Publications by Yuejiang Liu | EPFL Graph Search

Causal Triplet: An Open Challenge for Intervention-centric Causal Representation Learning

Recent years have seen a surge of interest in learning high-level causal representations from low-level image pairs under interventions. Yet, existing efforts are largely limited to simple synthetic settings that are far away from real-world problems. In this paper, we present Causal Triplet, a causal representation learning benchmark featuring not only visually more complex scenes, but also two crucial desiderata commonly overlooked in previous works: (i) an actionable counterfactual setting, where only certain object-level variables allow for counterfactual observations whereas others do not; (ii) an interventional downstream task with an emphasis on out-of-distribution robustness from the independent causal mechanisms principle. Through extensive experiments, we find that models built with the knowledge of disentangled or object-centric representations significantly outperform their distributed counterparts. However, recent causal representation learning methods still struggle to identify such latent structures, indicating substantial challenges and opportunities for future work. Our code and datasets will be available at https://sites.google.com/view/causaltriplet.

2023

On Pitfalls of Test-time Adaptation

Alexandre Massoud Alahi, Tao Lin, Yuejiang Liu, Hao Zhao

2023

Motion Style Transfer: Modular Low-Rank Adaptation for Deep Motion Forecasting

Alexandre Massoud Alahi, Parth Ashit Kothari, Danya Li, Yuejiang Liu

Deep motion forecasting models have achieved great success when trained on a massive amount of data. Yet, they often perform poorly when training data is limited. To address this challenge, we propose a transfer learning approach for efficiently adapting pre-trained forecasting models to new domains, such as unseen agent types and scene contexts. Unlike the conventional fine-tuning approach that updates the whole encoder, our main idea is to reduce the amount of tunable parameters that can precisely account for the target domain-specific motion style. To this end, we introduce two components that exploit our prior knowledge of motion style shifts: (i) a low-rank motion style adapter that projects and adjusts the style features at a low-dimensional bottleneck; and (ii) a modular adapter strategy that disentangles the features of scene context and motion history to facilitate a fine-grained choice of adaptation layers. Through extensive experimentation, we show that our proposed adapter design, coined MoSA, outperforms prior methods on several forecasting benchmarks.

2022

Learning Decoupled Representations for Human Pose Forecasting

Alexandre Massoud Alahi, Yuejiang Liu, Taylor Ferdinand Mordan, Behnam Parsaeifard, Saeed Saadatnejad

Human pose forecasting involves complex spatiotemporal interactions between body parts (e.g., arms, legs, spine). State-of-the-art approaches use Long Short-Term Memories (LSTMs) or Variational AutoEncoders (VAEs) to solve the problem. Yet, they do not effectively predict human motions when both global trajectory and local pose movements exist. We propose to learn decoupled representations for the global and local pose forecasting tasks. We also show that it is better to stop the prediction when the uncertainty in human motion increases. Our forecasting model outperforms all existing methods on the pose forecasting benchmark to date by over 20%. The code is available online: https://github.com/vita-epfl/decoupled-pose-prediction

IEEE2021

Social NCE: Contrastive Learning of Socially-aware Motion Representations

Alexandre Massoud Alahi, Yuejiang Liu

Learning socially-aware motion representations is at the core of recent advances in multi-agent problems, such as human motion forecasting and robot navigation in crowds. Despite promising progress, existing representations learned with neural networks still struggle to generalize in closed-loop predictions (e.g., output colliding trajectories). This issue largely arises from the non-i.i.d. nature of sequential prediction in conjunction with ill-distributed training data. Intuitively, if the training data only comes from human behaviors in safe spaces, i.e., from "positive" examples, it is difficult for learning algorithms to capture the notion of "negative" examples like collisions. In this work, we aim to address this issue by explicitly modeling negative examples through self-supervision: (i) we introduce a social contrastive loss that regularizes the extracted motion representation by discerning the ground-truth positive events from synthetic negative ones; (ii) we construct informative negative samples based on our prior knowledge of rare but dangerous circumstances. Our method substantially reduces the collision rates of recent trajectory forecasting, behavioral cloning and reinforcement learning algorithms, outperforming state-of-the-art methods on several benchmarks. Our code is available at https://github.com/vita-epfl/social-nce.

2021

TTT++: When Does Self-Supervised Test-Time Training Fail or Thrive?

Alexandre Massoud Alahi, Parth Ashit Kothari, Yuejiang Liu, Taylor Ferdinand Mordan, Bastien Germain C. Van Delft

Test-time training (TTT) through self-supervised learning (SSL) is an emerging paradigm to tackle distributional shifts. Despite encouraging results, it remains unclear when this approach thrives or fails. In this work, we first provide an in-depth look at its limitations and show that TTT can possibly deteriorate, instead of improving, the test-time performance in the presence of severe distribution shifts. To address this issue, we introduce a test-time feature alignment strategy utilizing offline feature summarization and online moment matching, which regularizes adaptation without revisiting training data. We further scale this strategy in the online setting through batch-queue decoupling to enable robust moment estimates even with limited batch size. Given aligned feature distributions, we then shed light on the strong potential of TTT by theoretically analyzing its performance post adaptation. This analysis motivates our use of more informative self-supervision in the form of contrastive learning for visual recognition problems. We empirically demonstrate that our modified version of test-time training, termed TTT++, outperforms state-of-the-art methods by significant margins on several benchmarks. Our result indicates that storing and exploiting extra information, in addition to model parameters, can be a promising direction towards robust test-time adaptation.

2021

, ,

The standard practice in Generative Adversarial Networks (GANs) discards the discriminator during sampling. However, this sampling method loses valuable information learned by the discriminator regarding the data distribution. In this work, we propose a collaborative sampling scheme between the generator and the discriminator for improved data generation. Guided by the discriminator, our approach refines the generated samples through gradient-based updates at a particular layer of the generator, shifting the generator distribution closer to the real data distribution. Additionally, we present a practical discriminator shaping method that can smoothen the loss landscape provided by the discriminator for effective sample refinement. Through extensive experiments on synthetic and image datasets, we demonstrate that our proposed method can improve generated samples both quantitatively and qualitatively, offering a new degree of freedom in GAN sampling.

2020

Crowd-Robot Interaction: Crowd-aware Robot Navigation with Attention-based Deep Reinforcement Learning

Alexandre Massoud Alahi, Changan Chen, Sven Kreiss, Yuejiang Liu

Mobility in an effective and socially-compliant manner is an essential yet challenging task for robots operating in crowded spaces. Recent works have shown the power of deep reinforcement learning techniques to learn socially cooperative policies. However, their cooperation ability deteriorates as the crowd grows since they typically relax the problem as a one-way Human-Robot interaction problem. In this work, we want to go beyond first-order Human-Robot interaction and more explicitly model Crowd-Robot Interaction (CRI). We propose to (i) rethink pairwise interactions with a self-attention mechanism, and (ii) jointly model Human-Robot as well as Human-Human interactions in the deep reinforcement learning framework. Our model captures the Human-Human interactions occurring in dense crowds that indirectly affects the robot's anticipation capability. Our proposed attentive pooling mechanism learns the collective importance of neighboring humans with respect to their future states. Various experiments demonstrate that our model can anticipate human dynamics and navigate in crowds with time efficiency, outperforming state-of-the-art methods.

IEEE2019

Real-Time Distributed Algorithms for Nonconvex Optimal Power Flow

Jean-Hubert Hours, Colin Neil Jones, Yuejiang Liu, Georgios Stathopoulos

The optimal power flow (OPF) problem, a fundamental problem in power systems, is generally nonconvex and computationally challenging for networks with an increasing number of smart devices and real-time control requirements. In this paper, we first investigate a fully distributed approach by means of the augmented Lagrangian and proximal alternating minimization method to solve the nonconvex OPF problem with a convergence guarantee. Given time-critical requirements, we then extend the algorithm to a distributed parametric tracking scheme with practical warm-starting and termination strategies, which aims to provide a closed-loop sub-optimal control policy while taking into account the grid information updated at the time of decision making. The effectiveness of the proposed algorithm for real-time nonconvex OPF problems is demonstrated in numerical simulations.

2017

Crowd-Robot Interaction: Crowd-aware Robot Navigation with Attention-based Deep Reinforcement Learning

Alexandre Massoud Alahi, Changan Chen, Sven Kreiss, Yuejiang Liu

Mobility in an effective and socially-compliant manner is an essential yet challenging task for robots operating in crowded spaces. Recent works have shown the power of deep reinforcement learning techniques to learn socially cooperative policies. However, their cooperation ability deteriorates as the crowd grows since they typically relax the problem as a one-way Human-Robot interaction problem. In this work, we want to go beyond first-order Human-Robot interaction and more explicitly model Crowd-Robot Interaction (CRI). We propose to (i) rethink pairwise interactions with a self-attention mechanism, and (ii) jointly model Human-Robot as well as Human-Human interactions in the deep reinforcement learning framework. Our model captures the Human-Human interactions occurring in dense crowds that indirectly affects the robot's anticipation capability. Our proposed attentive pooling mechanism learns the collective importance of neighboring humans with respect to their future states. Various experiments demonstrate that our model can anticipate human dynamics and navigate in crowds with time efficiency, outperforming state-of-the-art methods.

IEEE