance-artificial-intelligence
[to_ping] =>
[pinged] =>
[post_modified] => 2024-04-19 15:16:20
[post_modified_gmt] => 2024-04-19 22:16:20
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34173
[menu_order] => 3
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[2] => WP_Post Object
(
[ID] => 34124
[post_author] => 27
[post_date] => 2024-04-11 13:27:46
[post_date_gmt] => 2024-04-11 20:27:46
[post_content] => By Wayne Gillam | UW ECE News
[caption id="attachment_34126" align="alignright" width="575"] UW ECE alumnus and Affiliate Professor Jai Jaisimha (Ph.D. ‘96) has co-founded a new nonprofit organization, the Transparency Coalition.ai, which is advocating for transparency and regulation of the data used to train artificial intelligence. / Photo by Igor Omilaev, courtesy of the Transparency Coalition.ai[/caption]
Artificial intelligence holds great promise as well as possible peril for society. This rapidly evolving technology stands to accelerate advances in science, engineering, and healthcare as well as improve efficiency and productivity in a vast range of industries. But AI also has serious downsides, which include the potential to spread misinformation and disinformation online, exacerbate algorithmic biases, compromise individual privacy, and even obliterate copyright protections for intellectual property.
It is the peril of AI that drew the attention of UW ECE alumnus and Affiliate Professor Jai Jaisimha (Ph.D. ‘96), who has worked as a technology entrepreneur for decades and has held leadership positions in major tech companies that leverage AI, such as Amazon and Microsoft. Now, along with his co-founder Rob Eleveld, Jaisimha has created a new nonprofit organization, the Transparency Coalition.ai, which seeks to address these concerns by advocating for greater transparency and regulation of the data used to train and inform AI and generative AI models, such as ChatGPT.
“Rob and I were concerned about the rollout of generative AI and how it was going. We were already seeing evidence of societal harm. And we knew that we had the expertise needed to help address some of these issues,” Jaisimha said. “Because of my work in computer vision and imaging when I was at the UW, I spent years studying and understanding pattern recognition, how it worked, and the mathematics behind it. I built early versions of these AI models that are similar to the ones that are out there now.”
[caption id="attachment_34130" align="alignleft" width="250"] UW ECE alumnus and Affiliate Professor Jai Jaisimha[/caption]
The roots of Jaisimha’s knowledge about AI models and the data used to train them goes back to his time as a doctoral student working with his adviser, UW ECE Professor Emeritus Eve Riskin, who is now the dean of undergraduate education in the electrical and computer engineering department at the Stevens Institute of Technology. With oversight by Riskin, Jaisimha applied statistical pattern recognition techniques to better understand datasets and how to use them to make online browsing a more interactive and personalized experience. During this time, he also had an internship at a startup that did research for government agencies, which taught him the importance of data curation as well as algorithmic model construction, testing and validation.
These experiences as a student carried over into his professional life, which has been defined by his entrepreneurial ventures and corporate leadership.
“When you’re building a company, it’s not just about making a cool demo,” Jaisimha said. “It’s about making sure you have the entire pipeline of AI model building in place: data collection, data cleansing, being rigorous about validating and testing your results, and automating performance monitoring. That way, you end up building a robust system. Those are all things I learned at UW ECE.”
The Transparency Coalition.ai
According to Jaisimha, there is a large amount of research available about ethical and responsible AI development, and many companies have even formed entire departments focused on the ethical implementation of AI in their products. However, profit incentives have interfered with these good intentions. Teams focused on ethical AI in companies are often shrunk or even shut down. And thoughtful research in this area is too often ignored by industry leaders in favor of getting products to market faster and increasing profits.
Jaisimha and his colleagues at the Transparency Coalition.ai believe that the government has an important role to play in regulating AI and providing needed oversight in a highly competitive marketplace. They have chosen to focus their efforts on advocating for greater transparency in AI training data, which Jaisimha believes is key to promoting more ethical and responsible AI development.
Jaisimha said that focusing on AI training data is important because today’s generative AI models are ingesting large amounts of uncurated data. This includes copyrighted material and content behind paywalls, in social media and on personal websites, and even illicit and illegal content, such as child pornography. This indiscriminate ingestion of large amounts of uncurated data makes generative AI systems prone to frequent and well-documented “hallucinations,” where the system provides warped images or wrong and misleading answers when prompted.
"Because I’m an affiliate professor in UW ECE, I’ve been able to tap into the wider UW ecosystem, whether it’s the RAISE group, or computer science and engineering, or linguistics, or law. There’s a lot of good thinking happening across the university, and I’m looking for ways to bring it forward to legislators and to the public." — UW ECE alumnus and Affiliate Professor Jai Jaisimha
In contrast, a generative AI model with a transparent, curated dataset is less prone to hallucinations, greatly reduces the risk for societal harm, and is customized for solving specific problems. In many cases, the curation of training data also optimizes the AI model to accomplish tasks more efficiently. Because of these facts, Jaisimha believes that transparent, regulated, and curated AI training data will not only be good for society, in the long-term, it will be good for business too.
“If you want to solve real-world problems, you have to embrace the idea of either taking these big, generative AI models and refining them for a specific application or building the model in a way that acknowledges that need for customization from the beginning,” Jaisimha said. “We know it’s possible to implement some degree of constraint on AI training data. And I believe the result will be a thriving ecosystem of companies building technologies that will be more practical, useful, and focused on solving real-world problems.”
The Transparency Coalition.ai has set its sights on state-level advocacy, and it is reaching out to legislators in Washington and California. According to Jaisimha, state governments can often move to implement policy more rapidly than the federal government. Washington and California are also headquarters for a high number of leading technology companies working with artificial intelligence, which widens the impact of policies and legislation enacted in these states nationally and internationally.
Since its founding in October 2023, the Transparency Coalition.ai has already scored a significant victory. A bill establishing a new AI Task Force for the state of Washington was signed into law by Governor Jay Inslee last month. Among its other mandates, the Task Force will be considering appropriate regulation for AI training data, which is something the Transparency Coalition.ai advocated for among legislators involved in establishing the group.
At the UW, Jaisimha is collaborating with Professor Chirag Shah in the UW Information School, who runs the Responsibility in AI Systems and Experiences, or RAISE, team at the University. RAISE conducts high-quality research on ethical and responsible AI development, and Jaisimha is connecting the organization with Washington state legislators through the Transparency Coalition.ai to help better inform AI policy and regulation.
UW connections
Jaisimha became a UW ECE affiliate professor in September 2023, and going forward, he plans to retain and continue to grow his connections at the UW. In addition to being affiliated with RAISE and advising student teams in the Department’s Engineering Innovation and Entrepreneurship capstone program, known as ENGINE, he is involved in CoMotion at the UW. There, he is a mentor for students and faculty and is actively involved in the CoMotion Innovation Gap Fund, which helps to move UW inventions and innovations into commercial investment and development. Jaisimha also has several family connections to the University. His wife received her master’s degree in business administration from the Foster School of Business, and their eldest son will be graduating this year from UW ECE with a bachelor’s degree. Their youngest son is a sophomore studying informatics at the UW iSchool.
“We say in our family that we ‘bleed purple.’ The four of us are deeply connected to the UW, and I have a strong sense of needing to pay it forward,” Jaisimha said. “I see both my affiliate work at UW ECE and my mentorship work at CoMotion as a form of giving back. The Transparency Coalition.ai is also a form of giving back to the community.”
Jaisimha encourages students pursuing careers in AI development to not worry too much about whether what they are studying now will remain relevant in this fast-changing world. Instead, he notes there is a specific way of breaking down problems that engineering students learn, which is a useful skill to apply in many situations. He also recommends that students interested in AI explore edge technology, such as TinyML, which brings machine learning networks into resource-constrained devices. And he suggests that students consider taking statistics classes, which can help an ECE graduate stand out in fields related to AI and machine learning.
In regard to his work with the Transparency Coalition.ai, Jaisimha and his co-founder are continuing their outreach to state legislators and seeking donor support for their efforts. He encourages everyone who can to get involved and contact their state legislators to let them know about the need for AI transparency and oversight. To this end, the Transparency Coalition.ai recently installed a bill tracker on their website, which helps make it easy for people to see what bills related to AI are active in their state. Jaisimha also plans to continue to bring pertinent information about AI development to government officials going forward.
“Because I’m an affiliate professor in UW ECE, I’ve been able to tap into the wider UW ecosystem, whether it’s the RAISE group, or computer science and engineering, or linguistics, or law. There’s a lot of good thinking happening across the university, and I’m looking for ways to bring it forward to legislators and to the public,” Jaisimha said. “I’ve never stopped educating myself. Now that I’m back at the UW, I’m continuing my education and benefiting from being at one of the greatest institutions in the world.”
For more information about UW ECE Affiliate Professor Jai Jaisimha and his advocacy work, visit the Transparency Coalition.ai website.
[post_title] => A call for AI data transparency
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => ai-transparency
[to_ping] =>
[pinged] =>
[post_modified] => 2024-04-11 16:13:50
[post_modified_gmt] => 2024-04-11 23:13:50
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34124
[menu_order] => 4
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[3] => WP_Post Object
(
[ID] => 34045
[post_author] => 36
[post_date] => 2024-04-05 11:33:54
[post_date_gmt] => 2024-04-05 18:33:54
[post_content] => [caption id="attachment_34048" align="alignright" width="523"] UW ECE Assistant Professor Jungwon Choi's research focuses on designing power circuits that can receive electrical currents at high frequencies from a charging source and transfer the energy to the battery. Shown above, Choi with UW ECE graduate student Manas Palmal.[/caption]
Adapted from an article by Brooke Fisher, Photos by Dennis Wise | UW College of Engineering
Imagine rolling into a parking spot and your electric vehicle (EV) automatically begins to charge, quickly and without cables, thanks to a compact charging station on the ground. UW ECE Assistant Professor Jungwon Choi can do more than envision it — she’s developing the technology.
“Charging time is a barrier for people buying EVs,” Choi says. “I’m interested in how we can make more efficient power circuits to charge the battery in electric vehicles.”
To enhance EV charging, Choi is involved in research on many levels. In addition to advancing the design of spiral coils for high-frequency wireless charging — in which power is transmitted electromagnetically between coils located in a vehicle and charging station — her primary research focuses on designing power circuits that can receive electrical currents at high frequencies from a charging source and transfer the energy to the battery.
[caption id="attachment_34073" align="alignleft" width="418"] UW ECE graduate student Ghovindo Surya turns the knob on an oscilloscope to test the electrical currents received by the power circuit.[/caption]
“We want to have high efficiency,” Choi explains. “When we have 100% power at input and the battery receives only 80% power, then it’s lost as heat. It’s harmful for the system and energy is lost.”
A unique feature of power converters that Choi’s team is working to advance is the two-way flow of energy, which would enable EV batteries to store energy that could be utilized as backup power.
“In an emergency situation, or in case of a blackout, we could draw power from a vehicle into a house,” Choi says.
Learn more about how UW engineering research is driven to advance vehicle electrification on the UW College of Engineering website.
[post_title] => Leading the charge to enhance power transfer
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => leading-the-charge-to-enhance-power-transfer
[to_ping] =>
[pinged] =>
[post_modified] => 2024-04-05 11:33:54
[post_modified_gmt] => 2024-04-05 18:33:54
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34045
[menu_order] => 5
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[4] => WP_Post Object
(
[ID] => 34037
[post_author] => 27
[post_date] => 2024-03-29 10:34:16
[post_date_gmt] => 2024-03-29 17:34:16
[post_content] =>
[post_title] => Q&A: How to train AI when you don’t have enough data
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => jenq-neng-hwang-ai-training
[to_ping] =>
[pinged] =>
[post_modified] => 2024-03-29 10:46:24
[post_modified_gmt] => 2024-03-29 17:46:24
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34037
[menu_order] => 6
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[5] => WP_Post Object
(
[ID] => 33933
[post_author] => 27
[post_date] => 2024-03-14 09:11:19
[post_date_gmt] => 2024-03-14 16:11:19
[post_content] => By Anna Wirth-Singh | UW Department of Physics
[caption id="attachment_33938" align="alignright" width="550"] Ultra-thin meta-optics have the potential to make imaging systems lighter and thinner than ever. Using a new inverse design framework, a UW ECE-led research team has demonstrated broadband thermal imaging with meta-optics for applications ranging from consumer electronics to thermal sensing and night vision. Shown above, a side view of a fabricated wafer containing meta-optics held above a conventional refractive lens. The meta-optics were developed by a multi-institutional research team led by UW ECE and Physics Associate Professor Arka Majumdar. Photo by Anna Wirth-Singh.[/caption]
Long-wavelength infrared (LWIR) imaging holds critical significance across many applications, from consumer electronics to defense and national security. It finds applications in night vision, remote sensing, and long-range imaging. However, the conventional refractive lenses employed in these imaging systems are bulky and heavy, which is undesirable for almost all applications. Compounding this issue is the fact that many LWIR refractive lenses are crafted from expensive and limited-supply materials, such as germanium.
The next generation of optical systems demands lenses that are not only lighter and thinner than ever before, but also uphold uncompromising image quality. This demand has fueled a surge of efforts to develop ultra-thin sub-wavelength diffractive optics, known as meta-optics. Meta-optics, in their simplest form, consist of arrays of sub-wavelength scale nanopillars on a flat surface, with each pillar introducing a local phase shift to light passing through. By strategically arranging these pillars, the light can be controlled to produce steering and lensing. While conventional refractive lenses are close to a centimeter thick, meta-optics are about 500 microns thick, which dramatically reduces the overall thickness of the optics.
However, one challenge with meta-optics is strong chromatic aberrations. That is, light of different wavelengths interacts with the structure in different ways, and the result is typically a lens that cannot simultaneously focus light of different wavelengths in the same focal plane. Largely because of this issue, meta-optics have not yet fully replaced their refractive counterparts despite the benefits in size and weight reduction. In particular, the area of LWIR meta-optics is relatively unexplored compared to visible wavelength meta-optics, and the potential advantages of meta-optics over conventional refractive lenses are significant given the unique and extensive applications of this wavelength range.
[caption id="attachment_33941" align="alignleft" width="300"] UW ECE and Physics Associate Professor Arka Majumdar. Photo by Ryan Hoover.[/caption]
Now, in a new paper published in Nature Communications, a multi-institutional team of researchers, led by UW ECE and Physics Associate Professor Arka Majumdar, has introduced a new design framework termed “MTF-engineering.” The modulation transfer function, or MTF, describes how well a lens maintains image contrast as a function of spatial frequency. This framework addresses the challenges associated with broadband meta-optics to design and experimentally demonstrate thermal imaging with meta-optics in laboratory and real-world settings. The team built upon already successful inverse design techniques by developing a framework which optimizes both the pillar shape and the global arrangement simultaneously. UW ECE-affiliated team members included recent alumni Luocheng Huang (the paper’s lead author) and Zheyi Han, postdoctoral researchers Saswata Mukherjee, Johannes Fröch, and Quentin Tanguy as well as UW ECE Professor Karl Böhringer, who is the director of the Institute for Nano-Engineered Systems at the UW.
Leveraging artificial intelligence and a new inverse design framework
[caption id="attachment_33951" align="alignright" width="550"] Meta-optics, in their simplest form, consist of arrays of sub-wavelength scale pillars on a flat surface, with each pillar introducing a local phase shift to light passing through. By strategically arranging these pillars, the light can be controlled to produce steering and lensing. (Above) A full view of a fabricated wafer containing meta-optics. (Below) Scanning electron microscope images of the nanopillars contained within the team’s meta-optics. These meta-optics contain both complex light scatterers (left) and simple scatterers (right). Photos provided by Anna Wirth-Singh.[/caption]
One key innovation in the research team’s approach is the use of artificial intelligence — a deep neural network (DNN) model — to map between pillar shape and phase. In an inverse design process for large area optics, it is not computationally feasible to simulate how the light interacts with each pillar at each iteration. To solve this problem, the authors simulated a large library of nanopillars (also called “meta-atoms”) and used the simulated data to train a DNN. The DNN enabled a quick mapping between scatterer and phase in the optimization loop, allowing the inverse design of large-area optics containing millions of micron-scale pillars.
Another key innovation in this work is the figure of merit (FoM), leading to the framework being termed “MTF-engineering.” In inverse design, one defines an FoM and computationally optimizes the structure or arrangement to maximize the FoM. However, it is often not intuitive why the produced result is optimal. For this work, the authors leveraged their expertise in meta-optics to define an FoM that is intuitive.
Majumdar explained, “The figure of merit is related to the area under the MTF curve. The idea here is to pass as much information as possible through the lens, which is captured in the MTF. Then, combined with a light computational backend, we can achieve a high-quality image.” He continued, “The figure of merit reflects what we intuitively know about optics. This particular FoM is optimized when all the wavelengths perform equally well, thus constraining our optics to have uniform performance over the specified wavelengths without explicitly defining uniformity as an optimization criterion.”
This approach, combining intuition from meta-optics and a light computational backend, significantly improves performance compared to simple metalenses.
The authors fabricated their designed optics from a single silicon wafer, which is promising for future applications involving germanium-free LWIR imaging systems. While acknowledging that there is still room for improvement to achieve imaging quality comparable to commercial refractive lens systems, this work represents a significant step toward that goal.
The researchers have generously made their MTF-engineering framework, named “metabox,” available online via GitHub, inviting others to use it for designing their own meta-optics. The research team expressed excitement about the potential works that may emerge from the utilization of metabox in the broader scientific community.
This article is an adaptation of a blog post available at Springer Nature Research Communities. For more information about the research described above read, “Broadband thermal imaging using meta-optics” in Nature Communications or contact UW ECE and Physics Associate Professor Arka Majumdar.
[post_title] => Ultra-flat optics for broadband thermal imaging
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => ultra-flat-optics
[to_ping] =>
[pinged] =>
[post_modified] => 2024-03-14 09:14:23
[post_modified_gmt] => 2024-03-14 16:14:23
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=33933
[menu_order] => 7
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
)
[post_count] => 6
[current_post] => -1
[in_the_loop] =>
[post] => WP_Post Object
(
[ID] => 34205
[post_author] => 27
[post_date] => 2024-04-25 14:30:37
[post_date_gmt] => 2024-04-25 21:30:37
[post_content] =>
[post_title] => UPWARDS for the Future
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => upwards-for-the-future
[to_ping] =>
[pinged] =>
[post_modified] => 2024-04-25 14:57:27
[post_modified_gmt] => 2024-04-25 21:57:27
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34205
[menu_order] => 1
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[comment_count] => 0
[current_comment] => -1
[found_posts] => 882
[max_num_pages] => 147
[max_num_comment_pages] => 0
[is_single] =>
[is_preview] =>
[is_page] =>
[is_archive] => 1
[is_date] =>
[is_year] =>
[is_month] =>
[is_day] =>
[is_time] =>
[is_author] =>
[is_category] =>
[is_tag] =>
[is_tax] =>
[is_search] =>
[is_feed] =>
[is_comment_feed] =>
[is_trackback] =>
[is_home] =>
[is_404] =>
[is_embed] =>
[is_paged] =>
[is_admin] =>
[is_attachment] =>
[is_singular] =>
[is_robots] =>
[is_posts_page] =>
[is_post_type_archive] => 1
[query_vars_hash:WP_Query:private] => c64914061c8ecf9b16abe746203f6ad7
[query_vars_changed:WP_Query:private] => 1
[thumbnails_cached] =>
[allow_query_attachment_by_filename:protected] =>
[stopwords:WP_Query:private] =>
[compat_fields:WP_Query:private] => Array
(
[0] => query_vars_hash
[1] => query_vars_changed
)
[compat_methods:WP_Query:private] => Array
(
[0] => init_query_flags
[1] => parse_tax_query
)
)
)
UW ECE alumnus and Affiliate Professor Jai Jaisimha (Ph.D. ‘96) has co-founded a new nonprofit organization, the Transparency Coalition.ai, which is advocating for transparency and regulation of the data used to train artificial intelligence. / Photo by Igor Omilaev, courtesy of the Transparency Coalition.ai[/caption]
Artificial intelligence holds great promise as well as possible peril for society. This rapidly evolving technology stands to accelerate advances in science, engineering, and healthcare as well as improve efficiency and productivity in a vast range of industries. But AI also has serious downsides, which include the potential to spread misinformation and disinformation online, exacerbate algorithmic biases, compromise individual privacy, and even obliterate copyright protections for intellectual property.
It is the peril of AI that drew the attention of UW ECE alumnus and Affiliate Professor Jai Jaisimha (Ph.D. ‘96), who has worked as a technology entrepreneur for decades and has held leadership positions in major tech companies that leverage AI, such as Amazon and Microsoft. Now, along with his co-founder Rob Eleveld, Jaisimha has created a new nonprofit organization, the Transparency Coalition.ai, which seeks to address these concerns by advocating for greater transparency and regulation of the data used to train and inform AI and generative AI models, such as ChatGPT.
“Rob and I were concerned about the rollout of generative AI and how it was going. We were already seeing evidence of societal harm. And we knew that we had the expertise needed to help address some of these issues,” Jaisimha said. “Because of my work in computer vision and imaging when I was at the UW, I spent years studying and understanding pattern recognition, how it worked, and the mathematics behind it. I built early versions of these AI models that are similar to the ones that are out there now.”
[caption id="attachment_34130" align="alignleft" width="250"]
UW ECE alumnus and Affiliate Professor Jai Jaisimha[/caption]
The roots of Jaisimha’s knowledge about AI models and the data used to train them goes back to his time as a doctoral student working with his adviser, UW ECE Professor Emeritus Eve Riskin, who is now the dean of undergraduate education in the electrical and computer engineering department at the Stevens Institute of Technology. With oversight by Riskin, Jaisimha applied statistical pattern recognition techniques to better understand datasets and how to use them to make online browsing a more interactive and personalized experience. During this time, he also had an internship at a startup that did research for government agencies, which taught him the importance of data curation as well as algorithmic model construction, testing and validation.
These experiences as a student carried over into his professional life, which has been defined by his entrepreneurial ventures and corporate leadership.
“When you’re building a company, it’s not just about making a cool demo,” Jaisimha said. “It’s about making sure you have the entire pipeline of AI model building in place: data collection, data cleansing, being rigorous about validating and testing your results, and automating performance monitoring. That way, you end up building a robust system. Those are all things I learned at UW ECE.”
UW ECE Assistant Professor Jungwon Choi's research focuses on designing power circuits that can receive electrical currents at high frequencies from a charging source and transfer the energy to the battery. Shown above, Choi with UW ECE graduate student Manas Palmal.[/caption]
Adapted from an article by Brooke Fisher, Photos by Dennis Wise | UW College of Engineering
Imagine rolling into a parking spot and your electric vehicle (EV) automatically begins to charge, quickly and without cables, thanks to a compact charging station on the ground. UW ECE Assistant Professor Jungwon Choi can do more than envision it — she’s developing the technology.
“Charging time is a barrier for people buying EVs,” Choi says. “I’m interested in how we can make more efficient power circuits to charge the battery in electric vehicles.”
UW ECE graduate student Ghovindo Surya turns the knob on an oscilloscope to test the electrical currents received by the power circuit.[/caption]
“We want to have high efficiency,” Choi explains. “When we have 100% power at input and the battery receives only 80% power, then it’s lost as heat. It’s harmful for the system and energy is lost.”
A unique feature of power converters that Choi’s team is working to advance is the two-way flow of energy, which would enable EV batteries to store energy that could be utilized as backup power.
“In an emergency situation, or in case of a blackout, we could draw power from a vehicle into a house,” Choi says.
Learn more about how UW engineering research is driven to advance vehicle electrification on the UW College of Engineering website.
[post_title] => Leading the charge to enhance power transfer
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => leading-the-charge-to-enhance-power-transfer
[to_ping] =>
[pinged] =>
[post_modified] => 2024-04-05 11:33:54
[post_modified_gmt] => 2024-04-05 18:33:54
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34045
[menu_order] => 5
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[4] => WP_Post Object
(
[ID] => 34037
[post_author] => 27
[post_date] => 2024-03-29 10:34:16
[post_date_gmt] => 2024-03-29 17:34:16
[post_content] =>
[post_title] => Q&A: How to train AI when you don’t have enough data
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => jenq-neng-hwang-ai-training
[to_ping] =>
[pinged] =>
[post_modified] => 2024-03-29 10:46:24
[post_modified_gmt] => 2024-03-29 17:46:24
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=34037
[menu_order] => 6
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
[5] => WP_Post Object
(
[ID] => 33933
[post_author] => 27
[post_date] => 2024-03-14 09:11:19
[post_date_gmt] => 2024-03-14 16:11:19
[post_content] => By Anna Wirth-Singh | UW Department of Physics
[caption id="attachment_33938" align="alignright" width="550"]
Ultra-thin meta-optics have the potential to make imaging systems lighter and thinner than ever. Using a new inverse design framework, a UW ECE-led research team has demonstrated broadband thermal imaging with meta-optics for applications ranging from consumer electronics to thermal sensing and night vision. Shown above, a side view of a fabricated wafer containing meta-optics held above a conventional refractive lens. The meta-optics were developed by a multi-institutional research team led by UW ECE and Physics Associate Professor Arka Majumdar. Photo by Anna Wirth-Singh.[/caption]
Long-wavelength infrared (LWIR) imaging holds critical significance across many applications, from consumer electronics to defense and national security. It finds applications in night vision, remote sensing, and long-range imaging. However, the conventional refractive lenses employed in these imaging systems are bulky and heavy, which is undesirable for almost all applications. Compounding this issue is the fact that many LWIR refractive lenses are crafted from expensive and limited-supply materials, such as germanium.
The next generation of optical systems demands lenses that are not only lighter and thinner than ever before, but also uphold uncompromising image quality. This demand has fueled a surge of efforts to develop ultra-thin sub-wavelength diffractive optics, known as meta-optics. Meta-optics, in their simplest form, consist of arrays of sub-wavelength scale nanopillars on a flat surface, with each pillar introducing a local phase shift to light passing through. By strategically arranging these pillars, the light can be controlled to produce steering and lensing. While conventional refractive lenses are close to a centimeter thick, meta-optics are about 500 microns thick, which dramatically reduces the overall thickness of the optics.
However, one challenge with meta-optics is strong chromatic aberrations. That is, light of different wavelengths interacts with the structure in different ways, and the result is typically a lens that cannot simultaneously focus light of different wavelengths in the same focal plane. Largely because of this issue, meta-optics have not yet fully replaced their refractive counterparts despite the benefits in size and weight reduction. In particular, the area of LWIR meta-optics is relatively unexplored compared to visible wavelength meta-optics, and the potential advantages of meta-optics over conventional refractive lenses are significant given the unique and extensive applications of this wavelength range.
[caption id="attachment_33941" align="alignleft" width="300"]
UW ECE and Physics Associate Professor Arka Majumdar. Photo by Ryan Hoover.[/caption]
Now, in a new paper published in Nature Communications, a multi-institutional team of researchers, led by UW ECE and Physics Associate Professor Arka Majumdar, has introduced a new design framework termed “MTF-engineering.” The modulation transfer function, or MTF, describes how well a lens maintains image contrast as a function of spatial frequency. This framework addresses the challenges associated with broadband meta-optics to design and experimentally demonstrate thermal imaging with meta-optics in laboratory and real-world settings. The team built upon already successful inverse design techniques by developing a framework which optimizes both the pillar shape and the global arrangement simultaneously. UW ECE-affiliated team members included recent alumni Luocheng Huang (the paper’s lead author) and Zheyi Han, postdoctoral researchers Saswata Mukherjee, Johannes Fröch, and Quentin Tanguy as well as UW ECE Professor Karl Böhringer, who is the director of the Institute for Nano-Engineered Systems at the UW.
Meta-optics, in their simplest form, consist of arrays of sub-wavelength scale pillars on a flat surface, with each pillar introducing a local phase shift to light passing through. By strategically arranging these pillars, the light can be controlled to produce steering and lensing. (Above) A full view of a fabricated wafer containing meta-optics. (Below) Scanning electron microscope images of the nanopillars contained within the team’s meta-optics. These meta-optics contain both complex light scatterers (left) and simple scatterers (right). Photos provided by Anna Wirth-Singh.[/caption]
One key innovation in the research team’s approach is the use of artificial intelligence — a deep neural network (DNN) model — to map between pillar shape and phase. In an inverse design process for large area optics, it is not computationally feasible to simulate how the light interacts with each pillar at each iteration. To solve this problem, the authors simulated a large library of nanopillars (also called “meta-atoms”) and used the simulated data to train a DNN. The DNN enabled a quick mapping between scatterer and phase in the optimization loop, allowing the inverse design of large-area optics containing millions of micron-scale pillars.
Another key innovation in this work is the figure of merit (FoM), leading to the framework being termed “MTF-engineering.” In inverse design, one defines an FoM and computationally optimizes the structure or arrangement to maximize the FoM. However, it is often not intuitive why the produced result is optimal. For this work, the authors leveraged their expertise in meta-optics to define an FoM that is intuitive.
Majumdar explained, “The figure of merit is related to the area under the MTF curve. The idea here is to pass as much information as possible through the lens, which is captured in the MTF. Then, combined with a light computational backend, we can achieve a high-quality image.” He continued, “The figure of merit reflects what we intuitively know about optics. This particular FoM is optimized when all the wavelengths perform equally well, thus constraining our optics to have uniform performance over the specified wavelengths without explicitly defining uniformity as an optimization criterion.”
This approach, combining intuition from meta-optics and a light computational backend, significantly improves performance compared to simple metalenses.
The authors fabricated their designed optics from a single silicon wafer, which is promising for future applications involving germanium-free LWIR imaging systems. While acknowledging that there is still room for improvement to achieve imaging quality comparable to commercial refractive lens systems, this work represents a significant step toward that goal.
The researchers have generously made their MTF-engineering framework, named “metabox,” available online via GitHub, inviting others to use it for designing their own meta-optics. The research team expressed excitement about the potential works that may emerge from the utilization of metabox in the broader scientific community.
This article is an adaptation of a blog post available at Springer Nature Research Communities. For more information about the research described above read, “Broadband thermal imaging using meta-optics” in Nature Communications or contact UW ECE and Physics Associate Professor Arka Majumdar.
[post_title] => Ultra-flat optics for broadband thermal imaging
[post_excerpt] =>
[post_status] => publish
[comment_status] => closed
[ping_status] => closed
[post_password] =>
[post_name] => ultra-flat-optics
[to_ping] =>
[pinged] =>
[post_modified] => 2024-03-14 09:14:23
[post_modified_gmt] => 2024-03-14 16:14:23
[post_content_filtered] =>
[post_parent] => 0
[guid] => https://www.ece.uw.edu/?post_type=spotlight&p=33933
[menu_order] => 7
[post_type] => spotlight
[post_mime_type] =>
[comment_count] => 0
[filter] => raw
)
)
[_numposts:protected] => 6
[_rendered:protected] => 1
[_classes:protected] => Array
(
[0] => view-block
[1] => block--spotlight-robust-news
)
[_finalHTML:protected] => https://www.washington.edu/news/2024/04/22/uw-leads-international-group-in-semiconductor-research-and-workforce-development/
