Searcharxiv⌕ Search

arXiv subjects

Miao Liu

Publications and source records attributed to Miao Liu.

At least 91 records · Page 5Linked to original sources
arXiv2024

Listen to Look into the Future: Audio-Visual Egocentric Gaze Anticipation

Egocentric gaze anticipation serves as a key building block for the emerging capability of Augmented Reality. Notably, gaze behavior is driven by both visual cues and audio signals during daily activities. Motivated by this observation, we introduce the first model that leverages both the video and audio modalities for egocentric gaze anticipation. Specifically, we propose a Contrastive Spatial-Temporal Separable (CSTS) fusion approach that adopts two modules to separately capture audio-visual correlations in spatial and temporal dimensions, and applies a contrastive loss on the re-weighted audio-visual features from fusion modules for representation learning. We conduct extensive ablation studies and thorough analysis using two egocentric video datasets: Ego4D and Aria, to validate our model design. We demonstrate the audio improves the performance by +2.5% and +2.4% on the two datasets. Our model also outperforms the prior state-of-the-art methods by at least +1.9% and +1.6%. Moreover, we provide visualizations to show the gaze anticipation results and provide additional insights into audio-visual representation learning. The code and data split are available on our website (https://bolinlai.github.io/CSTS-EgoGazeAnticipation/).

cs.CV↗
arXiv2024

LEGO: Learning EGOcentric Action Frame Generation via Visual Instruction Tuning

Generating instructional images of human daily actions from an egocentric viewpoint serves as a key step towards efficient skill transfer. In this paper, we introduce a novel problem -- egocentric action frame generation. The goal is to synthesize an image depicting an action in the user's context (i.e., action frame) by conditioning on a user prompt and an input egocentric image. Notably, existing egocentric action datasets lack the detailed annotations that describe the execution of actions. Additionally, existing diffusion-based image manipulation models are sub-optimal in controlling the state transition of an action in egocentric image pixel space because of the domain gap. To this end, we propose to Learn EGOcentric (LEGO) action frame generation via visual instruction tuning. First, we introduce a prompt enhancement scheme to generate enriched action descriptions from a visual large language model (VLLM) by visual instruction tuning. Then we propose a novel method to leverage image and text embeddings from the VLLM as additional conditioning to improve the performance of a diffusion model. We validate our model on two egocentric datasets -- Ego4D and Epic-Kitchens. Our experiments show substantial improvement over prior image manipulation models in both quantitative and qualitative evaluation. We also conduct detailed ablation studies and analysis to provide insights in our method. More details of the dataset and code are available on the website (https://bolinlai.github.io/Lego_EgoActGen/).

cs.CV↗
arXiv2024

Mitigating Gradient Bias in Multi-objective Learning: A Provably Convergent Stochastic Approach

Machine learning problems with multiple objective functions appear either in learning with multiple criteria where learning has to make a trade-off between multiple performance metrics such as fairness, safety and accuracy; or, in multi-task learning where multiple tasks are optimized jointly, sharing inductive bias between them. This problems are often tackled by the multi-objective optimization framework. However, existing stochastic multi-objective gradient methods and its variants (e.g., MGDA, PCGrad, CAGrad, etc.) all adopt a biased noisy gradient direction, which leads to degraded empirical performance. To this end, we develop a stochastic Multi-objective gradient Correction (MoCo) method for multi-objective optimization. The unique feature of our method is that it can guarantee convergence without increasing the batch size even in the non-convex setting. Simulations on multi-task supervised and reinforcement learning demonstrate the effectiveness of our method relative to state-of-the-art methods.

cs.LG↗
arXiv2024

Contextual Moral Value Alignment Through Context-Based Aggregation

Developing value-aligned AI agents is a complex undertaking and an ongoing challenge in the field of AI. Specifically within the domain of Large Language Models (LLMs), the capability to consolidate multiple independently trained dialogue agents, each aligned with a distinct moral value, into a unified system that can adapt to and be aligned with multiple moral values is of paramount importance. In this paper, we propose a system that does contextual moral value alignment based on contextual aggregation. Here, aggregation is defined as the process of integrating a subset of LLM responses that are best suited to respond to a user input, taking into account features extracted from the user's input. The proposed system shows better results in term of alignment to human value compared to the state of the art.

cs.AI↗
arXiv2024

GPTFF: A high-accuracy out-of-the-box universal AI force field for arbitrary inorganic materials

This study introduces a novel AI force field, namely graph-based pre-trained transformer force field (GPTFF), which can simulate arbitrary inorganic systems with good precision and generalizability. Harnessing a large trove of the data and the attention mechanism of transformer algorithms, the model can accurately predict energy, atomic forces, and stress with Mean Absolute Error (MAE) values of 32 meV/atom, 71 meV/Å, and 0.365 GPa, respectively. The dataset used to train the model includes 37.8 million single-point energies, 11.7 billion force pairs, and 340.2 million stresses. We also demonstrated that GPTFF can be universally used to simulate various physical systems, such as crystal structure optimization, phase transition simulations, and mass transport.

cond-mat.mtrl-sci↗
arXiv2024

A Neuro-Symbolic Approach to Multi-Agent RL for Interpretability and Probabilistic Decision Making

Multi-agent reinforcement learning (MARL) is well-suited for runtime decision-making in optimizing the performance of systems where multiple agents coexist and compete for shared resources. However, applying common deep learning-based MARL solutions to real-world problems suffers from issues of interpretability, sample efficiency, partial observability, etc. To address these challenges, we present an event-driven formulation, where decision-making is handled by distributed co-operative MARL agents using neuro-symbolic methods. The recently introduced neuro-symbolic Logical Neural Networks (LNN) framework serves as a function approximator for the RL, to train a rules-based policy that is both logical and interpretable by construction. To enable decision-making under uncertainty and partial observability, we developed a novel probabilistic neuro-symbolic framework, Probabilistic Logical Neural Networks (PLNN), which combines the capabilities of logical reasoning with probabilistic graphical models. In PLNN, the upward/downward inference strategy, inherited from LNN, is coupled with belief bounds by setting the activation function for the logical operator associated with each neural network node to a probability-respecting generalization of the Fréchet inequalities. These PLNN nodes form the unifying element that combines probabilistic logic and Bayes Nets, permitting inference for variables with unobserved states. We demonstrate our contributions by addressing key MARL challenges for power sharing in a system-on-chip application.

cs.AI↗
arXiv2024

Near optimal constructions of frameproof codes

Frameproof codes are a class of secure codes that were originally introduced in the pioneering work of Boneh and Shaw in the context of digital fingerprinting. They can be used to enhance the security and credibility of digital content. Let $M_{c,l}(q)$ denote the largest cardinality of a $q$-ary $c$-frameproof code with length $l$. Based on an intriguing observation that relates $M_{c,l}(q)$ to the renowned Erdős Matching Conjecture in extremal set theory, in 2003, Blackburn posed an open problem on the precise value of the limit $R_{c,l}=\lim_{q\rightarrow\infty}\frac{M_{c,l}(q)}{q^{\lceil l/c \rceil}}$. By combining several ideas from the probabilistic method, we present a lower bound for $M_{c,l}(q)$, which, together with an upper bound of Blackburn, completely determines $R_{c,l}$ for {\it all} fixed $c,l$, and resolves the above open problem in the full generality. We also present an improved upper bound for $M_{c,l}(q)$.

cs.IT↗
arXiv2024

Animated Stickers: Bringing Stickers to Life with Video Diffusion

We introduce animated stickers, a video diffusion model which generates an animation conditioned on a text prompt and static sticker image. Our model is built on top of the state-of-the-art Emu text-to-image model, with the addition of temporal layers to model motion. Due to the domain gap, i.e. differences in visual and motion style, a model which performed well on generating natural videos can no longer generate vivid videos when applied to stickers. To bridge this gap, we employ a two-stage finetuning pipeline: first with weakly in-domain data, followed by human-in-the-loop (HITL) strategy which we term ensemble-of-teachers. It distills the best qualities of multiple teachers into a smaller student model. We show that this strategy allows us to specifically target improvements to motion quality while maintaining the style from the static image. With inference optimizations, our model is able to generate an eight-frame video with high-quality, interesting, and relevant motion in under one second.

cs.CV↗
arXiv2024

Approximate generalized Steiner systems and near-optimal constant weight codes

Constant weight codes (CWCs) and constant composition codes (CCCs) are two important classes of codes that have been studied extensively in both combinatorics and coding theory for nearly sixty years. In this paper we show that for {\it all} fixed odd distances, there exist near-optimal CWCs and CCCs asymptotically achieving the classic Johnson-type upper bounds. Let $A_q(n,w,d)$ denote the maximum size of $q$-ary CWCs of length $n$ with constant weight $w$ and minimum distance $d$. One of our main results shows that for {\it all} fixed $q,w$ and odd $d$, one has $\lim_{n\rightarrow\infty}\frac{A_q(n,d,w)}{\binom{n}{t}}=\frac{(q-1)^t}{\binom{w}{t}}$, where $t=\frac{2w-d+1}{2}$. This implies the existence of near-optimal generalized Steiner systems originally introduced by Etzion, and can be viewed as a counterpart of a celebrated result of Rödl on the existence of near-optimal Steiner systems. Note that prior to our work, very little is known about $A_q(n,w,d)$ for $q\ge 3$. A similar result is proved for the maximum size of CCCs. We provide different proofs for our two main results, based on two strengthenings of the well-known Frankl-Rödl-Pippenger theorem on the existence of near-optimal matchings in hypergraphs: the first proof follows by Kahn's linear programming variation of the above theorem, and the second follows by the recent independent work of Delcour-Postle, and Glock-Joos-Kim-Kühn-Lichev on the existence of near-optimal matchings avoiding certain forbidden configurations. We also present several intriguing open questions for future research.

math.CO↗
arXiv2023

On the Convergence and Sample Complexity Analysis of Deep Q-Networks with $ε$-Greedy Exploration

This paper provides a theoretical understanding of Deep Q-Network (DQN) with the $\varepsilon$-greedy exploration in deep reinforcement learning. Despite the tremendous empirical achievement of the DQN, its theoretical characterization remains underexplored. First, the exploration strategy is either impractical or ignored in the existing analysis. Second, in contrast to conventional Q-learning algorithms, the DQN employs the target network and experience replay to acquire an unbiased estimation of the mean-square Bellman error (MSBE) utilized in training the Q-network. However, the existing theoretical analysis of DQNs lacks convergence analysis or bypasses the technical challenges by deploying a significantly overparameterized neural network, which is not computationally efficient. This paper provides the first theoretical convergence and sample complexity analysis of the practical setting of DQNs with $ε$-greedy policy. We prove an iterative procedure with decaying $ε$ converges to the optimal Q-value function geometrically. Moreover, a higher level of $ε$ values enlarges the region of convergence but slows down the convergence, while the opposite holds for a lower level of $ε$ values. Experiments justify our established theoretical insights on DQNs.

cs.LG↗
arXiv2023

MatChat: A Large Language Model and Application Service Platform for Materials Science

The prediction of chemical synthesis pathways plays a pivotal role in materials science research. Challenges, such as the complexity of synthesis pathways and the lack of comprehensive datasets, currently hinder our ability to predict these chemical processes accurately. However, recent advancements in generative artificial intelligence (GAI), including automated text generation and question-answering systems, coupled with fine-tuning techniques, have facilitated the deployment of large-scale AI models tailored to specific domains. In this study, we harness the power of the LLaMA2-7B model and enhance it through a learning process that incorporates 13,878 pieces of structured material knowledge data. This specialized AI model, named MatChat, focuses on predicting inorganic material synthesis pathways. MatChat exhibits remarkable proficiency in generating and reasoning with knowledge in materials science. Although MatChat requires further refinement to meet the diverse material design needs, this research undeniably highlights its impressive reasoning capabilities and innovative potential in the field of materials science. MatChat is now accessible online and open for use, with both the model and its application framework available as open source. This study establishes a robust foundation for collaborative innovation in the integration of generative AI in materials science.

cond-mat.mtrl-sci↗
arXiv2023

Weberite Na$_2$MM'F$_7$ (M,M'=Redox-Active Metal) as Promising Fluoride-Based Sodium-Ion Battery Cathodes

Sodium-ion batteries are a viable alternative to lithium-ion technology due to the plentiful sodium resources. However, certain commercialization challenges, such as low specific energies and poor cycling performance of current Na-ion cathodes, still need to be addressed. To overcome these hurdles, this study explored the potential of a novel class of fluoride-based materials, specifically trigonal-type Na$_2$MM'F$_7$ (M and M' are redox-active metals) belonging to the weberite-type compounds, as promising candidates for Na-ion cathodes. Through a comprehensive assessment utilizing ab initio calculations, twelve prospective compounds were identified, demonstrating high thermodynamic stability, large gravimetric capacities (>170 mAh/g), and low net Na-ion migration barriers (<600 meV). Significantly, ten out of the twelve screened compounds exhibit high specific energies exceeding 580 Wh/kg (approximately equals to the specific energy of LiFePO$_4$), indicating their exceptional electrochemical performance. This study will pave the way for further advancements in fluoride-based electrode materials.

physics.chem-ph↗
arXiv2023

Coagent Networks Revisited

Coagent networks formalize the concept of arbitrary networks of stochastic agents that collaborate to take actions in a reinforcement learning environment. Prominent examples of coagent networks in action include approaches to hierarchical reinforcement learning (HRL), such as those using options, which attempt to address the exploration exploitation trade-off by introducing abstract actions at different levels by sequencing multiple stochastic networks within the HRL agents. We first provide a unifying perspective on the many diverse examples that fall under coagent networks. We do so by formalizing the rules of execution in a coagent network, enabled by the novel and intuitive idea of execution paths in a coagent network. Motivated by parameter sharing in the hierarchical option-critic architecture, we revisit the coagent network theory and achieve a much shorter proof of the policy gradient theorem using our idea of execution paths, without any assumption on how parameters are shared among coagents. We then generalize our setting and proof to include the scenario where coagents act asynchronously. This new perspective and theorem also lead to more mathematically accurate and performant algorithms than those in the existing literature. Lastly, by running nonstationary RL experiments, we survey the performance and properties of different generalizations of option-critic models.

cs.LG↗
arXiv2023

Energy landscape and phase competition of CsV3Sb5-, CsV6Sb6-, and TbMn6Sn6-type Kagome materials

Finding viable Kagome lattices is vital for materializing novel phenomena in quantum materials. In this work, we performed element substitutions on CsV3Sb5 with space group P6/mmm, TbMn6Sn6 with space group P6/mmm, and CsV6Sb6 with space group R-3 m, respectively, as the parent compounds. A total of 4158 materials were obtained through element substitutions, and these materials were then calculated via density function theory in high-throughput mode. Afterward, 48 materials were identified with high thermodynamic stability (E_hull<5meV/atom). Furthermore, we compared the thermodynamic stability of three different phases with the same elemental composition and predicted some competing phases that may arise during material synthesis. Finally, by calculating the electronic structures of these materials, we attempted to identify patterns in the electronic structure variations as the elements change. This work provides guidance for discovering promising AM3X5/AM6X6 Kagome materials from a vast phase space.

cond-mat.supr-con↗
arXiv2023

Towards Few-shot Coordination: Revisiting Ad-hoc Teamplay Challenge In the Game of Hanabi

Cooperative Multi-agent Reinforcement Learning (MARL) algorithms with Zero-Shot Coordination (ZSC) have gained significant attention in recent years. ZSC refers to the ability of agents to coordinate zero-shot (without additional interaction experience) with independently trained agents. While ZSC is crucial for cooperative MARL agents, it might not be possible for complex tasks and changing environments. Agents also need to adapt and improve their performance with minimal interaction with other agents. In this work, we show empirically that state-of-the-art ZSC algorithms have poor performance when paired with agents trained with different learning methods, and they require millions of interaction samples to adapt to these new partners. To investigate this issue, we formally defined a framework based on a popular cooperative multi-agent game called Hanabi to evaluate the adaptability of MARL methods. In particular, we created a diverse set of pre-trained agents and defined a new metric called adaptation regret that measures the agent's ability to efficiently adapt and improve its coordination performance when paired with some held-out pool of partners on top of its ZSC performance. After evaluating several SOTA algorithms using our framework, our experiments reveal that naive Independent Q-Learning (IQL) agents in most cases adapt as quickly as the SOTA ZSC algorithm Off-Belief Learning (OBL). This finding raises an interesting research question: How to design MARL algorithms with high ZSC performance and capability of fast adaptation to unseen partners. As a first step, we studied the role of different hyper-parameters and design choices on the adaptability of current MARL algorithms. Our experiments show that two categories of hyper-parameters controlling the training data diversity and optimization process have a significant impact on the adaptability of Hanabi agents.

cs.LG↗
arXiv2023

Selenium / Tellurium Two-Dimensional Structures: from Isovalent Se Dopants in Te to Atomically Thin Se Films

Two-dimensional (2D) elemental semiconductors have great potential for device applications, but their performance is limited by the lack of efficient doping methods. Here, combining molecular beam epitaxy, scanning tunneling microscopy/spectroscopy, X-ray photoelectron spectroscopy, and density functional theory calculations, we investigate the evolution of the structural and electronic properties of 2D selenium/tellurium films with increased Se dosages on graphene/6H-SiC(0001) substrates. We found that Se atoms form isovalent dopants by replacing surface Te atoms, which introduces efficient electron doping and lowers the work function of Te films. With the Se dosage increasing, two types of elemental 2D crystalline Se structures, trigonal Se and Se8 molecular assembly films, are obtained on ultrathin Te films, which are distinct from the amorphous Se acquired by depositing Se directly on graphene/6H-SiC(0001). Our results shed light on tuning the electronic properties of 2D elemental semiconductors by isovalent doping and constructing heterostructures of isovalent 2D elemental materials.

cond-mat.mtrl-sci↗
arXiv2023

Electron-phonon interactions in LuH$_2$, LuH$_3$, and LuN

This paper presents the calculation results of electron-phonon interactions within the LuH$_2$, LuH$_3$, and LuN systems under 0 GPa and 10 GPa via density functional theory at the GGA-PBE level. The purpose of this work is to provide useful data that may be of the interests of the superconducting community as it was reported that the Lu-H-N compound is likely to be a room-temperature superconductor under 1 GPa [Nature, 615, 244 (2023)].

cond-mat.supr-con↗