Unsupervised anomaly detection is often framed around two widely studied paradigms. Deep one-class classification, exemplified by Deep SVDD, learns compact latent representations of normality, while density estimators realized by normalizing flows directly model the likelihood of nominal data. In this work, we show that uniformly scaling flows (USFs), normalizing flows with a constant Jacobian determinant, precisely connect these approaches. Specifically, we prove how training a USF via maximum-likelihood reduces to a Deep SVDD objective with a unique regularization that inherently prevents representational collapse. This theoretical bridge implies that USFs inherit both the density faithfulness of flows and the distance-based reasoning of one-class methods. We further demonstrate that USFs induce a tighter alignment between negative log-likelihood and latent norm than either Deep SVDD or non-USFs, and how recent hybrid approaches combining one-class objectives with VAEs can be naturally extended to USFs. Consequently, we advocate using USFs as a drop-in replacement for non-USFs in modern anomaly detection architectures. Empirically, this substitution yields consistent performance gains and substantially improved training stability across multiple benchmarks and model backbones for both image-level and pixel-level detection. These results unify two major anomaly detection paradigms, advancing both theoretical understanding and practical performance.

Recently, there has been an increased interest in NeRF methods which reconstruct differentiable representation of three-dimensional scenes. One of the main limitations of such methods is their inability to assess the confidence of the model in its predictions. In this paper, we propose a new neural network model for the formation of extended vector representations, called uSF, which allows the model to predict not only color and semantic label of each point, but also estimate the corresponding values of uncertainty. We show that with a small number of images available for training, a model quantifying uncertainty performs better than a model without such functionality. Code of the uSF approach is publicly available at https://github.com/sevashasla/usf/.

Transfer in Reinforcement Learning (RL) refers to the idea of applying knowledge gained from previous tasks to solving related tasks. Learning a universal value function (Schaul et al., 2015), which generalizes over goals and states, has previously been shown to be useful for transfer. However, successor features are believed to be more suitable than values for transfer (Dayan, 1993; Barreto et al., 2017), even though they cannot directly generalize to new goals. In this paper, we propose (1) Universal Successor Features (USFs) to capture the underlying dynamics of the environment while allowing generalization to unseen goals and (2) a flexible end-to-end model of USFs that can be trained by interacting with the environment. We show that learning USFs is compatible with any RL algorithm that learns state values using a temporal difference method. Our experiments in a simple gridworld and with two MuJoCo environments show that USFs can greatly accelerate training when learning multiple tasks and can effectively transfer knowledge to new tasks.

A large public research university figured out how to tap the power of artificial intelligence and human intelligence to produce impressive gains in student success. Over the past ten years, the University of South Florida (USF) has experienced dramatic improvement in student success, as measured by the first-year persistence rate, the four-year graduation rate, and the six-year graduation rate. Each of those metrics has improved substantially. To promote persistence and completion, the university implemented a wide array of programs, practices, and policies based on a campus Student Success Task Force Report released in April 2010. These initiatives included many of the standard student success initiatives in place at many other colleges and universities, including living learning communities, the professionalization of academic advising, gradual increases in admissions requirements, course redesigns, expansion of on-campus housing, and promotion of student engagement.

San Francisco is known as a hub of tech innovation, making USF an ideal place to study computer and data science. The location gives students the opportunity to connect professionally with companies everyone knows: Google, Twitter, Facebook – the list goes on. But what opportunities does USF offer students to participate in peer reviewed scholarship, a place where current students and faculty can connect over tech R&D on campus? As of Fall 2018, the answer comes in the form of the weekly MAGICS Lab meetings, a way to gain valuable mentorship and learn about emerging technologies, a place where undergraduate, graduate students, and faculty all have the opportunity to learn, research, and publish together. This group welcomes all skill-levels, from novice to seasoned researchers alike.

The RoboCup Rescue Physical Agent League Competition was held in the summer of 2001 in conjunction with the AAAI Mobile Robot Competition Urban Search and Rescue event, eerily preceding the September 11 World Trade Center (WTC) disaster. Four teams responded to the WTC disaster through the auspices of the Center for Robot-Assisted Search and Rescue (CRASAR), directed by John Blitch. The four teams were Foster- Miller and iRobot (both robot manufacturers from the Boston area), the United States Navy's Space Warfare Center (SPAWAR) group from San Diego, and the University of South Florida (USF). Blitch, through his position as program manager for the Defense Advanced Research Projects Agency (DARPA) Tactical Mobile Robots Program, was a supporter of the competition; he also served as a member of the rules committee and a judge. USF participated by chairing the rules committee, judging, assisting with the logistics, providing commentary, and demonstrating tethered and wireless robots whenever entrants had to skip around during the competition. Based on our experiences and history, we were asked to comment on the validity of the competition. The CRASAR collective experience suggests that most of the basic rules of the competition matched reality because the rules accurately reflected deployment scenarios, but the National Institute of Standards and Technology (NIST) Standard Test Course, and hardware or software approaches forwarded by competitors in last summer's event, missed the mark. This article briefly reviews the types of robots and missions used by CRASAR at the WTC site, then discusses the robotassisted search and rescue effort in terms of lessons for the competition.