More Like this

1. HaLP: Hallucinating Latent Positives for Skeleton-based Self-Supervised Learning of Actions

   https://doi.org/10.1109/CVPR52729.2023.01807  

   Shah, Anshul ; Roy, Aniket ; Shah, Ketul ; Mishra, Shlok ; Jacobs, David ; Cherian, Anoop ; Chellappa, Rama ( June 2023 , IEEE)

   Supervised learning of skeleton sequence encoders for action recognition has received significant attention in recent times. However, learning such encoders without labels continues to be a challenging problem. While prior works have shown promising results by applying contrastive learning to pose sequences, the quality of the learned representations is often observed to be closely tied to data augmentations that are used to craft the positives. However, augmenting pose sequences is a difficult task as the geometric constraints among the skeleton joints need to be enforced to make the augmentations realistic for that action. In this work, we propose a new contrastive learning approach to train models for skeleton-based action recognition without labels. Our key contribution is a simple module, HaLP – to Hallucinate Latent Positives for contrastive learning. Specifically, HaLP explores the latent space of poses in suitable directions to generate new positives. To this end, we present a novel optimization formulation to solve for the synthetic positives with an explicit control on their hardness. We propose approximations to the objective, making them solvable in closed form with minimal overhead. We show via experiments that using these generated positives within a standard contrastive learning framework leads to consistent improvements across benchmarks such as NTU-60, NTU- 120, and PKU-II on tasks like linear evaluation, transfer learning, and kNN evaluation. Our code can be found at https://github.com/anshulbshah/HaLP. 

   more » « less
2. SkeletonMAE: Spatial-Temporal Masked Autoencoders for Self-supervised Skeleton Action Recognition

   Wenhan Wu, Yilei Hua ( July 2023 , Workshop in International Conference on Multimedia and Expo (ICME), 2023)

   Self-supervised skeleton-based action recognition has attracted more attention in recent years. By utilizing the unlabeled data, more generalizable features can be learned to alleviate the overfitting problem and reduce the demand for massive labeled training data. Inspired by the MAE [1], we propose a spatial-temporal masked autoencoder framework for self-supervised 3D skeleton-based action recognition (SkeletonMAE). Following MAE's masking and reconstruction pipeline, we utilize a skeleton-based encoder-decoder transformer architecture to reconstruct the masked skeleton sequences. A novel masking strategy, named Spatial-Temporal Masking, is introduced in terms of both joint-level and frame-level for the skeleton sequence. This pre-training strategy makes the encoder output generalizable skeleton features with spatial and temporal dependencies. Given the unmasked skeleton sequence, the encoder is fine-tuned for the action recognition task. Extensive ex- periments show that our SkeletonMAE achieves remarkable performance and outperforms the state-of-the-art methods on both NTU RGB+D 60 and NTU RGB+D 120 datasets. 

   more » « less
3. Part Aware Contrastive Learning for Self-Supervised Action Recognition

   Yilei Hua, Wenhan Wu ( October 2023 , THE 32nd INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE)

   In recent years, remarkable results have been achieved in self-supervised action recognition using skeleton sequences with contrastive learning. It has been observed that the semantic distinction of human action features is often represented by local body parts, such as legs or hands, which are advantageous for skeleton-based action recognition. This paper proposes an attention-based contrastive learning framework for skeleton representation learning, called SkeAttnCLR, which integrates local similarity and global features for skeleton-based action representations. To achieve this, a multi-head attention mask module is employed to learn the soft attention mask features from the skeletons, suppressing non-salient local features while accentuating local salient features, thereby bringing similar local features closer in the feature space. Additionally, ample contrastive pairs are generated by expanding contrastive pairs based on salient and non-salient features with global features, which guide the network to learn the semantic representations of the entire skeleton. Therefore, with the attention mask mechanism, SkeAttnCLR learns local features under different data augmentation views. The experiment results demonstrate that the inclusion of local feature similarity significantly enhances skeleton-based action representation. Our proposed SkeAttnCLR outperforms state-of-the-art methods on NTURGB+D, NTU120-RGB+D, and PKU-MMD datasets. The code and settings are available at this repository: https://github.com/GitHubOfHyl97/SkeAttnCLR. 

   more » « less
4. Self-supervised 4D Spatio-temporal Feature Learning via Order Prediction of Sequential Point Cloud Clips

   Wang, H ; Yang, L ; Rong, X ; Feng, J ; Tian, Y. ( June 2021 , Winter Conference on Applications of Computer Vision (WACV), 2021.)

   null (Ed.)

   Recently 3D scene understanding attracts attention for many applications, however, annotating a vast amount of 3D data for training is usually expensive and time consuming. To alleviate the needs of ground truth, we propose a self-supervised schema to learn 4D spatio-temporal features (i.e. 3 spatial dimensions plus 1 temporal dimension) from dynamic point cloud data by predicting the temporal order of sampled and shuffled point cloud clips. 3D sequential point cloud contains precious geometric and depth information to better recognize activities in 3D space compared to videos. To learn the 4D spatio-temporal features, we introduce 4D convolution neural networks to predict the temporal order on a self-created large scale dataset, NTU- PCLs, derived from the NTU-RGB+D dataset. The efficacy of the learned 4D spatio-temporal features is verified on two tasks: 1) Self-supervised 3D nearest neighbor retrieval; and 2) Self-supervised representation learning transferred for action recognition on smaller 3D dataset. Our extensive experiments prove the effectiveness of the proposed self-supervised learning method which achieves comparable results w.r.t. the fully-supervised methods on action recognition on MSRAction3D dataset. 

   more » « less
5. An LSTM-based adversarial variational autoencoder framework for self-supervised neural decoding of behavioral choices

   https://doi.org/10.1088/1741-2552/ad3eb3  

   Salsabilian, Shiva ; Lee, Christian ; Margolis, David ; Najafizadeh, Laleh ( April 2024 , Journal of Neural Engineering)

   Objective. This paper presents data-driven solutions to address two challenges in the problem of linking neural data and behavior: 1) unsupervised analysis of behavioral data and automatic label generation from behavioral observations, and 2) extraction of subject-invariant features for the development of generalized neural decoding models. Approach. For behavioral analysis and label generation, an unsupervised method, which employs an autoencoder to transform behavior data into a cluster-friendly feature space is presented. The model iteratively refines the assigned clusters with soft clustering assignment loss, and gradually improves the learned feature representations. To address subject variability in decoding neural activity, adversarial learning in combination with a long short-term memory-based adversarial variational autoencoder (LSTM-AVAE) model is employed. By using an adversary network to constrain the latent representations, the model captures shared information among subjects' neural activity, making it proper for cross-subject transfer learning. Main results. The proposed approach is evaluated using cortical recordings of Thy1-GCaMP6s transgenic mice obtained via widefield calcium imaging during a motivational licking behavioral experiment. The results show that the proposed model achieves an accuracy of 89.7% in cross-subject neural decoding, outperforming other well-known autoencoder-based feature learning models. These findings suggest that incorporating an adversary network eliminates subject dependency in representations, leading to improved cross-subject transfer learning performance, while also demonstrating the effectiveness of LSTM-based models in capturing the temporal dependencies within neural data. Significance. Results demonstrate the feasibility of the proposed framework in unsupervised clustering and label generation of behavioral data, as well as achieving high accuracy in cross-subject neural decoding, indicating its potentials for relating neural activity to behavior.

    

   more » « less