Large language models (LLMs) often exhibit societal biases in their outputs, prompting ethical concerns regarding fairness and harm. In this work, we propose KLAAD (KL-Attention Alignment Debiasing), an attention-based debiasing framework that implicitly aligns attention distributions between stereotypical and anti-stereotypical sentence pairs without directly modifying model weights. KLAAD introduces a composite training objective combining Cross-Entropy, KL divergence, and Triplet losses, guiding the model to consistently attend across biased and unbiased contexts while preserving fluency and coherence. Experimental evaluation of KLAAD demonstrates improved bias mitigation on both the BBQ and BOLD benchmarks, with minimal impact on language modeling quality. The results indicate that attention-level alignment offers a principled solution for mitigating bias in generative language models.

We present a novel tuning procedure for random forests (RFs) that improves the accuracy of estimated quantiles and produces valid, relatively narrow prediction intervals. While RFs are typically used to estimate mean responses (conditional on covariates), they can also be used to estimate quantiles by estimating the full distribution of the response. However, standard approaches for building RFs often result in excessively biased quantile estimates. To reduce this bias, our proposed tuning procedure minimizes "quantile coverage loss" (QCL), which we define as the estimated bias of the marginal quantile coverage probability estimate based on the out-of-bag sample. We adapt QCL tuning to handle censored data and demonstrate its use with random survival forests. We show that QCL tuning results in quantile estimates with more accurate coverage probabilities than those achieved using default parameter values or traditional tuning (using MSPE for uncensored data and C-index for censored data), while also reducing the estimated MSE of these coverage probabilities. We discuss how the superior performance of QCL tuning is linked to its alignment with the estimation goal. Finally, we explore the validity and width of prediction intervals created using this method.

Federated learning combines local updates from clients to produce a global model, which is susceptible to poisoning attacks. Most previous defense strategies relied on vectors derived from projections of local updates on a Euclidean space; however, these methods fail to accurately represent the functionality and structure of local models, resulting in inconsistent performance. Here, we present a new paradigm to defend against poisoning attacks in federated learning using functional mappings of local models based on intermediate outputs. Experiments show that our mechanism is robust under a broad range of computing conditions and advanced attack scenarios, enabling safer collaboration among data-sensitive participants via federated learning.

In early February, Israel sent six quadcopter drones containing explosives into a facility near the city of Kermanshah that was Iran's main manufacturing and storage plant for military drones, according to a senior intelligence official briefed on the operation. That Israeli attack destroyed dozens of Iran's drones. Iran retaliated by firing ballistic missiles at a housing complex in northern Iraq that it said had been used by Israeli agents to plot attacks against Iran. In June 2021, another attack using a quadcopter drone -- which explodes on impact -- was also launched from within the country. It struck the Iran Centrifuge Technology Company, or TESA, in the city of Karaj.

For many Asians, heightened xenophobia and the rise in hate crimes during 2020, and now through 2021, added extra stress and trauma to their everyday lives. In a now too-familiar story, Chanhee Choi, a South Korean student at the University of Washington, was attacked in downtown Seattle by a racist assailant, ranting about Chinese people and the coronavirus. Afterward, she decided to do something that only she could have done to bring awareness to the issue. She decided to make a game about it. "It was around the beginning of the pandemic, in 2020," said Choi. "I was walking down the street in downtown Seattle. At the moment I was just back from a trip home to see my family. There, everyone was wearing masks, but here, nobody did it. I was the only one wearing a mask because I just came from South Korea, so I was worried about being around others, if it was possible to get coronavirus. I was just protecting myself, but I didn't expect that someone could judge me or have a problem, or think wearing a mask makes me look like I'm sick. Suddenly one guy started yelling at me like, 'Are you Chinese? He raised his fist to my face. I looked around for help and everyone turned away, like they didn't want to see me. I felt like I was the only Asian in the city, even though Seattle has so many. I was there by myself, knowing what he was doing to me. I had never felt this kind of fear in the United States. Since that happened, I don't go downtown alone now. At the time I noticed that every time Trump was on the news, he mentioned the China virus. But why did that happen to me? That was my first question. I wanted to share this kind of feeling and sadness, so others could try to understand the experience that I had."

While previous approaches perform untargeted attacks on the state of the agent, we propose a method to perform targeted attacks to lure an agent into consistently following a desired policy. We place ourselves in a realistic setting, where attacks are performed on observations of the environment rather than the internal state of the agent and develop constant attacks instead of per-observation ones. We illustrate our method by attacking deep RL agents playing Atari games and show that universal additive masks can be applied not only to degrade performance but to take control of an agent.

In this work, we study the robustness of a CNN+RNN based image captioning system being subjected to adversarial noises. We propose to fool an image captioning system to generate some targeted partial captions for an image polluted by adversarial noises, even the targeted captions are totally irrelevant to the image content. A partial caption indicates that the words at some locations in this caption are observed, while words at other locations are not restricted.It is the first work to study exact adversarial attacks of targeted partial captions. Due to the sequential dependencies among words in a caption, we formulate the generation of adversarial noises for targeted partial captions as a structured output learning problem with latent variables. Both the generalized expectation maximization algorithm and structural SVMs with latent variables are then adopted to optimize the problem. The proposed methods generate very successful at-tacks to three popular CNN+RNN based image captioning models. Furthermore, the proposed attack methods are used to understand the inner mechanism of image captioning systems, providing the guidance to further improve automatic image captioning systems towards human captioning.

Successful candidates will have a Doctoral degree (Ph.D.), publications, and demonstrated research competencies and capabilities commensurate with appointment levels in the department(s) of interest, as well as demonstrated interest in and experience with collaborative teaming and/or transdisciplinary efforts Successful candidates will be expected to develop and maintain externally funded research programs (individual and collaborative), engage in both undergraduate and graduate education, and contribute their leadership, partnering and innovative thinking towards global prominence in their respective discipline. Teaching opportunities will vary by department and teaching qualifications will be considered for fit within respective department(s).

Matrix completion is a problem that arises in many data-analysis settings where the input consists of a partially-observed matrix (e.g., recommender systems, traffic matrix analysis etc.). Classical approaches to matrix completion assume that the input partially-observed matrix is low rank. The success of these methods depends on the number of observed entries and the rank of the matrix; the larger the rank, the more entries need to be observed in order to accurately complete the matrix. In this paper, we deal with matrices that are not necessarily low rank themselves, but rather they contain low-rank submatrices. We propose Targeted, which is a general framework for completing such matrices. In this framework, we first extract the low-rank submatrices and then apply a matrix-completion algorithm to these low-rank submatrices as well as the remainder matrix separately. Although for the completion itself we use state-of-the-art completion methods, our results demonstrate that Targeted achieves significantly smaller reconstruction errors than other classical matrix-completion methods. One of the key technical contributions of the paper lies in the identification of the low-rank submatrices from the input partially-observed matrices.

Amid Syria's five-year-old civil war and Iraq's push to expel the Islamic State group from its major cities, President Barack Obama has quietly reneged on promises of "no boots on the ground" in recent years. But another American ground battle lingers just outside of the spotlight, in Somalia. A campaign involving private contractors, drone strikes and up to 300 U.S. Special Operations troops against the al Qaeda offshoot group al-Shabab has been escalating there over the past year, the New York Times reported Sunday, citing "senior American military officials." Operations in the country, located in the eastern "Horn of Africa," are expected to expand, according to the Times, on top of efforts that have involved the Navy's SEAL Team 6, weekly raids with troops from nearby Kenya and Uganda and interrogation of prisoners. The American use of force there hasn't exactly been welcome. At the end of September, for instance, Somalia's Security Minister Osman Issa accused the U.S. of killing 22 Somali soldiers in an airstrike, the result of bad intelligence information.