Datasets:

dwb2023

/

arxiv-papers-dataset

like 0

Dynamics of An Information Theoretic Analog of Two Masses on a Spring

2024-07-03T12:47:16+00:00

2024-07-03T12:47:16+00:00

http://arxiv.org/pdf/2407.03074v1

http://arxiv.org/abs/2407.03074v1

In this letter we investigate an information theoretic analogue of the classic two masses on spring system, arising from a physical interpretation of Friston's free energy principle in the theory of learning in a system of agents. Using methods from classical mechanics on manifolds, we define a kinetic energy term using the Fisher metric on distributions and a potential energy function defined in terms of stress on the agents' beliefs. The resulting Lagrangian (Hamiltonian) produces a variation of the classic DeGroot dynamics. In the two agent case, the potential function is defined using the Jeffrey's divergence and the resulting dynamics are characterized by a non-linear spring. These dynamics produce trajectories that resemble flows on tori but are shown numerically to produce chaos near the boundary of the space. We then investigate persuasion as an information theoretic control problem where analysis indicates that manipulating peer pressure with a fixed target is a more stable approach to altering an agent's belief than providing a slowly changing belief state that approaches the target.

physics.soc-ph

http://arxiv.org/abs/2407.03074v1

Teams of LLM Agents can Exploit Zero-Day Vulnerabilities

2024-06-02T16:25:26+00:00

2024-06-02T16:25:26+00:00

http://arxiv.org/pdf/2406.01637v1

http://arxiv.org/abs/2406.01637v1

LLM agents have become increasingly sophisticated, especially in the realm of cybersecurity. Researchers have shown that LLM agents can exploit real-world vulnerabilities when given a description of the vulnerability and toy capture-the-flag problems. However, these agents still perform poorly on real-world vulnerabilities that are unknown to the agent ahead of time (zero-day vulnerabilities). In this work, we show that teams of LLM agents can exploit real-world, zero-day vulnerabilities. Prior agents struggle with exploring many different vulnerabilities and long-range planning when used alone. To resolve this, we introduce HPTSA, a system of agents with a planning agent that can launch subagents. The planning agent explores the system and determines which subagents to call, resolving long-term planning issues when trying different vulnerabilities. We construct a benchmark of 15 real-world vulnerabilities and show that our team of agents improve over prior work by up to 4.5$\times$.

cs.MA

http://arxiv.org/abs/2406.01637v1

Unified Opinion Dynamic Modeling as Concurrent Set Relations in Rewriting Logic

2024-02-14T08:52:06+00:00

2024-02-14T08:52:06+00:00

http://arxiv.org/pdf/2402.09021v1

http://arxiv.org/abs/2402.09021v1

Social media platforms have played a key role in weaponizing the polarization of social, political, and democratic processes. This is, mainly, because they are a medium for opinion formation. Opinion dynamic models are a tool for understanding the role of specific social factors on the acceptance/rejection of opinions because they can be used to analyze certain assumptions on human behaviors. This work presents a framework that uses concurrent set relations as the formal basis to specify, simulate, and analyze social interaction systems with dynamic opinion models. Standard models for social learning are obtained as particular instances of the proposed framework. It has been implemented in the Maude system as a fully executable rewrite theory that can be used to better understand how opinions of a system of agents can be shaped. This paper also reports an initial exploration in Maude on the use of reachability analysis, probabilistic simulation, and statistical model checking of important properties related to opinion dynamic models.

cs.LO

http://arxiv.org/abs/2402.09021v1

Characterising and Verifying the Core in Concurrent Multi-Player Mean-Payoff Games (Full Version)

2023-11-27T14:53:26+00:00

2023-11-27T14:53:26+00:00

http://arxiv.org/pdf/2311.15883v1

http://arxiv.org/abs/2311.15883v1

Concurrent multi-player mean-payoff games are important models for systems of agents with individual, non-dichotomous preferences. Whilst these games have been extensively studied in terms of their equilibria in non-cooperative settings, this paper explores an alternative solution concept: the core from cooperative game theory. This concept is particularly relevant for cooperative AI systems, as it enables the modelling of cooperation among agents, even when their goals are not fully aligned. Our contribution is twofold. First, we provide a characterisation of the core using discrete geometry techniques and establish a necessary and sufficient condition for its non-emptiness. We then use the characterisation to prove the existence of polynomial witnesses in the core. Second, we use the existence of such witnesses to solve key decision problems in rational verification and provide tight complexity bounds for the problem of checking whether some/every equilibrium in a game satisfies a given LTL or GR(1) specification. Our approach is general and can be adapted to handle other specifications expressed in various fragments of LTL without incurring additional computational costs.

cs.GT

http://arxiv.org/abs/2311.15883v1

Continuification control of large-scale multiagent systems under limited sensing and structural perturbations

2023-03-23T13:26:43+00:00

2023-03-23T13:26:43+00:00

http://arxiv.org/pdf/2303.13246v1

http://arxiv.org/abs/2303.13246v1

We investigate the stability and robustness properties of a continuification-based strategy for the control of large-scale multiagent systems. Within continuation-based strategy, one transforms the microscopic, agent-level description of the system dynamics into a macroscopic continuum-level, for which a control action can be synthesized to steer the macroscopic dynamics towards a desired distribution. Such an action is ultimately discretized to obtain a set of deployable control inputs for the agents to achieve the goal. The mathematical proof of convergence toward the desired distribution typically relies on the assumptions that no disturbance is present and that each agent possesses global knowledge of all the others' positions. Here, we analytically and numerically address the possibility of relaxing these assumptions for the case of a one-dimensional system of agents moving in a ring. We offer compelling evidence in favor of the use of a continuification-based strategy when agents only possess a finite sensing capability and spatio-temporal perturbations affect the macroscopic dynamics of the ensemble. We also discuss some preliminary results about the role of an integral action in the macroscopic control solution.

eess.SY

http://arxiv.org/abs/2303.13246v1

Control of multi-agent systems: results, open problems, and applications

2023-02-23T20:06:45+00:00

2023-02-23T20:06:45+00:00

http://arxiv.org/pdf/2302.12308v1

http://arxiv.org/abs/2302.12308v1

The purpose of this review paper is to present some recent results on the modeling and control of large systems of agents. We focus on particular applications where the agents are capable of independent actions instead of simply reacting to external forces. In the literature, such agents were referred to as autonomous, intelligent, self-propelled, greedy, and others. The main applications we have in mind are social systems (as opinion dynamics), pedestrian movements (also called crowd dynamics), animal groups, and vehicular traffic. Also, the control problems posed by such systems are new and require innovative methods. We illustrate some ideas, developed recently, including the use of sparse controls, limiting the total variation of controls, and defining new control problems for measures. After reviewing various approaches, we discuss some future research directions of potential interest. The latter encompasses both new types of equations, as well as new types of limiting procedures to connect several scales at which a system can be represented. We conclude by illustrating a recent real-life experiment using autonomous vehicles on an open highway to smooth traffic waves. This opens the door to a new era of interventions to control in real-time multi-agent systems and to increase the societal impact of such interventions guided by control research.

math.OC

http://arxiv.org/abs/2302.12308v1

Co-evolution of Social and Non-Social Guilt

2023-02-20T09:40:49+00:00

2023-02-20T09:40:49+00:00

http://arxiv.org/pdf/2302.09859v1

http://arxiv.org/abs/2302.09859v1

Building ethical machines may involve bestowing upon them the emotional capacity to self-evaluate and repent on their actions. While reparative measures, such as apologies, are often considered as possible strategic interactions, the explicit evolution of the emotion of guilt as a behavioural phenotype is not yet well understood. Here, we study the co-evolution of social and non-social guilt of homogeneous or heterogeneous populations, including well-mixed, lattice and scale-free networks. Socially aware guilt comes at a cost, as it requires agents to make demanding efforts to observe and understand the internal state and behaviour of others, while non-social guilt only requires the awareness of the agents' own state and hence incurs no social cost. Those choosing to be non-social are however more sensitive to exploitation by other agents due to their social unawareness. Resorting to methods from evolutionary game theory, we study analytically, and through extensive numerical and agent-based simulations, whether and how such social and non-social guilt can evolve and deploy, depending on the underlying structure of the populations, or systems, of agents. The results show that, in both lattice and scale-free networks, emotional guilt prone strategies are dominant for a larger range of the guilt and social costs incurred, compared to the well-mixed population setting, leading therefore to significantly higher levels of cooperation for a wider range of the costs. In structured population settings, both social and non-social guilt can evolve and deploy through clustering with emotional prone strategies, allowing them to be protected from exploiters, especially in case of non-social (less costly) strategies. Overall, our findings provide important insights into the design and engineering of self-organised and distributed cooperative multi-agent systems.

cs.MA

http://arxiv.org/abs/2302.09859v1

Improving generalization in reinforcement learning through forked agents

2022-12-13T09:47:28+00:00

2022-12-29T17:21:47+00:00

http://arxiv.org/pdf/2212.06451v3

http://arxiv.org/abs/2212.06451v3

An eco-system of agents each having their own policy with some, but limited, generalizability has proven to be a reliable approach to increase generalization across procedurally generated environments. In such an approach, new agents are regularly added to the eco-system when encountering a new environment that is outside of the scope of the eco-system. The speed of adaptation and general effectiveness of the eco-system approach highly depends on the initialization of new agents. In this paper we propose different initialization techniques, inspired from Deep Neural Network initialization and transfer learning, and study their impact.

cs.AI

http://arxiv.org/abs/2212.06451v3

On the optimal control of kinetic epidemic models with uncertain social features

2022-10-17T15:57:16+00:00

2023-06-08T11:49:50+00:00

http://arxiv.org/pdf/2210.09201v3

http://arxiv.org/abs/2210.09201v3

It is recognized that social heterogeneities in terms of the contact distribution have a strong influence on the spread of infectious diseases. Nevertheless, few data are available on the group composition of social contacts, and their statistical description does not possess universal patterns and may vary spatially and temporally. It is therefore essential to design robust control strategies, mimicking the effects of non-pharmaceutical interventions, to limit efficiently the number of infected cases. In this work, starting from a recently introduced kinetic model for epidemiological dynamics that takes into account the impact of social contacts of individuals, we consider an uncertain contact formation dynamics leading to slim-tailed as well as fat-tailed distributions of contacts. Hence, we analyse the effects of an optimally robust control strategy of the system of agents. Thanks to classical methods of kinetic theory, we couple uncertainty quantification methods with the introduced mathematical model to assess the effects of social limitations. Finally, using the proposed modelling approach and starting from available data, we show the effectiveness of the proposed selective measures to dampen uncertainties together with the epidemic trends.

math.OC

http://arxiv.org/abs/2210.09201v3

AB/Push-Pull Method for Distributed Optimization in Time-Varying Directed Networks

2022-09-14T23:39:51+00:00

2022-09-26T16:10:18+00:00

http://arxiv.org/pdf/2209.06974v2

http://arxiv.org/abs/2209.06974v2

In this paper, we study the distributed optimization problem for a system of agents embedded in time-varying directed communication networks. Each agent has its own cost function and agents cooperate to determine the global decision that minimizes the summation of all individual cost functions. We consider the so-called push-pull gradient-based algorithm (termed as AB/Push-Pull) which employs both row- and column-stochastic weights simultaneously to track the optimal decision and the gradient of the global cost while ensuring consensus and optimality. We show that the algorithm converges linearly to the optimal solution over a time-varying directed network for a constant stepsize when the agent's cost function is smooth and strongly convex. The linear convergence of the method has been shown in Saadatniaki et al. (2020), where the multi-step consensus contraction parameters for row- and column-stochastic mixing matrices are not directly related to the underlying graph structure, and the explicit range for the stepsize value is not provided. With respect to Saadatniaki et al. (2020), the novelty of this work is twofold: (1) we establish the one-step consensus contraction for both row- and column-stochastic mixing matrices with the contraction parameters given explicitly in terms of the graph diameter and other graph properties; and (2) we provide explicit upper bounds for the stepsize value in terms of the properties of the cost functions, the mixing matrices, and the graph connectivity structure.

math.OC

http://arxiv.org/abs/2209.06974v2

Taming Multi-Agent Reinforcement Learning with Estimator Variance Reduction

2022-09-02T13:44:00+00:00

2023-06-22T14:19:41+00:00

http://arxiv.org/pdf/2209.01054v2

http://arxiv.org/abs/2209.01054v2

Centralised training with decentralised execution (CT-DE) serves as the foundation of many leading multi-agent reinforcement learning (MARL) algorithms. Despite its popularity, it suffers from a critical drawback due to its reliance on learning from a single sample of the joint-action at a given state. As agents explore and update their policies during training, these single samples may poorly represent the actual joint-policy of the system of agents leading to high variance gradient estimates that hinder learning. To address this problem, we propose an enhancement tool that accommodates any actor-critic MARL method. Our framework, Performance Enhancing Reinforcement Learning Apparatus (PERLA), introduces a sampling technique of the agents' joint-policy into the critics while the agents train. This leads to TD updates that closely approximate the true expected value under the current joint-policy rather than estimates from a single sample of the joint-action at a given state. This produces low variance and precise estimates of expected returns, minimising the variance in the critic estimators which typically hinders learning. Moreover, as we demonstrate, by eliminating much of the critic variance from the single sampling of the joint policy, PERLA enables CT-DE methods to scale more efficiently with the number of agents. Theoretically, we prove that PERLA reduces variance in value estimates similar to that of decentralised training while maintaining the benefits of centralised training. Empirically, we demonstrate PERLA's superior performance and ability to reduce estimator variance in a range of benchmarks including Multi-agent Mujoco, and StarCraft II Multi-agent Challenge.

cs.MA

http://arxiv.org/abs/2209.01054v2

Agency problem and mean field system of agents with moral hazard, synergistic effects and accidents

2022-07-22T13:53:50+00:00

2022-07-22T13:53:50+00:00

http://arxiv.org/pdf/2207.11087v1

http://arxiv.org/abs/2207.11087v1

We investigate the existence of an optimal policy to monitor a mean field systems of agents managing a risky project under moral hazard with accidents modeled by L\'evy processes magnified by the law of the project. We provide a general method to find both a mean field equilibrium for the agents and the optimal compensation policy under general, sufficient and necessary assumptions on all the parameters. We formalize the problem as a bilevel optimization with the probabilistic version of a mean field games which can be reduced to a controlled McKean-Vlasov SDE with jumps. We apply our results to an optimal energy demand-response problem with a crowd of consumers subjected to power cut/shortage when the variability of the energy consumption is too high under endogenous or exogenous strains. In this example, we get explicit solution to the mean field game and to the McKean-Vlasov equation with jumps

math.OC

http://arxiv.org/abs/2207.11087v1

Topological states and continuum model for swarmalators without force reciprocity

2022-05-31T12:26:08+00:00

2022-05-31T12:26:08+00:00

http://arxiv.org/pdf/2205.15739v1

http://arxiv.org/abs/2205.15739v1

Swarmalators are systems of agents which are both self-propelled particles and oscillators. Each particle is endowed with a phase which modulates its interaction force with the other particles. In return, relative positions modulate phase synchronization between interacting particles. In the present model, there is no force reciprocity: when a particle attracts another one, the latter repels the former. This results in a pursuit behavior. In this paper, we derive a hydrodynamic model of this swarmalator system and show that it has explicit doubly-periodic travelling-wave solutions in two space dimensions. These special solutions enjoy non-trivial topology quantified by the index of the phase vector along a period in either dimension. Stability of these solutions is studied by investigating the conditions for hyperbolicity of the model. Numerical solutions of both the particle and hydrodynamic models are shown. They confirm the consistency of the hydrodynamic model with the particle one for small times or large phase-noise but also reveal the emergence of intriguing patterns in the case of small phase-noise.

math-ph

http://arxiv.org/abs/2205.15739v1

Improving generalization to new environments and removing catastrophic forgetting in Reinforcement Learning by using an eco-system of agents

2022-04-13T17:52:54+00:00

2022-07-01T08:56:03+00:00

http://arxiv.org/pdf/2204.06550v2

http://arxiv.org/abs/2204.06550v2

Adapting a Reinforcement Learning (RL) agent to an unseen environment is a difficult task due to typical over-fitting on the training environment. RL agents are often capable of solving environments very close to the trained environment, but when environments become substantially different, their performance quickly drops. When agents are retrained on new environments, a second issue arises: there is a risk of catastrophic forgetting, where the performance on previously seen environments is seriously hampered. This paper proposes a novel approach that exploits an eco-system of agents to address both concerns. Hereby, the (limited) adaptive power of individual agents is harvested to build a highly adaptive eco-system.

cs.AI

http://arxiv.org/abs/2204.06550v2

The consensus problem for opinion dynamics with local average random interactions

2022-04-12T10:29:38+00:00

2022-04-12T10:29:38+00:00

http://arxiv.org/pdf/2204.05689v1

http://arxiv.org/abs/2204.05689v1

We study the consensus formation for an agents based model, generalizing that originally proposed by Krause \cite{Kr}, by allowing the communication channels between any couple of agents to be switched on or off randomly, at each time step, with a probability law depending on the proximity of the agents' opinions. Namely, we consider a system of agents sharing their opinions according to the following updating protocol. At time $t+1$ the opinion $X_{i}\left( t+1\right) \in\left[ 0,1\right] $ of agent $i$ is updated at the weighted average of the opinions of the agents communicating with it at time $t.$ The weights model the confidence level an agent assign to the opinions of the other agents and are kept fixed by the system dynamics, but the set of agents communicating with any agent $i$ at time $t+1$ is randomly updated in such a way that the agent $j$ can be chosen to belong to this set independently of the other agents with a probability that is a non increasing function of $\left\vert X_{i}\left( t\right) -X_{j}\left( t\right) \right\vert .$ This condition models the fact that a communication among the agents is more likely to happen if their opinions are close. We prove that the system reaches consensus, i.e. as the time tends to infinity the agents' opinions will reach the same value exponentially fast.

cs.SI

http://arxiv.org/abs/2204.05689v1

Phase Transition of the 3-Majority Dynamics with Uniform Communication Noise

2021-12-07T07:37:46+00:00

2021-12-13T10:29:09+00:00

http://arxiv.org/pdf/2112.03543v2

http://arxiv.org/abs/2112.03543v2

Communication noise is a common feature in several real-world scenarios where systems of agents need to communicate in order to pursue some collective task. In particular, many biologically inspired systems that try to achieve agreements on some opinion must implement resilient dynamics that are not strongly affected by noisy communications. In this work, we study the popular 3-Majority dynamics, an opinion dynamics which has been proved to be an efficient protocol for the majority consensus problem, in which we introduce a simple feature of uniform communication noise, following (d'Amore et al. 2020). We prove that in the fully connected communication network of n agents and in the binary opinion case, the process induced by the 3-Majority dynamics exhibits a phase transition. For a noise probability $p < 1/3$, the dynamics reaches in logarithmic time an almost-consensus metastable phase which lasts for a polynomial number of rounds with high probability. Furthermore, departing from previous analyses, we further characterize this phase by showing that there exists an attractive equilibrium value $s_{\text{eq}} \in [n]$ for the bias of the system, i.e. the difference between the majority community size and the minority one. Moreover, the agreement opinion turns out to be the initial majority one if the bias towards it is of magnitude $\Omega(\sqrt{n\log n})$ in the initial configuration. If, instead, $p > 1/3$, no form of consensus is possible, and any information regarding the initial majority opinion is lost in logarithmic time with high probability. Despite more communications per-round are allowed, the 3-Majority dynamics surprisingly turns out to be less resilient to noise than the Undecided-State dynamics (d'Amore et al. 2020), whose noise threshold value is $p = 1/2$.

cs.DC

http://arxiv.org/abs/2112.03543v2

Bootstrapping of memetic from genetic evolution via inter-agent selection pressures

2021-04-07T21:31:05+00:00

2021-04-07T21:31:05+00:00

http://arxiv.org/pdf/2104.03404v1

http://arxiv.org/abs/2104.03404v1

We create an artificial system of agents (attention-based neural networks) which selectively exchange messages with each-other in order to study the emergence of memetic evolution and how memetic evolutionary pressures interact with genetic evolution of the network weights. We observe that the ability of agents to exert selection pressures on each-other is essential for memetic evolution to bootstrap itself into a state which has both high-fidelity replication of memes, as well as continuing production of new memes over time. However, in this system there is very little interaction between this memetic 'ecology' and underlying tasks driving individual fitness - the emergent meme layer appears to be neither helpful nor harmful to agents' ability to learn to solve tasks. Sourcecode for these experiments is available at https://github.com/GoodAI/memes

cs.AI

http://arxiv.org/abs/2104.03404v1

Hybrid Information-driven Multi-agent Reinforcement Learning

2021-02-01T17:28:39+00:00

2021-02-01T17:28:39+00:00

http://arxiv.org/pdf/2102.01004v1

http://arxiv.org/abs/2102.01004v1

Information theoretic sensor management approaches are an ideal solution to state estimation problems when considering the optimal control of multi-agent systems, however they are too computationally intensive for large state spaces, especially when considering the limited computational resources typical of large-scale distributed multi-agent systems. Reinforcement learning (RL) is a promising alternative which can find approximate solutions to distributed optimal control problems that take into account the resource constraints inherent in many systems of distributed agents. However, the RL training can be prohibitively inefficient, especially in low-information environments where agents receive little to no feedback in large portions of the state space. We propose a hybrid information-driven multi-agent reinforcement learning (MARL) approach that utilizes information theoretic models as heuristics to help the agents navigate large sparse state spaces, coupled with information based rewards in an RL framework to learn higher-level policies. This paper presents our ongoing work towards this objective. Our preliminary findings show that such an approach can result in a system of agents that are approximately three orders of magnitude more efficient at exploring a sparse state space than naive baseline metrics. While the work is still in its early stages, it provides a promising direction for future research.

cs.MA

http://arxiv.org/abs/2102.01004v1

Synthesizing Decentralized Controllers with Graph Neural Networks and Imitation Learning

2020-12-29T18:59:14+00:00

2022-03-23T13:50:12+00:00

http://arxiv.org/pdf/2012.14906v4

http://arxiv.org/abs/2012.14906v4

Dynamical systems consisting of a set of autonomous agents face the challenge of having to accomplish a global task, relying only on local information. While centralized controllers are readily available, they face limitations in terms of scalability and implementation, as they do not respect the distributed information structure imposed by the network system of agents. Given the difficulties in finding optimal decentralized controllers, we propose a novel framework using graph neural networks (GNNs) to \emph{learn} these controllers. GNNs are well-suited for the task since they are naturally distributed architectures and exhibit good scalability and transferability properties. We show that GNNs learn appropriate decentralized controllers by means of imitation learning, leverage their permutation invariance properties to successfully scale to larger teams and transfer to unseen scenarios at deployment time. The problems of flocking and multi-agent path planning are explored to illustrate the potential of GNNs in learning decentralized controllers.

cs.LG

http://arxiv.org/abs/2012.14906v4

Guidance of Agents in Cyclic Pursuit

2020-07-02T08:04:14+00:00

2020-07-02T08:04:14+00:00

http://arxiv.org/pdf/2007.00949v1

http://arxiv.org/abs/2007.00949v1

This report studies the emergent behavior of systems of agents performing cyclic pursuit controlled by an external broadcast signal detected by a random set of the agents. Two types of cyclic pursuit are analyzed: 1)linear cyclic pursuit, where each agent senses the relative position of its target or leading agent 2)non-linear cyclic pursuit, where the agents can sense only bearing to their leading agent and colliding agents merge and continue on the path of the pursued agent (a so-called "bugs" model). Cyclic pursuit is, in both cases, a gathering algorithm, which has been previously analyzed. The novelty of our work is the derivation of emergent behaviours, in both linear and non-linear cyclic pursuit, in the presence of an exogenous broadcast control detected by a random subset of agents. We show that the emergent behavior of the swarm depends on the type of cyclic pursuit. In the linear case, the agents asymptotically align in the desired direction and move with a common speed which is a proportional to the ratio of the number of agents detecting the broadcast control to the total number of agents in the swarm, for any magnitude of input (velocity) signal. In the non-linear case, the agents gather and move with a shared velocity, which equals the input velocity signal, independently of the number of agents detecting the broadcast signal.

cs.RO

http://arxiv.org/abs/2007.00949v1

Learning interaction kernels in heterogeneous systems of agents from multiple trajectories

2019-10-10T19:54:04+00:00

2020-07-15T00:48:14+00:00

http://arxiv.org/pdf/1910.04832v3

http://arxiv.org/abs/1910.04832v3

Systems of interacting particles or agents have wide applications in many disciplines such as Physics, Chemistry, Biology and Economics. These systems are governed by interaction laws, which are often unknown: estimating them from observation data is a fundamental task that can provide meaningful insights and accurate predictions of the behaviour of the agents. In this paper, we consider the inverse problem of learning interaction laws given data from multiple trajectories, in a nonparametric fashion, when the interaction kernels depend on pairwise distances. We establish a condition for learnability of interaction kernels, and construct estimators that are guaranteed to converge in a suitable $L^2$ space, at the optimal min-max rate for 1-dimensional nonparametric regression. We propose an efficient learning algorithm based on least squares, which can be implemented in parallel for multiple trajectories and is therefore well-suited for the high dimensional, big data regime. Numerical simulations on a variety examples, including opinion dynamics, predator-swarm dynamics and heterogeneous particle dynamics, suggest that the learnability condition is satisfied in models used in practice, and the rate of convergence of our estimator is consistent with the theory. These simulations also suggest that our estimators are robust to noise in the observations, and produce accurate predictions of dynamics in relative large time intervals, even when they are learned from data collected in short time intervals.

stat.ML

http://arxiv.org/abs/1910.04832v3

The sharp, the flat and the shallow: Can weakly interacting agents learn to escape bad minima?

2019-05-10T12:37:51+00:00

2019-05-10T12:37:51+00:00

http://arxiv.org/pdf/1905.04121v1

http://arxiv.org/abs/1905.04121v1

An open problem in machine learning is whether flat minima generalize better and how to compute such minima efficiently. This is a very challenging problem. As a first step towards understanding this question we formalize it as an optimization problem with weakly interacting agents. We review appropriate background material from the theory of stochastic processes and provide insights that are relevant to practitioners. We propose an algorithmic framework for an extended stochastic gradient Langevin dynamics and illustrate its potential. The paper is written as a tutorial, and presents an alternative use of multi-agent learning. Our primary focus is on the design of algorithms for machine learning applications; however the underlying mathematical framework is suitable for the understanding of large scale systems of agent based models that are popular in the social sciences, economics and finance.

stat.ML

http://arxiv.org/abs/1905.04121v1

Nonparametric inference of interaction laws in systems of agents from trajectory data

2018-12-14T16:15:36+00:00

2019-03-23T18:18:39+00:00

http://arxiv.org/pdf/1812.06003v4

http://arxiv.org/abs/1812.06003v4

Inferring the laws of interaction between particles and agents in complex dynamical systems from observational data is a fundamental challenge in a wide variety of disciplines. We propose a non-parametric statistical learning approach to estimate the governing laws of distance-based interactions, with no reference or assumption about their analytical form, from data consisting trajectories of interacting agents. We demonstrate the effectiveness of our learning approach both by providing theoretical guarantees, and by testing the approach on a variety of prototypical systems in various disciplines. These systems include homogeneous and heterogeneous agents systems, ranging from particle systems in fundamental physics to agent-based systems modeling opinion dynamics under the social influence, prey-predator dynamics, flocking and swarming, and phototaxis in cell dynamics.

cs.LG

http://arxiv.org/abs/1812.06003v4

A Path Integral Approach to Business Cycle Models with Large Number of Agents

2018-10-16T10:03:35+00:00

2018-10-16T10:03:35+00:00

http://arxiv.org/pdf/1810.07178v1

http://arxiv.org/abs/1810.07178v1

This paper presents an analytical treatment of economic systems with an arbitrary number of agents that keeps track of the systems' interactions and agents' complexity. This formalism does not seek to aggregate agents. It rather replaces the standard optimization approach by a probabilistic description of both the entire system and agents'behaviors. This is done in two distinct steps. A first step considers an interacting system involving an arbitrary number of agents, where each agent's utility function is subject to unpredictable shocks. In such a setting, individual optimization problems need not be resolved. Each agent is described by a time-dependent probability distribution centered around his utility optimum. The entire system of agents is thus defined by a composite probability depending on time, agents' interactions and forward-looking behaviors. This dynamic system is described by a path integral formalism in an abstract space-the space of the agents' actions-and is very similar to a statistical physics or quantum mechanics system. We show that this description, applied to the space of agents'actions, reduces to the usual optimization results in simple cases. Compared to a standard optimization, such a description markedly eases the treatment of systems with small number of agents. It becomes however useless for a large number of agents. In a second step therefore, we show that for a large number of agents, the previous description is equivalent to a more compact description in terms of field theory. This yields an analytical though approximate treatment of the system. This field theory does not model the aggregation of a microeconomic system in the usual sense. It rather describes an environment of a large number of interacting agents. From this description, various phases or equilibria may be retrieved, along with individual agents' behaviors and their interactions with the environment. For illustrative purposes, this paper studies a Business Cycle model with a large number of agents.

econ.GN

http://arxiv.org/abs/1810.07178v1

Interoceptive robustness through environment-mediated morphological development

2018-04-06T13:33:37+00:00

2018-06-20T01:39:25+00:00

http://arxiv.org/pdf/1804.02257v2

http://arxiv.org/abs/1804.02257v2

Typically, AI researchers and roboticists try to realize intelligent behavior in machines by tuning parameters of a predefined structure (body plan and/or neural network architecture) using evolutionary or learning algorithms. Another but not unrelated longstanding property of these systems is their brittleness to slight aberrations, as highlighted by the growing deep learning literature on adversarial examples. Here we show robustness can be achieved by evolving the geometry of soft robots, their control systems, and how their material properties develop in response to one particular interoceptive stimulus (engineering stress) during their lifetimes. By doing so we realized robots that were equally fit but more robust to extreme material defects (such as might occur during fabrication or by damage thereafter) than robots that did not develop during their lifetimes, or developed in response to a different interoceptive stimulus (pressure). This suggests that the interplay between changes in the containing systems of agents (body plan and/or neural architecture) at different temporal scales (evolutionary and developmental) along different modalities (geometry, material properties, synaptic weights) and in response to different signals (interoceptive and external perception) all dictate those agents' abilities to evolve or learn capable and robust strategies.

cs.AI

http://arxiv.org/abs/1804.02257v2

Multi-Object Hallucination in Vision-Language Models

2024-07-08T17:59:57+00:00

2024-07-08T17:59:57+00:00

http://arxiv.org/pdf/2407.06192v1

http://arxiv.org/abs/2407.06192v1

Large vision language models (LVLMs) often suffer from object hallucination, producing objects not present in the given images. While current benchmarks for object hallucination primarily concentrate on the presence of a single object class rather than individual entities, this work systematically investigates multi-object hallucination, examining how models misperceive (e.g., invent nonexistent objects or become distracted) when tasked with focusing on multiple objects simultaneously. We introduce Recognition-based Object Probing Evaluation (ROPE), an automated evaluation protocol that considers the distribution of object classes within a single image during testing and uses visual referring prompts to eliminate ambiguity. With comprehensive empirical studies and analysis of potential factors leading to multi-object hallucination, we found that (1) LVLMs suffer more hallucinations when focusing on multiple objects compared to a single object. (2) The tested object class distribution affects hallucination behaviors, indicating that LVLMs may follow shortcuts and spurious correlations.(3) Hallucinatory behaviors are influenced by data-specific factors, salience and frequency, and model intrinsic behaviors. We hope to enable LVLMs to recognize and reason about multiple objects that often occur in realistic visual scenes, provide insights, and quantify our progress towards mitigating the issues.

cs.CV

http://arxiv.org/abs/2407.06192v1

Tailor3D: Customized 3D Assets Editing and Generation with Dual-Side Images

2024-07-08T17:59:55+00:00

2024-07-08T17:59:55+00:00

http://arxiv.org/pdf/2407.06191v1

http://arxiv.org/abs/2407.06191v1

Recent advances in 3D AIGC have shown promise in directly creating 3D objects from text and images, offering significant cost savings in animation and product design. However, detailed edit and customization of 3D assets remains a long-standing challenge. Specifically, 3D Generation methods lack the ability to follow finely detailed instructions as precisely as their 2D image creation counterparts. Imagine you can get a toy through 3D AIGC but with undesired accessories and dressing. To tackle this challenge, we propose a novel pipeline called Tailor3D, which swiftly creates customized 3D assets from editable dual-side images. We aim to emulate a tailor's ability to locally change objects or perform overall style transfer. Unlike creating 3D assets from multiple views, using dual-side images eliminates conflicts on overlapping areas that occur when editing individual views. Specifically, it begins by editing the front view, then generates the back view of the object through multi-view diffusion. Afterward, it proceeds to edit the back views. Finally, a Dual-sided LRM is proposed to seamlessly stitch together the front and back 3D features, akin to a tailor sewing together the front and back of a garment. The Dual-sided LRM rectifies imperfect consistencies between the front and back views, enhancing editing capabilities and reducing memory burdens while seamlessly integrating them into a unified 3D representation with the LoRA Triplane Transformer. Experimental results demonstrate Tailor3D's effectiveness across various 3D generation and editing tasks, including 3D generative fill and style transfer. It provides a user-friendly, efficient solution for editing 3D assets, with each editing step taking only seconds to complete.

cs.CV

http://arxiv.org/abs/2407.06191v1

4D Contrastive Superflows are Dense 3D Representation Learners

2024-07-08T17:59:54+00:00

2024-07-08T17:59:54+00:00

http://arxiv.org/pdf/2407.06190v1

http://arxiv.org/abs/2407.06190v1

In the realm of autonomous driving, accurate 3D perception is the foundation. However, developing such models relies on extensive human annotations -- a process that is both costly and labor-intensive. To address this challenge from a data representation learning perspective, we introduce SuperFlow, a novel framework designed to harness consecutive LiDAR-camera pairs for establishing spatiotemporal pretraining objectives. SuperFlow stands out by integrating two key designs: 1) a dense-to-sparse consistency regularization, which promotes insensitivity to point cloud density variations during feature learning, and 2) a flow-based contrastive learning module, carefully crafted to extract meaningful temporal cues from readily available sensor calibrations. To further boost learning efficiency, we incorporate a plug-and-play view consistency module that enhances the alignment of the knowledge distilled from camera views. Extensive comparative and ablation studies across 11 heterogeneous LiDAR datasets validate our effectiveness and superiority. Additionally, we observe several interesting emerging properties by scaling up the 2D and 3D backbones during pretraining, shedding light on the future research of 3D foundation models for LiDAR-based perception.

cs.CV

http://arxiv.org/abs/2407.06190v1

Video-STaR: Self-Training Enables Video Instruction Tuning with Any Supervision

2024-07-08T17:59:42+00:00

2024-07-08T17:59:42+00:00

http://arxiv.org/pdf/2407.06189v1

http://arxiv.org/abs/2407.06189v1

The performance of Large Vision Language Models (LVLMs) is dependent on the size and quality of their training datasets. Existing video instruction tuning datasets lack diversity as they are derived by prompting large language models with video captions to generate question-answer pairs, and are therefore mostly descriptive. Meanwhile, many labeled video datasets with diverse labels and supervision exist - however, we find that their integration into LVLMs is non-trivial. Herein, we present Video Self-Training with augmented Reasoning (Video-STaR), the first video self-training approach. Video-STaR allows the utilization of any labeled video dataset for video instruction tuning. In Video-STaR, an LVLM cycles between instruction generation and finetuning, which we show (I) improves general video understanding and (II) adapts LVLMs to novel downstream tasks with existing supervision. During generation, an LVLM is prompted to propose an answer. The answers are then filtered only to those that contain the original video labels, and the LVLM is then re-trained on the generated dataset. By only training on generated answers that contain the correct video labels, Video-STaR utilizes these existing video labels as weak supervision for video instruction tuning. Our results demonstrate that Video-STaR-enhanced LVLMs exhibit improved performance in (I) general video QA, where TempCompass performance improved by 10%, and (II) on downstream tasks, where Video-STaR improved Kinetics700-QA accuracy by 20% and action quality assessment on FineDiving by 15%.

cs.CV

http://arxiv.org/abs/2407.06189v1

CrowdMoGen: Zero-Shot Text-Driven Collective Motion Generation

2024-07-08T17:59:36+00:00

2024-07-08T17:59:36+00:00

http://arxiv.org/pdf/2407.06188v1

http://arxiv.org/abs/2407.06188v1

Crowd Motion Generation is essential in entertainment industries such as animation and games as well as in strategic fields like urban simulation and planning. This new task requires an intricate integration of control and generation to realistically synthesize crowd dynamics under specific spatial and semantic constraints, whose challenges are yet to be fully explored. On the one hand, existing human motion generation models typically focus on individual behaviors, neglecting the complexities of collective behaviors. On the other hand, recent methods for multi-person motion generation depend heavily on pre-defined scenarios and are limited to a fixed, small number of inter-person interactions, thus hampering their practicality. To overcome these challenges, we introduce CrowdMoGen, a zero-shot text-driven framework that harnesses the power of Large Language Model (LLM) to incorporate the collective intelligence into the motion generation framework as guidance, thereby enabling generalizable planning and generation of crowd motions without paired training data. Our framework consists of two key components: 1) Crowd Scene Planner that learns to coordinate motions and dynamics according to specific scene contexts or introduced perturbations, and 2) Collective Motion Generator that efficiently synthesizes the required collective motions based on the holistic plans. Extensive quantitative and qualitative experiments have validated the effectiveness of our framework, which not only fills a critical gap by providing scalable and generalizable solutions for Crowd Motion Generation task but also achieves high levels of realism and flexibility.

cs.CV

http://arxiv.org/abs/2407.06188v1

JeDi: Joint-Image Diffusion Models for Finetuning-Free Personalized Text-to-Image Generation

2024-07-08T17:59:02+00:00

2024-07-08T17:59:02+00:00

http://arxiv.org/pdf/2407.06187v1

http://arxiv.org/abs/2407.06187v1

Personalized text-to-image generation models enable users to create images that depict their individual possessions in diverse scenes, finding applications in various domains. To achieve the personalization capability, existing methods rely on finetuning a text-to-image foundation model on a user's custom dataset, which can be non-trivial for general users, resource-intensive, and time-consuming. Despite attempts to develop finetuning-free methods, their generation quality is much lower compared to their finetuning counterparts. In this paper, we propose Joint-Image Diffusion (\jedi), an effective technique for learning a finetuning-free personalization model. Our key idea is to learn the joint distribution of multiple related text-image pairs that share a common subject. To facilitate learning, we propose a scalable synthetic dataset generation technique. Once trained, our model enables fast and easy personalization at test time by simply using reference images as input during the sampling process. Our approach does not require any expensive optimization process or additional modules and can faithfully preserve the identity represented by any number of reference images. Experimental results show that our model achieves state-of-the-art generation quality, both quantitatively and qualitatively, significantly outperforming both the prior finetuning-based and finetuning-free personalization baselines.

cs.CV

http://arxiv.org/abs/2407.06187v1