Yilun Liu, Minggui He, Feiyu Yao, Yuhe Ji, Shimin Tao, Jingzhou Du, Justin Li, Jian Gao, Zhang Li, Hao Yang, Boxing Chen, Osamu Yoshie. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Generative AI models have substantially improved the realism of synthetic media, yet their misuse through sophisticated DeepFakes poses significant risks. Despite recent advances in deepfake detection, fairness remains inadequately addressed, enabling deepfake markers to exploit biases against specific populations. While previous studies have emphasized group-level fairness, individual fairness (i.e., ensuring similar predictions for similar individuals) remains largely unexplored. In this work, we identify for the first time that the original principle of individual fairness fundamentally fails in the context of deepfake detection, revealing a critical gap previously unexplored in the literature. To mitigate it, we propose the first generalizable framework that can be integrated into existing deepfake detectors to enhance individual fairness and generalization. Extensive experiments conducted on leading deepfake datasets demonstrate that our approach significantly improves individual fairness while maintaining robust detection performance, outperforming state-of-the-art methods. The code is available at: https://github.com/Purdue-M2/Individual-Fairness-Deepfake-Detection.

The accurate estimation of cell type proportions in tissues is crucial for various downstream analyses. With the increasing availability of single-cell sequencing data, numerous deconvolution methods that use single-cell RNA sequencing data as a reference have been developed. However, a unified understanding of how these deconvolution approaches perform in practical applications is still lacking. To address this, we systematically assessed the accuracy and robustness of nine deconvolution methods that use single-cell RNA sequencing data as a reference, evaluating them on real bulk data with cell proportions verified through flow cytometry, as well as simulated bulk data generated from five single-cell RNA sequencing datasets. Our study highlights the importance of several factors-including reference dataset construction strategies, dataset size, cell type subdivision, and cell type inconsistency-on the accuracy and robustness of deconvolution results. We also propose a set of recommended guidelines for software users in diverse scenarios.

Introductory CS courses often present computational concepts in a multitude of varied contexts. The contents of these examples and assignments present an untapped opportunity to reach out to under-represented minorities. For CS in particular, the ubiquity of technology in modern society makes it imperative that students consider how seemingly innocuous technical decisions could selectively disadvantage particular populations. If the entirety of the CS community - from K-12 to industry - is to be inclusive, we must train students to be aware of these issues from their first CS course. This BOF aims to foster discussion of how examples and assignments can illustrate CS concepts and simultaneously acknowledge the challenges that under-represented students may be facing. At least three such avenues are possible: 1) Choosing multi-cultural contexts for homework questions and in-class examples; 2) Assigning projects with socioeconomic implications, such as the Schelling Segregation Model (a 2014 Nifty Assignment); and 3) Showing how inclusion is easily subverted by majority-Caucasian data sets or "purely technical" decisions such as representing gender with a Boolean. The long-term goal is to establish a network of instructors interested in contributing assignments and activities, and to create a resource suitable for broader dissemination in the future.

This paper presents the first functional evaluation of spontaneous, uncued retrieval from long-term memory in a cognitive architecture. The key insight is that current deliberate cued retrieval mechanisms require the agent to have knowledge of when and what to retrieve --- knowledge that may be missing or incorrect. Spontaneous uncued retrieval eliminates these requirements through automatic retrievals that use the agent's problem solving context as a heuristic for relevance, thus supplementing deliberate cued retrieval. Using constraints derived from this insight, we sketch the space of spontaneous retrieval mechanisms and describe an implementation of spontaneous retrieval in Soar together with an agent that takes advantage of that mechanism. Empirical evidence is provided in the Missing Link word-puzzle domain, where agents using spontaneous retrieval out-perform agents without that capability, leading us to conclude that spontaneous retrieval can be a useful mechanism and is worth further exploration.

Maintaining and pursuing multiple goals over varying time scales is an important ability for artificial agents in many cognitive architectures. Goals that remain suspended for long periods, however, are prone to be forgotten. This paper presents a class of preemptive strategies that allow agents to selectively retain goals in memory and to recover forgotten goals. Preemptive strategies work by retrieving and rehearsing goals at triggers, which are either periodic or are predictive of the opportunity to act. Since cognitive architectures contain common hierarchies of memory systems and share similar forgetting mechanisms, these strategies work across multiple architectures. We evaluate their effectiveness in a simulated mobile robot controlled by Soar, and demonstrate how preemptive strategies can be adapted to different environments and agents.

Episodic memory endows agents with numerous general cognitive capabilities, such as action modeling and virtual sensing. However, for long-lived agents, there are numerous unexplored computational challenges in supporting useful episodic-memory functions while maintaining real-time reactivity. In this paper, we review the implementation of episodic memory in Soar and present an expansive evaluation of that system. We demonstrate useful applications of episodic memory across a variety of domains, including games, mobile robotics, planning, and linguistics. In these domains, we characterize properties of environments, tasks, and episodic cues that affect performance, and evaluate the ability of Soar’s episodic memory to support hours to days of real-time operation.

One issue facing agents that accumulate large bodies of knowledge is determining whether they have knowl- edge that is relevant to its current goals. Performing comprehensive searches of long-term memory in every situation can be computationally expensive and disrup- tive to task reasoning. In this paper, we demonstrate that the recognition judgment — a heuristic for whether memory structures have been previously perceived — can serve as a low-cost indicator of the existence of potentially relevant knowledge. We present an approach for computing both context-dependent and context- independent recognition judgments using processes and data shared with declarative memories. We then de- scribe an initial, efficient implementation in the Soar cognitive architecture and evaluate our system in a word sense disambiguation task, showing that it reduces the number of memory searches without degrading agent performance.