United States

Language models (LMs) are shown to have commonsense knowledge of the physical world, which is fundamental for completing tasks in everyday situations. However, it is still an open question whether LMs have the ability to generate grounded, executable plans for embodied tasks. It is very challenging because LMs do not have ``eyes&#39;&#39; or ``hands&#39;&#39; to observe and reason about the realistic environment. In this work, we show the first study on this important research question. We first present a novel problem formulation named G-PlanET, which takes as input a high-level goal and a table of objects in a specific environment.
The expected output is a plan consisting of step-by-step instructions for agents to execute.  To enable the study of this problem, we establish an evaluation protocol and devise a dedicated metric for assessing the quality of plans. In our extensive experiments, we show that adding flattened tables for encoding environments and using an iterative decoding strategy can both improve the LMs&#39; ability for grounded planning. Our analysis of the results also leads to interesting non-trivial findings.  

AAAI 2023

On Grounded Planning for Embodied Tasks with Language Models

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-23 is the Thirty-Seventh AAAI Conference on Artificial Intelligence. The theme of this conference is to create collaborative bridges within and beyond AI. Like previous AAAI conferences, AAAI-23 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and two new activities: a Bridge Program and a Lab Program. Many of these activities are tailored to the theme of bridges and all are selected according to the highest standards, with additional programs for students and young researchers. 
AAAI is providing you with a conference planner, which you can use to help organize your itinerary of activities. This includes talks to attend in person, talks to attend remotely, breaks with colleagues and your site seeing activities. To access this conference planner, please go to [https://aaai-2023.takemobi.io](https://aaai-2023.takemobi.io).

In order to access this site, you need to register. If you haven't already, please register [here](https://aaai.org/Conferences/AAAI-23/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.

Language models (LMs) are shown to have commonsense knowledge of the physical world, which is fundamental for completing tasks in everyday situations. However, it is still an open question whether LMs have the ability to generate grounded, executable plans for embodied tasks. It is very challenging because LMs do not have ``eyes'' or ``hands'' to observe and reason about the realistic environment. In this work, we show the first study on this important research question. We first present a novel problem formulation named G-PlanET, which takes as input a high-level goal and a table of objects in a specific environment.
The expected output is a plan consisting of step-by-step instructions for agents to execute.  To enable the study of this problem, we establish an evaluation protocol and devise a dedicated metric for assessing the quality of plans. In our extensive experiments, we show that adding flattened tables for encoding environments and using an iterative decoding strategy can both improve the LMs' ability for grounded planning. Our analysis of the results also leads to interesting non-trivial findings.  

poster

Graph Contrastive Learning (GCL) has recently drawn much research interest for learning generalizable node representations in a self-supervised manner. In general, the contrastive learning process in GCL is performed on top of the representations learned by a graph neural network (GNN) backbone, which transforms and propagates the node contextual information based on its local neighborhoods. However, nodes sharing similar characteristics may not always be geographically close, which poses a great challenge for unsupervised GCL efforts due to their inherent limitations in capturing such global graph knowledge. In this work, we address their inherent limitations by proposing a simple yet effective framework -- Simple Neural Networks with Structural and Semantic Contrastive Learning} (S^3-CL). Notably, by virtue of the proposed structural and semantic contrastive learning algorithms, even a simple neural network can learn expressive node representations that preserve valuable global structural and semantic patterns. Our experiments demonstrate that the node representations learned by S^3-CL) achieve superior performance on different downstream tasks compared with the state-of-the-art unsupervised GCL methods.

Eliciting Structural and Semantic Global Knowledge in Unsupervised Graph Contrastive Learning

Generative adversarial networks (GANs) are known for their strong abilities to capture the underlying distribution and generate realistic-looking samples. Since the seminal work of GAN, many variants have been proposed. However, existing GANs are mostly established on the assumption that the training dataset is clean. But in many real-world applications, this may not hold, that is, the training dataset is contaminated by a proportion of undesired instances. When training on such a dataset, existing GANs will learn a mixture distribution of desirable and contaminated instances rather than the desired distribution. To learn the target distribution from contaminated datasets, we propose the purified generative adversarial network (PuriGAN). The discriminator is augmented with the capability to distinguish between target and contaminated instances by leveraging an extra dataset solely composed of contamination instances, which guarantees that the generator can be prevented from generating contaminated data. We prove that under some mild conditions, the proposed PuriGAN is able to converge to the distribution of desired instances. Experimental results on several image datasets demonstrate that the proposed PuriGAN is able to generate much better images from the desired distribution over competitive baselines under various kinds of conditions. Moreover, we also apply PuriGAN to two downstream applications, semi-supervised anomaly detection with contaminated datasets and PU-learning, with the experimental results showing that PuriGAN is able to deliver the best performance over comparable baselines on both tasks.

Leveraging Contaminated Datasets to Learn Clean-Data Distribution with Purified Generative Adversarial Networks

Humans and animals engage in rich social interactions. It is often theorized that a relatively small number of basic social interactions give rise to the full range of behavior observed. But no computational theory explaining how social interactions combine together has been proposed before. We do so here. We take a model, the Social MDP, which is able to express a range of social interactions, and extend it to represent linear combinations of social interactions. Practically for robotics applications, such models are now able to not just express that an agent should help another agent, but to express goal-centric social interactions. Perhaps an agent is helping someone get dressed, but preventing them from falling, and is happy to exchange stories in the meantime. How an agent responds socially, should depend on what it thinks the other agent is doing at that point in time. To encode this notion, we take linear combinations of social interactions as defined in Social MDPs, and compute the weights on those combinations on the fly depending on the estimated goals of other agents. This new model, the Linear Social MDP, enables zero-shot reasoning about complex social interactions, provides a mathematical basis for the long-standing intuition that social interactions should compose, and leads to interesting new behaviors that we validate using human observers. Complex social interactions are part of the future of intelligent agents, and having principled mathematical models built on a foundation like MDPs will make it possible to bring social interactions to every robotic application.

Zero-shot linear combinations of grounded social interactions with Linear Social MDPs

Recently, transformer architecture has gained great success in the computer vision community, such as image classification, object detection, etc. Nonetheless, its application for 3D vision remains to be explored, given that point cloud is inherently sparse, irregular, and unordered. Furthermore, existing point transformer frameworks usually feed raw point cloud of N×3 dimension into transformers, which limits the point processing scale because of their quadratic computational costs to the input size N. In this paper, we rethink the structure of point transformer. Instead of directly applying transformer to points, our network (TransLO) can process tens of thousands of points simultaneously by projecting points onto a 2D surface and then feeding them into a local transformer with linear complexity. Specifically, it is mainly composed of two components: Window-based Masked transformer with Self Attention (WMSA) to capture long-range dependencies; Masked Cross-Frame Attention (MCFA) to associate two frames and predict pose estimation. To deal with the sparsity issue of point cloud, we propose a binary mask to remove invalid and dynamic points. To our knowledge, this is the first transformer-based LiDAR odometry network. The experiment results on the KITTI odometry dataset show that our average rotation and translation RMSE achieves 0.500°/100m and 0.993% respectively. The performance of our network surpasses all recent learning-based methods and even outperforms LOAM on most evaluation sequences.

TransLO: A Window-Based Masked Point Transformer Framework for Large-Scale LiDAR Odometry

Topic models and all their variants analyse text by learning
meaningful representations through word co-occurrences. As
pointed out by Williamson et al. (2010), such models implic-
itly assume that the probability of a topic to be active and
its proportion within each document are positively correlated.
This correlation can be strongly detrimental in the case of doc-
uments created over time, simply because recent documents
are likely better described by new and hence rare topics. In this
work we leverage recent advances in neural variational infer-
ence and present an alternative neural approach to the dynamic
Focused Topic Model. Indeed, we develop a neural model for
topic evolution which exploits sequences of Bernoulli random
variables in order to track the appearances of topics, thereby
decoupling their activities from their proportions. We evaluate
our model on three different datasets (the UN general debates,
the collection of NEURIPS papers, and the ACL Anthology
dataset) and show that it (i) outperforms state-of-the-art topic
models in generalization tasks and (ii) performs comparably to
them on prediction tasks, while employing roughly the same
number of parameters, and converging about two times faster.

Neural Dynamic Focused Topic Model

In sequential machine teaching, a teacher’s objective is to provide the optimal sequence of inputs to sequential learners in order to guide them towards the best model. However, this teaching objective considers a restricted class of learners with fixed inductive biases. In this paper, we extend the machine teaching framework to learners that can improve their inductive biases, represented as latent internal states, in order to generalize to new datasets.
We introduce a novel framework in which learners’ inductive biases may change with the teaching interaction, which affects the learning performance in future tasks. In order to teach such learners, we propose a multi-objective control approach that takes the future performance of the learner after teaching into account. This framework provides tools for modelling learners with internal states, humans and meta-learning algorithms alike. Furthermore, we distinguish manipulative teaching, which can be done by effectively hiding data and also used for indoctrination, from teaching to learn which aims to help the learner become better at learning from new datasets in the absence of a teacher. Our empirical results demonstrate that our framework is able to reduce the number of required tasks for online meta-learning, and increases independent learning performance of simulated human users in future tasks.

Teaching to Learn: Sequential Teaching of Learners with Internal States

We propose and show the efficacy of a new method to address generic inverse problems. Inverse modeling is the task whereby one seeks to determine the hidden parameters of a natural system that produce a given set of observed measurements. Recent work has shown impressive results using deep learning, but we note that there is a trade-off between model performance and computational time. For some applications, the computational time at inference for the best performing inverse modeling method may be overly prohibitive to its use. In seeking a faster, high-performing model, we present a new method that leverages multiple manifolds as a mixture of backward (e.g., inverse) models in a forward-backward model architecture. These multiple backwards models all share a common forward model, and their training is mitigated by generating training examples from the forward model. The proposed method thus has two innovations: 1) the multiple Manifold Mixture Network (MMN) architecture, and 2) the training procedure involving augmenting backward model training data using the forward model. We demonstrate the advantages of our method by comparing to several baselines on four benchmark inverse problems, and we furthermore provide analysis to motivate its design.

Mixture Manifold Networks: A Computationally Efficient Baseline for Inverse Modeling

Graph Neural Networks (GNNs) have drawn significant attentions over the years and been broadly applied to essential applications requiring solid robustness or vigorous security standards, such as product recommendation and user behavior modeling. Under these scenarios, exploiting GNN's vulnerabilities and further downgrading its performance become extremely incentive for adversaries. Previous attackers mainly focus on structural perturbations or node injections to the existing graphs, guided by gradients from the surrogate models. Although they deliver promising results, several limitations still exist. For the structural perturbation attack, to launch a proposed attack, adversaries need to manipulate the existing graph topology, which is impractical in most circumstances. Whereas for the node injection attack, though being more practical, current approaches require training surrogate models to simulate a white-box setting, which results in significant performance downgrade when the surrogate architecture diverges from the actual victim model. To bridge these gaps, in this paper, we study the problem of black-box node injection attack, without training a potentially misleading surrogate model. Specifically, we model the node injection attack as a Markov decision process and propose Gradient-free Graph Advantage Actor Critic, namely G2A2C, a reinforcement learning framework in the fashion of advantage actor critic. By directly querying the victim model, G2A2C learns to inject highly malicious nodes with extremely limited attacking budgets, while maintaining a similar node feature distribution. Through our comprehensive experiments over eight acknowledged benchmark datasets with different characteristics, we demonstrate the superior performance of our proposed G2A2C over the existing state-of-the-art attackers. Source code is publicly available at: https://github.com/jumxglhf/G2A2C.

Let Graph be the Go Board: Gradient-free Node Injection Attack for Graph Neural Networks via Reinforcement Learning

Consider an undirected graph $G=(V,E)$ model for a communication network, where  each edge is owned by a selfish agent, who reports the cost for offering the use of her edge. Note that each edge agent may misreport her own cost for the use of the edge for her own benefit. In such a non-cooperative setting, we aim at designing an approximately truthful mechanism for establishing a Steiner tree, a minimum cost tree spanning over all the terminals. We present a truthful-in-expectation mechanism that achieves the approximation ratio $\ln4+\epsilon \approx 1.39$, which matches the current best algorithmic ratio for STP \cite{GORZ12,BGRS13,TZ21}.

Truthful Mechanisms for Steiner Tree Problems

Deep learning-based digital watermarking frameworks have been widely studied recently. Most existing methods adopt an ``encoder-noise layer-decoder''-based architecture where the embedding and extraction processes are accomplished separately by the encoder and the decoder. However, one potential drawback of such a framework is that the encoder and the decoder may not be well coupled, result in the fact that the encoder may embed some redundant features into the host image thus influencing the invisibility and robustness of the whole algorithm. To address this limitation, this paper proposes a flow-based robust watermarking framework. The basic component of such framework is an invertible up-down-sampling neural block that can realize the embedding and extraction simultaneously. As a consequence, the encoded feature could keep high consistency with the feature that the decoder needed, which effectively avoids the embedding of redundant features. In addition, to ensure the robustness of black-box distortion, an invertible noise layer (INL) is designed to simulate the distortion and is served as a noise layer in the training stage. Benefiting from its reversibility, INL is also applied as a preprocessing before extraction to eliminate the distortion, which further improves the robustness of the algorithm. Extensive experiments demonstrate the superiority of the proposed framework in terms of visual quality and robustness. Compared with the state-of-the-art architecture, the visual quality (measured by PSNR) of the proposed framework improves by 2dB and the extraction accuracy after JPEG compression (QF=50) improves by more than 4%. Besides, the robustness against black-box distortions can be greatly achieved with more than 95% extraction accuracy.

On Grounded Planning for Embodied Tasks with Language Models

Next from AAAI 2023

Eliciting Structural and Semantic Global Knowledge in Unsupervised Graph Contrastive Learning

Similar lecture

DeAR: A Deep-learning-based Audio Re-recording Resilient Watermarking