United States

Attempting to align AI capabilities and value structures by means of value elicitation from humans, such as through Reinforcement Learning from Human Feedback (RLHF), is a computational challenge that raises both psychological and philosophical questions. Adopting an evolutionary perspective on the emergence of value structures in humans and machine learning systems can offer a bridge between qualitative and quantitative aspects of alignment. Here, evolutionary dynamics are applied to a game-theoretic model of RLHF. This allows for formal reasoning about the process and capabilities that result from alignment training, even where quantitative benchmarks cannot be clearly defined. A simple parametrized game model of RLHF, subject to replicator dynamics, shows how the success of the training method is sensitive to bias in human judgments. Under ideal conditions, RHLF training leads to aligned behavior. If the choice pattern of the human judge is biased, the training instead incentivizes misalignment. This application shows that evolutionary analyses can contribute to improving the prospects for safety and support successful cooperation between humans and AI systems in deployment.

AAAI 2025

An Evolutionary Perspective on AI Alignment (Student Abstract)

### Welcome!
The 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) will take place in Bangkok, Thailand from August 11th to 16th, 2024. Our Virtual Poster Sessions will take place online Thursday, August 22, 2024.

You are required to register for this event. **Please register [here](https://2024.aclweb.org/registration). **

If you have already registered, please check your inbox for an email from Underline granting you access to ACL 2024 content.

Please register!

ACL 2024

The 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) will take place in Bangkok, Thailand from August 11th to 16th, 2024. More information will be announced soon.

poster

We are pleased to announce the Thirty-Ninth AAAI Conference on Artificial Intelligence (AAAI-25), which will be held in Philadelphia, Pennsylvania at the Pennsylvania Convention Center from February 25 to March 4, 2025.

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-25 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and a range of other activities to be announced.

### [Invited Speakers](https://aaai.org/conference/aaai/aaai-25/aaai-25-invited-speakers/)

Register [here](https://aaai.org/conference/aaai/aaai-25/registration/)

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-25 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and a range of other activities to be announced.



Recent advancements in single-cell sequencing technologies enable the measurement of multiple modalities in individual cells, offering insights into the transcriptome and regulome in various biological systems and human diseases in an unprecedented resolution. However, effectively using these ultra-high-dimensional and large-scale multiomic data to understand gene regulation remains challenging. Inspired by the success of adapting large language models into the genomics field, we develop scMBERT, a BERT framework-based pre-trained deep learning model using single-cell multiomic data. We showed that scMBERT increases model flexibility and performance in downstream tasks like cell type annotation and batch-effect correction, demonstrating the potential of leveraging multiomic data to improve single-cell genomic data analyses.

scMBERT: A Pre-Trained Deep Learning Model for Single-Cell Multiomic Data Representation and Prediction (Student Abstract)

Counterfactual explanations in Explainable AI (XAI) identify which features to change to alter an outcome, but existing methods adjust only the features of a single agent. We present a new approach to re-evaluating rankings that is based on predictions of future features of the other agents in a ranking system. It uses an algorithm that provides a more realistic counterfactual explanation of changing the ranking of a particular agent. Computer experiments demonstrated that the proposed algorithm can capture the time variation of the entire ranking system in the inference results.

Counterfactual Explanations of Time Varying Rankings (Student Abstract)

This paper addresses contention window optimization for multi-access scenarios. 
Our investigation into state-of-the-art models revealed that a limited number of nodes dominate the communication channels. Such monopolization issues are critical in networks as they can lead to significant disruptions. To mitigate this monopolization problem, we propose an imitation learning-based backoff mechanism. The proposed model is a reinforcement learning-based contention window optimization method. It imitates the expert's policy to ensure fair policy convergence for the agent and includes opportunities for weight adjustment to boost performance. The proposed model shows a fairness improvement of approximately 20% to 41% across various scenarios.

Imitation Learning Backoff: Reinforcement Learning-based Channel Access for Guaranteeing Fairness (Student Abstract)

Multi-agent reinforcement learning (MARL) trains multiple agents in shared environments. Recently, MARL models have significantly improved performance by leveraging sequential decision-making processes. Although these models can enhance performance, they do not explicitly con-sider the importance of the order in which agents make decisions. We propose AOAD-MAT, a novel model incorporating action decision sequence into learning. AOAD-MAT uses a Transformer-based actor-critic architecture to dynamically adjust agent action order. It introduces a subtask predicting the next agent to act, integrated into a PPO-based loss function. Experiments on StarCraft Multi-Agent Challenge and Multi-Agent MuJoCo benchmarks show AOAD-MAT out-performs existing models, demonstrating the effectiveness of adjusting agent order in MARL.

Sequential Order Adjustment of Action Decisions for Multi-Agent Transformer (Student Abstract)

Artificial neural networks (ANNs) struggle with continual learning, sacrificing performance on previously learned tasks to acquire new task knowledge. Here we propose a new approach allowing to mitigate catastrophic forgetting during continuous task learning. Typically a new task is trained until it reaches maximal performance, causing complete catastrophic forgetting of the previous tasks. In our new approach, termed Optimal Stopping (OS), network training on each new task continues only while the mean validation accuracy across all the tasks (current and previous) increases. The stopping criterion creates an explicit balance: lower performance on new tasks is accepted in exchange for preserving knowledge of previous tasks, resulting in higher overall network performance. The overall performance is further improved when OS is combined with Sleep Replay Consolidation (SRC), wherein the network converts to a Spiking Neural Network (SNN) and undergoes unsupervised learning modulated by Hebbian plasticity. During the SRC, the network spontaneously replays activation patterns from previous tasks, helping to maintain and restore prior task performance. This combined approach offers a promising avenue for enhancing the robustness and longevity of learned representations in continual learning models, achieving over twice the mean accuracy of baseline continuous learning while maintaining stable performance across tasks.

When to Learn and When to Stop: Quitting at the Optimal Time (Student Abstract)

The recent success of transformer language models owes much to their conversational fluency, which includes linguistic and morphological proficiency. An affine Taylor approximation has been found to be a good approximation for transformer computations over certain factual and encyclopedic relations. We show that the truly linear approximation W s, where s is a early layer representation of the base form and W is a local model derivative, is necessary and sufficient
to approximate morphological derivation, achieving above 80% top-1 accuracy across most morphological tasks in the Bigger Analogy Test Set. We argue that many morphological forms in transformer models are likely linearly encoded.

Linear Decoding of Morphology Relations in Language Models (Student Abstract)

FuSE-MET addresses critical challenges in deploying human activity recognition (HAR) systems in uncontrolled environments by effectively managing noisy labels, sparse data, and undefined activity vocabularies. By integrating BERT-based word embeddings with domain-specific knowledge (i.e., MET values), FuSE-MET optimizes label merging, reducing label complexity and improving classification accuracy. Our approach outperforms the state-of-the-art techniques, including ChatGPT-4, by balancing semantic meaning and physical intensity.

Domain-Informed Label Fusion Surpasses LLMs in Free-Living Activity Classification (Student Abstract)

Neural networks have emerged as powerful tools across various applications, yet their decision-making process often remains opaque, leading to them being perceived as "black boxes." This opacity raises concerns about their interpretability and reliability, especially in safety-critical scenarios. Network inversion techniques offer a solution by allowing us to peek inside these black boxes, revealing the features and patterns learned by the networks behind their decision-making processes and thereby provide valuable insights into how neural networks arrive at their conclusions, making them more interpretable and trustworthy. This paper presents a simple yet effective approach to network inversion using a meticulously conditioned generator that learns the data distribution in the input space of the trained neural network, enabling the reconstruction of inputs that would most likely lead to the desired outputs. To capture the diversity in the input space for a given output, instead of simply revealing the conditioning labels to the generator, we encode the conditioning label information into vectors and intermediate matrices and further minimize the cosine similarity between features of the generated images.

Network Inversion of Convolutional Neural Nets (Student Abstract)

We applied Risk-averse Reinforcement Learning (RL) to optimize investment portfolios while incorporating risk constraints. Given that portfolios must adhere to risk constraints set by investors and regulators, enforcing hard constraints is essential for practical portfolio optimization. Traditional techniques often lack the flexibility to model the complexities of dynamic financial markets. To address this, we used the Augmented Lagrangian Multiplier (ALM) to impose constraints on the agent, reducing risk during decision-making. Our risk-constrained RL algorithm demonstrated no constraint violations during testing and outperformed other Risk-averse RL methods, indicating its potential for optimizing portfolios for risk-averse investors.

Augmented Lagrangian Risk-constrained Reinforcement Learning for Portfolio Optimization (Student Abstract)

This paper proposes a novel quantum style transfer (QST) in hybrid quantum-classical computing. QST leverages quantum computing's ability to process high-dimensional data efficiently. Our approach aims to decrease both inference time and complexity while maintaining performance, presenting a viable solution that enhances the scalability and efficiency of image generation technologies.

Downloads

Next from AAAI 2025

scMBERT: A Pre-Trained Deep Learning Model for Single-Cell Multiomic Data Representation and Prediction (Student Abstract)

Similar lecture

Back to Basics: Revisiting REINFORCE-Style Optimization for Learning from Human Feedback in LLMs

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from AAAI 2025

scMBERT: A Pre-Trained Deep Learning Model for Single-Cell Multiomic Data Representation and Prediction (Student Abstract)

Similar lecture

Back to Basics: Revisiting REINFORCE-Style Optimization for Learning from Human Feedback in LLMs

Downloads