Corey Clark

Eli Laird, Ayesh Madushanka, Elfi Kraka et al.

Progress in graph neural networks has grown rapidly in recent years, with many new developments in drug discovery, medical diagnosis, and recommender systems. While this progress is significant, many networks are `black boxes' with little understanding of the `what' exactly the network is learning. Many high-stakes applications, such as drug discovery, require human-intelligible explanations from the models so that users can recognize errors and discover new knowledge. Therefore, the development of explainable AI algorithms is essential for us to reap the benefits of AI. We propose an explainability algorithm for GNNs called eXplainable Insight (XInsight) that generates a distribution of model explanations using GFlowNets. Since GFlowNets generate objects with probabilities proportional to a reward, XInsight can generate a diverse set of explanations, compared to previous methods that only learn the maximum reward sample. We demonstrate XInsight by generating explanations for GNNs trained on two graph classification tasks: classifying mutagenic compounds with the MUTAG dataset and classifying acyclic graphs with a synthetic dataset that we have open-sourced. We show the utility of XInsight's explanations by analyzing the generated compounds using QSAR modeling, and we find that XInsight generates compounds that cluster by lipophilicity, a known correlate of mutagenicity. Our results show that XInsight generates a distribution of explanations that uncovers the underlying relationships demonstrated by the model. They also highlight the importance of generating a diverse set of explanations, as it enables us to discover hidden relationships in the model and provides valuable guidance for further analysis.

Kassi Nzalasse, Rishav Raj, Eli Laird et al.

As Artificial Intelligence applications expand, the evaluation of models faces heightened scrutiny. Ensuring public readiness requires evaluation datasets, which differ from training data by being disjoint and ethically sourced in compliance with privacy regulations. The performance and fairness of face recognition systems depend significantly on the quality and representativeness of these evaluation datasets. This data is sometimes scraped from the internet without user's consent, causing ethical concerns that can prohibit its use without proper releases. In rare cases, data is collected in a controlled environment with consent, however, this process is time-consuming, expensive, and logistically difficult to execute. This creates a barrier for those unable to conjure the immense resources required to gather ethically sourced evaluation datasets. To address these challenges, we introduce the Synthetic Identity Generation pipeline, or SIG, that allows for the targeted creation of ethical, balanced datasets for face recognition evaluation. Our proposed and demonstrated pipeline generates high-quality images of synthetic identities with controllable pose, facial features, and demographic attributes, such as race, gender, and age. We also release an open-source evaluation dataset named ControlFace10k, consisting of 10,008 face images of 3,336 unique synthetic identities balanced across race, gender, and age, generated using the proposed SIG pipeline. We analyze ControlFace10k along with a non-synthetic BUPT dataset using state-of-the-art face recognition algorithms to demonstrate its effectiveness as an evaluation tool. This analysis highlights the dataset's characteristics and its utility in assessing algorithmic bias across different demographic groups.

Steph Buongiorno, Lawrence Jake Klinkert, Tanishq Chawla et al.

This research introduces Procedural Artificial Narrative using Generative AI (PANGeA), a structured approach for leveraging large language models (LLMs), guided by a game designer's high-level criteria, to generate narrative content for turn-based role-playing video games (RPGs). Distinct from prior applications of LLMs used for video game design, PANGeA innovates by not only generating game level data (which includes, but is not limited to, setting, key items, and non-playable characters (NPCs)), but by also fostering dynamic, free-form interactions between the player and the environment that align with the procedural game narrative. The NPCs generated by PANGeA are personality-biased and express traits from the Big 5 Personality Model in their generated responses. PANGeA addresses challenges behind ingesting free-form text input, which can prompt LLM responses beyond the scope of the game narrative. A novel validation system that uses the LLM's intelligence evaluates text input and aligns generated responses with the unfolding narrative. Making these interactions possible, PANGeA is supported by a server that hosts a custom memory system that supplies context for augmenting generated responses thus aligning them with the procedural narrative. For its broad application, the server has a REST interface enabling any game engine to integrate directly with PANGeA, as well as an LLM interface adaptable with local or private LLMs. PANGeA's ability to foster dynamic narrative generation by aligning responses with the procedural narrative is demonstrated through an empirical study and ablation test of two versions of a demo game. These are, a custom, browser-based GPT and a Unity demo. As the results show, PANGeA holds potential to assist game designers in using LLMs to generate narrative-consistent content even when provided varied and unpredictable, free-form text input.

Lawrence J. Klinkert, Stephanie Buongiorno, Corey Clark

This research explores the potential of Large Language Models (LLMs) to utilize psychometric values, specifically personality information, within the context of video game character development. Affective Computing (AC) systems quantify a Non-Player character's (NPC) psyche, and an LLM can take advantage of the system's information by using the values for prompt generation. The research shows an LLM can consistently represent a given personality profile, thereby enhancing the human-like characteristics of game characters. Repurposing a human examination, the International Personality Item Pool (IPIP) questionnaire, to evaluate an LLM shows that the model can accurately generate content concerning the personality provided. Results show that the improvement of LLM, such as the latest GPT-4 model, can consistently utilize and interpret a personality to represent behavior.

Eli J. Laird, Corey Clark

World models enable agents to plan within imagined environments by predicting future states conditioned on past observations and actions. However, their ability to plan over long horizons is limited by the effective memory span of the backbone architecture. This limitation leads to perceptual drift in long rollouts, hindering the model's capacity to perform loop closures within imagined trajectories. In this work, we investigate the effective memory span of transformer-based world models through an analysis of several memory augmentation mechanisms. We introduce a taxonomy that distinguishes between memory encoding and memory injection mechanisms, motivating their roles in extending the world model's memory through the lens of residual stream dynamics. Using a state recall evaluation task, we measure the memory recall of each mechanism and analyze its respective trade-offs. Our findings show that memory mechanisms improve the effective memory span in vision transformers and provide a path to completing loop closures within a world model's imagination.

Hunjae Lee, Corey Clark

In decoder-only (causal) transformers, the computation graph created by causal masking routes information through both direct-path attention and indirect paths formed by intermediate tokens. We denote these indirect paths between token pairs as their runways. We argue that certain failure modes of causal transformers as observed by a growing body of recent works are likely exacerbated by a misalignment between these two information propagation modes. We formalize runway cascade as a phenomenon whereby this misalignment results in redundancies and irrelevant information cascading to token representations despite adequately learned attention patterns. As a solution, we propose runway-aware rewiring as a more explicit way of incorporating runway context directly into each token's direct-path attention. This mechanism re-wires the attention pattern for each token based on a summary of its runway landscape, enabling awareness of accumulating representational influences and allowing for more balanced information propagation. Our proposed methodology introduces no additional parameters and can seamlessly be integrated into standard attention mechanism. Empirically, our rewired transformer results in steady improvements in general language modeling as well as noticeably stronger information retrieval and extrapolation abilities compared to standard transformers.

Hunjae Lee, Corey Clark

Variable count is among the main scalability bottlenecks for transformer modeling in multivariate time series (MTS) data. On top of this, a growing consensus in the field points to indiscriminate inter-variable mixing as a potential source of noise-accumulation and performance degradation. This is likely exacerbated by sparsity of informative signals characteristic of many MTS systems coupled with representational misalignment stemming from indiscriminate information mixing between (heterogeneous) variables. While scalability and performance are often seen as competing interests in transformer design, we show that both can be improved simultaneously in MTS by strategically constraining the representational capacity of inter-variable mixing. Our proposed method, transformer with Delegate Token Attention (DELTAformer), constrains inter-variable modeling through what we call delegate tokens which are then used to perform full, unconstrained, inter-temporal modeling. Delegate tokens act as an implicit regularizer that forces the model to be highly selective about what inter-variable information is allowed to propagate through the network. Our results show that DELTAformer scales linearly with variable-count while actually outperforming standard transformers, achieving state-of-the-art performance across benchmarks and baselines. In addition, DELTAformer can focus on relevant signals better than standard transformers in noisy MTS environments and overall exhibit superior noise-resilience. Overall, results across various experiments confirm that by aligning our model design to leverage domain-specific challenges in MTS to our advantage, DELTAformer can simultaneously achieve linear scaling while actually improving its performance against standard, quadratic transformers.

Hunjae Lee, Corey Clark

Standard attention mechanisms in transformers employ static token representations that remain unchanged across all pair-wise computations in each layer. This limits their representational alignment with the potentially diverse relational dynamics of each token-pair interaction. While they excel in domains with relatively homogeneous relationships, standard attention's static relational learning struggles to capture the diverse, heterogeneous inter-channel dependencies of multivariate time series (MTS) data--where different channel-pair interactions within a single system may be governed by entirely different physical laws or temporal dynamics. To better align the attention mechanism for such domain phenomena, we propose attention with dynamic relational priming (prime attention). Unlike standard attention where each token presents an identical representation across all of its pair-wise interactions, prime attention tailors each token dynamically (or per interaction) through learnable modulations to best capture the unique relational dynamics of each token pair, optimizing each pair-wise interaction for that specific relationship. This representational plasticity of prime attention enables effective extraction of relationship-specific information in MTS while maintaining the same asymptotic computational complexity as standard attention. Our results demonstrate that prime attention consistently outperforms standard attention across benchmarks, achieving up to 6.5\% improvement in forecasting accuracy. In addition, we find that prime attention achieves comparable or superior performance using up to 40\% less sequence length compared to standard attention, further demonstrating its superior relational modeling capabilities.

Steph Buongiorno, Corey Clark

External knowledge graphs (KGs) can be used to augment large language models (LLMs), while simultaneously providing an explainable knowledge base of facts that can be inspected by a human. This approach may be particularly valuable in domains where explainability is critical, like human trafficking data analysis. However, creating KGs can pose challenges. KGs parsed from documents may comprise explicit connections (those directly stated by a document) but miss implicit connections (those obvious to a human although not directly stated). To address these challenges, this preliminary research introduces the GAME-KG framework, standing for "Gaming for Augmenting Metadata and Enhancing Knowledge Graphs." GAME-KG is a federated approach to modifying explicit as well as implicit connections in KGs by using crowdsourced feedback collected through video games. GAME-KG is shown through two demonstrations: a Unity test scenario from Dark Shadows, a video game that collects feedback on KGs parsed from US Department of Justice (DOJ) Press Releases on human trafficking, and a following experiment where OpenAI's GPT-4 is prompted to answer questions based on a modified and unmodified KG. Initial results suggest that GAME-KG can be an effective framework for enhancing KGs, while simultaneously providing an explainable set of structured facts verified by humans.