In recent years, the rapid development of artificial intelligence (AI) especially multi-modal Large Language Models (MLLMs), has enabled it to understand text, images, videos, and other multimedia data, allowing AI systems to execute various tasks based on human-provided prompts. However, AI-powered bots have increasingly been able to bypass most existing CAPTCHA systems, posing significant security threats to web applications. This makes the design of new CAPTCHA mechanisms an urgent priority. We observe that humans are highly sensitive to shifts and abrupt changes in videos, while current AI systems still struggle to comprehend and respond to such situations effectively. Based on this observation, we design and implement BounTCHA, a CAPTCHA mechanism that leverages human perception of boundaries in video transitions and disruptions. By utilizing AI's capability to expand original videos with prompts, we introduce unexpected twists and changes to create a pipeline for generating short videos for CAPTCHA purposes. We develop a prototype and conduct experiments to collect data on humans' time biases in boundary identification. This data serves as a basis for distinguishing between human users and bots. Additionally, we perform a detailed security analysis of BounTCHA, demonstrating its resilience against various types of attacks. We hope that BounTCHA will act as a robust defense, safeguarding millions of web applications in the AI-driven era.

Multiple clustering aims to discover various latent structures of data from different aspects. Deep multiple clustering methods have achieved remarkable performance by exploiting complex patterns and relationships in data. However, existing works struggle to flexibly adapt to diverse user-specific needs in data grouping, which may require manual understanding of each clustering. To address these limitations, we introduce Multi-Sub, a novel end-to-end multiple clustering approach that incorporates a multi-modal subspace proxy learning framework in this work. Utilizing the synergistic capabilities of CLIP and GPT-4, Multi-Sub aligns textual prompts expressing user preferences with their corresponding visual representations. This is achieved by automatically generating proxy words from large language models that act as subspace bases, thus allowing for the customized representation of data in terms specific to the user's interests. Our method consistently outperforms existing baselines across a broad set of datasets in visual multiple clustering tasks.

Despite the high prevalence and burden of mental health conditions, there is a global shortage of mental health providers. Artificial Intelligence (AI) methods have been proposed as a way to address this shortage, by supporting providers with less extensive training as they deliver care. To this end, we developed the AI-Assisted Provider Platform (A2P2), a text-based virtual therapy interface that includes a response suggestion feature, which supports providers in delivering protocolized therapies empathetically. We studied providers with and without expertise in mental health treatment delivering a therapy session using the platform with (intervention) and without (control) AI-assistance features. Upon evaluation, the AI-assisted system significantly decreased response times by 29.34% (p=0.002), Both groups rated the system as having excellent usability. Introduction Mental health conditions are highly prevalent and exert a considerable burden on society, with a global estimated cost of 125.3 million disability-adjusted life years in 2019 The utility of empathy-related AI support for provider selection of professionally crafted text messages remains unevaluated and is the focus of the current work. Response retrieval has several benefits over generation including the safety and controllability of the responses.

This paper studies extractable memorization: training data that an adversary can efficiently extract by querying a machine learning model without prior knowledge of the training dataset. We show an adversary can extract gigabytes of training data from open-source language models like Pythia or GPT-Neo, semi-open models like LLaMA or Falcon, and closed models like ChatGPT. Existing techniques from the literature suffice to attack unaligned models; in order to attack the aligned ChatGPT, we develop a new divergence attack that causes the model to diverge from its chatbot-style generations and emit training data at a rate 150x higher than when behaving properly. Our methods show practical attacks can recover far more data than previously thought, and reveal that current alignment techniques do not eliminate memorization.

David B. Wake, a pioneer in the fields of evolutionary morphology, evolutionary developmental biology (evo-devo), and organismal diversification, died on 29 April. He was 84. Wake was a career-long visionary in organismal biology who led evolutionary biologists to examine not only how organisms are different but also how they become different. As a graduate student, he set the framework for his career by detailing the evolutionary relationships and morphological diversity of salamanders. He then delved into functional morphology (how organismal structures work), evolutionary development (how developmental pathways influence diversification of form), and speciation (how species come to be). One of the most influential and integrative biodiversity scientists of his era, Dave was boundlessly curious about all aspects of evolution and unusually open-minded about new techniques and analyses. Dave was born on 8 June 1936 and raised in South Dakota. He attended Pacific Lutheran College in Tacoma, Washington, where he became fascinated by salamanders after uncovering some while looking for insects for an entomology course. After receiving his BA in biology in 1958, he joined the lab of herpetologist Jay Savage at the University of Southern California (USC). At USC, he met Marvalee Hendricks, a fellow graduate student and scholar of caecilians, another understudied group of amphibians. Dave and Marvalee married in 1962 and became collaborators in life and in science. Dave completed his MS in 1960 and his PhD in 1964, both in biology at USC. He joined the faculty at the University of Chicago for 5 years and then moved to the University of California, Berkeley (UC Berkeley), where he was director of the Museum of Vertebrate Zoology from 1971 to 1998 and professor of integrative biology until his retirement in 2003. Marvalee joined the faculty at UC Berkeley as a tenure-track professor soon after Dave. From the time I first heard of them, they were known as “the Wakes,” and for many of us, they were an early model for successful dual careers in academia. An unapologetic organismal biologist, Dave used salamanders as a model taxon (as opposed to a model organism) to ask questions, answer some, and rework others, creating a cycle of ever-deepening inquiry into their evolution. Rather than focusing on a single model organism or particular evolutionary mechanisms, Dave developed extensive knowledge of many salamander species, enough to use the whole taxon as a model platform to inform his many research foci. In doing so, Dave achieved an unprecedented level of integration across approaches to address evolutionary mechanisms and their consequences for diversification. His exemplary integration inspired a series of papers by James Griesemer in the field of history and philosophy of science. Dave's salamander research transcended boundaries of methodologies, specialties, lines of inquiry, and disciplines. He developed a distinctive form of scientific problem-solving and iterative questioning that synergistically increased our general understanding of evolution. Dave developed expertise in phylogenetics, morphology, development, ecology, neurobiology, behavior, and physiology, and his discoveries were groundbreaking in many fields. Coupling knowledge of adult morphology, ontogeny, embryology, function, and selection, he developed predictions about the retention and/or loss of morphological structures during development and became one of the first to frame these findings as part of the nascent field of evo-devo. By combining detailed spatial knowledge of morphological variation, biogeography, behavior, and genomics, he contributed a classic example of speciation in action with his exploration of the salamander ring species Ensatina . These discoveries, none of which could have been made without integrating approaches, span micro- to macroevolution and are now classics in evolutionary biology. His accomplishments led to many accolades, including election to the American Philosophical Society, the American Academy of Arts and Sciences, and the National Academy of Sciences. In the late 1980s, Dave was an early proponent of action in response to the alarming global decline in amphibians. He chaired the first Declining Amphibian Populations Task Force and raised awareness of the predicament posed to amphibians by anthropogenic changes in climate and landscapes. As with his own research, he promoted diverse approaches in finding the causes of this large-scale biodiversity loss, to the benefit of both the scientific community and amphibians. I joined the Wake lab in 1996 as a postdoc. It was an exciting time, with many field trips and countless discussions of species concepts, salamander tongues, and amphibian declines. Dave had a work ethic that amazed everyone and often left us challenging our expectations of ourselves. His leadership style emphasized showing, not telling. That work ethic resulted in some humorous “Wakeisms”—when lab members took vacations that he perceived as just a bit long, he began lab meetings by naming those who were absent and stating that “they must be gallivanting around the world.” To this day, former Wake lab members refer to vacationing as doing just that. Dave set high standards for research and expected us to meet them. He was honest in his delivery of criticism but somehow made it feel like it was for our own good. He was an unwavering supporter of those who worked with him and incredibly loyal to his students and colleagues. No matter how busy, he eagerly welcomed visiting students and early-career scientists, always showing interest in their stories and offering advice about their research and professional development. Nearly 15 years ago, at a symposium organized around the potential of new genomic approaches in herpetology, Dave regaled the audience with what he saw as the biggest questions still to be answered by integrating this new approach. He genuinely reveled in witnessing the advancement of science, not only through his own work but also through that of his lab members and anyone else who stepped up to the plate. He concluded with a typically positive outlook on science: “I only wish I had another 50 years to live, just to see what you are all going to discover!”

We are writing to scholars such as you with expertise in various areas of service systems, analytics, artificial intelligence, innovation, mobile systems and cognition in hopes that you will consider submitting a paper to our minitrack. The deadline for submitting papers to HICSS-53 is June 15, 2019. Please consider submitting your work if it is related to any of the specific topics listed and/or if you feel it addresses visions of the future of this track. We expect a range of concepts, tools, methods, philosophies and theories to be discussed. We thank you, in advance, for your valuable contribution to HICSS-53.

The Philips Ultra Fast Scanner includes automatic tissue detection and allows for continuous loading without interrupting the scanning process.

Domain generation algorithms (DGAs) are commonly leveraged by malware to create lists of domain names which can be used for command and control (C&C) purposes. Approaches based on machine learning have recently been developed to automatically detect generated domain names in real-time. In this work, we present a novel DGA called CharBot which is capable of producing large numbers of unregistered domain names that are not detected by state-of-the-art classifiers for real-time detection of DGAs, including the recently published methods FANCI (a random forest based on human-engineered features) and LSTM.MI (a deep learning approach). CharBot is very simple, effective and requires no knowledge of the targeted DGA classifiers. We show that retraining the classifiers on CharBot samples is not a viable defense strategy. We believe these findings show that DGA classifiers are inherently vulnerable to adversarial attacks if they rely only on the domain name string to make a decision. Designing a robust DGA classifier may, therefore, necessitate the use of additional information besides the domain name alone. To the best of our knowledge, CharBot is the simplest and most efficient black-box adversarial attack against DGA classifiers proposed to date.

Becky Friedman stands outside a rapture party, May 21, 2011 at Dorky's Arcade in Tacoma, Wash. First the bad news: The world is going to end one day and there is nothing we can do about it. The good news: We probably have a billion years to enjoy ourselves here before that happens. Humans have always been obsessed with The End. Since the dawn of civilization, people claiming to know when the big day is coming have stirred up trouble, sparked panics and started cults.

Economists have become increasingly interested in studying the nature of production functions in social policy applications, with the goal of improving productivity. Traditionally models have assumed workers are homogenous inputs. However, in practice, substantial variability in productivity means the marginal productivity of labor depends substantially on which new workers are hired--which requires not an estimate of a causal effect, but rather a prediction. Annoyed at being automatically tagged with Facebook's facial-recognition system? Wearing a pair of tie-dye-looking glasses could help.