Are you interested in performing high-impact interdisciplinary research in Artificial Intelligence and its alignment with humans and society? The University of Amsterdam has recently started a flagship project on Human-Aligned Video AI (HAVA). The HAVA Lab will address fundamental questions about what defines human alignment with video AI, how to make this computable, and what determines its societal acceptance. 

Video AI holds the promise to explore what is unreachable, monitor what is imperceivable and to protect what is most valuable. New species have become identifiable in our deep oceans, the visually impaired profit from automated speech transcriptions of visual scenery, and elderly caregivers may be supported with an extra pair of eyes, to name just three of the many, many application examples. This is no longer wishful thinking. Broad uptake of video-AI for science, for business, and for wellbeing awaits at the horizon, thanks to a decade of phenomenal progress in machine deep learning. 

However, the same video-AI is also accountable for self-driving cars crashing into pedestrians, deep fakes that make us believe misinformation, and mass-surveillance systems that monitor our behaviour. The research community’s over-concentration on recognition accuracy has neglected human-alignment for societal acceptance. The HAVA Lab is an intern-disciplinary lab that will study how to make the much-needed digital transformation towards human-aligned video AI. 

The HAVA Lab will host 7 PhD positions working together with researchers from all 7 faculties of the university, from video AI and its alignment with human cognition, ethics, and law, to its embedding in medical domains, public safety, and business. The lab has 9 supervisors in total spanning all 7 faculties of the university for maximum interdisciplinarity. Depending on the specific topic, the PhD students also have a strong link to the working environment and faculty of their respective supervisors.  The HAVA Lab has been given a unique central location at the library, an ideal hub for interdisciplinary collaborations. The PI of the lab is prof. dr. Cees Snoek.

Five of the seven PhD positions have been filled, we are looking to fill two more PhD positions, with the following interdisciplinary focus:

  • One PhD Position on human-aligned video-AI for public safety, which will be supervised by prof. dr. Marie Rosenkrantz Lindegaard and prof. dr. Cees Snoek.
  • One PhD Position on human-aligned video-AI for surgical skills, which will be supervised by prof. dr. Marlies Schijven and prof. dr. Cees Snoek.

For more details on the vacancies pls check: https://vacatures.uva.nl/UvA/job/Two-PhD-Positions-on-Human-aligned-Video-AI/794802002/

The ICCV 2023 paper Bayesian Prompt Learning for Image-Language Model Generalization by Mohammad Mahdi Derakhshani, Enrique Sanchez, Adrian Bulat, Victor Guilherme Turrisi da Costa, Cees G M Snoek, Georgios Tzimiropoulos, Brais Martinez is now available. Foundational image-language models have generated considerable interest due to their efficient adaptation to downstream tasks by prompt learning. Prompt learning treats part of the language model input as trainable while freezing the rest, and optimizes an Empirical Risk Minimization objective. However, Empirical Risk Minimization is known to suffer from distributional shifts which hurt generalizability to prompts unseen during training. By leveraging the regularization ability of Bayesian methods, we frame prompt learning from the Bayesian perspective and formulate it as a variational inference problem. Our approach regularizes the prompt space, reduces overfitting to the seen prompts and improves the prompt generalization on unseen prompts. Our framework is implemented by modeling the input prompt space in a probabilistic manner, as an a priori distribution which makes our proposal compatible with prompt learning approaches that are unconditional or conditional on the image. We demonstrate empirically on 15 benchmarks that Bayesian prompt learning provides an appropriate coverage of the prompt space, prevents learning spurious features, and exploits transferable invariant features. This results in better generalization of unseen prompts, even across different datasets and domains.

The ICCV 2023 paper Detecting Objects with Graph Priors and Graph Refinement by Aritra Bhowmik, Martin R Oswald, Yu Wang, Nora Baka, Cees G M Snoek is now available. The goal of this paper is to detect objects by exploiting their interrelationships. Rather than relying on predefined and labeled graph structures, we infer a graph prior from object co-occurrence statistics. The key idea of our paper is to model object relations as a function of initial class predictions and co-occurrence priors to generate a graph representation of an image for improved classification and bounding box regression. We additionally learn the object-relation joint distribution via energy based modeling. Sampling from this distribution generates a refined graph representation of the image which in turn produces improved detection performance. Experiments on the Visual Genome and MS-COCO datasets demonstrate our method is detector agnostic, end-to-end trainable, and especially beneficial for rare object classes. What is more, we establish a consistent improvement over object detectors like DETR and Faster-RCNN, as well as state-of-the-art methods modeling object interrelationships.

The ICCV 2023 paper Tubelet-Contrastive Self-Supervision for Video-Efficient Generalization by Fida Mohammad Thoker, Hazel Doughty, Cees G M Snoek is now available. We propose a self-supervised method for learning motion-focused video representations. Existing approaches minimize distances between temporally augmented videos, which maintain high spatial similarity. We instead propose to learn similarities between videos with identical local motion dynamics but an otherwise different appearance. We do so by adding synthetic motion trajectories to videos which we refer to as tubelets. By simulating different tubelet motions and applying transformations, such as scaling and rotation, we introduce motion patterns beyond what is present in the pretraining data. This allows us to learn a video representation that is remarkably data-efficient: our approach maintains performance when using only 25% of the pretraining videos. Experiments on 10 diverse downstream settings demonstrate our competitive performance and generalizability to new domains and fine-grained actions.

The ICCV 2023 paper Self-Ordering Point Clouds by Pengwan Yang, Cees G M Snoek, and Yuki M Asano is now available. In this paper we address the task of finding representative subsets of points in a 3D point cloud by means of a point-wise ordering. Only a few works have tried to address this challenging vision problem, all with the help of hard to obtain point and cloud labels. Different from these works, we introduce the task of point-wise ordering in 3D point clouds through self-supervision, which we call self-ordering. We further contribute the first end-to-end trainable network that learns a point-wise ordering in a self-supervised fashion. It utilizes a novel differentiable point scoring-sorting strategy and it constructs an hierarchical contrastive scheme to obtain self-supervision signals. We extensively ablate the method and show its scalability and superior performance even compared to supervised ordering methods on multiple datasets and tasks including zero-shot ordering of point clouds from unseen categories.

The ICCV 2023 paper Order-preserving Consistency Regularization for Domain Adaptation and Generalization by Mengmeng Jing, Xiantong Zhen, Jingjing Li, and Cees G M Snoek is now available. Deep learning models fail on cross-domain challenges if the model is oversensitive to domain-specific attributes, e.g., lightning, background, camera angle, etc. To alleviate this problem, data augmentation coupled with consistency regularization are commonly adopted to make the model less sensitive to domain-specific attributes. Consistency regularization enforces the model to output the same representation or prediction for two views of one image. These constraints, however, are either too strict or not order-preserving for the classification probabilities. In this work, we propose the Order-preserving Consistency Regularization (OCR) for cross-domain tasks. The order-preserving property for the prediction makes the model robust to task-irrelevant transformations. As a result, the model becomes less sensitive to the domain-specific attributes. The comprehensive experiments show that our method achieves clear advantages on five different cross-domain tasks.

The ICCV 2023 paper entitled Time Does Tell: Self-Supervised Time-Tuning of Dense Image Representations by Mohammadreza Salehi, Efstratios Gavves, Cees G M Snoek, Yuki M Asano is now available. Spatially dense self-supervised learning is a rapidly growing problem domain with promising applications for unsupervised segmentation and pretraining for dense downstream tasks. Despite the abundance of temporal data in the form of videos, this information-rich source has been largely overlooked. Our paper aims to address this gap by proposing a novel approach that incorporates temporal consistency in dense self-supervised learning. While methods designed solely for images face difficulties in achieving even the same performance on videos, our method improves not only the representation quality for videos-but also images. Our approach, which we call time-tuning, starts from image-pretrained models and fine-tunes them with a novel self-supervised temporal-alignment clustering loss on unlabeled videos. This effectively facilitates the transfer of high-level information from videos to image representations. Time-tuning improves the state-of-the-art by 8-10% for unsupervised semantic segmentation on videos and matches it for images. We believe this method paves the way for further self-supervised scaling by leveraging the abundant availability of videos.

The ICML 2023 paper Unlocking Slot Attention by Changing Optimal Transport Costs by Yan Zhang, David W Zhang, Simon Lacoste-Julien, Gertjan J Burghouts, Cees G M Snoek is now available. Slot attention is a powerful method for object-centric modeling in images and videos. However, its set-equivariance limits its ability to handle videos with a dynamic number of objects because it cannot break ties. To overcome this limitation, we first establish a connection between slot attention and optimal transport. Based on this new perspective we propose MESH (Minimize Entropy of Sinkhorn): a cross-attention module that combines the tiebreaking properties of unregularized optimal transport with the speed of regularized optimal transport. We evaluate slot attention using MESH on multiple object-centric learning benchmarks and find significant improvements over slot attention in every setting.

Make slot attention more powerful by taking an optimal transport perspective.

The ICML 2023 paper MetaModulation: Learning Variational Feature Hierarchies for Few-Shot Learning with Fewer Tasks by Wenfang Sun, Yingjun Du, Xiantong Zhen, Fan Wang, Ling Wang, and Cees G M Snoek is now available. Meta-learning algorithms are able to learn a new task using previously learned knowledge, but they often require a large number of meta-training tasks which may not be readily available. To address this issue, we propose a method for few-shot learning with fewer tasks, which we call MetaModulation. The key idea is to use a neural network to increase the density of the meta-training tasks by modulating batch normalization parameters during meta-training. Additionally, we modify parameters at various network levels, rather than just a single layer, to increase task diversity. To account for the uncertainty caused by the limited training tasks, we propose a variational MetaModulation where the modulation parameters are treated as latent variables. We also introduce learning variational feature hierarchies by the variational MetaModulation, which modulates features at all layers and can consider task uncertainty and generate more diverse tasks. The ablation studies illustrate the advantages of utilizing a learnable task modulation at different levels and demonstrate the benefit of incorporating probabilistic variants in few-task meta-learning. Our MetaModulation and its variational variants consistently outperform state-of-the-art alternatives on four few-task meta-learning benchmarks.

The ICLR2023 cam-ready ‘Energy-Based Test Sample Adaptation for Domain Generalization‘ by Zehao Xiao, Xiantong Zhen, Shanghai Liao and Cees Snoek is now available. In this paper, we propose energy-based sample adaptation at test time for domain generalization. Where previous works adapt their models to target domains, we adapt the unseen target samples to source-trained models. To this end, we design a discriminative energy-based model, which is trained on source domains to jointly model the conditional distribution for classification and data distribution for sample adaptation. The model is optimized to simultaneously learn a classifier and an energy function. To adapt target samples to source distributions, we iteratively update the samples by energy minimization with stochastic gradient Langevin dynamics. Moreover, to preserve the categorical information in the sample during adaptation, we introduce a categorical latent variable into the energy-based model. The latent variable is learned from the original sample before adaptation by variational inference and fixed as a condition to guide the sample update. Experiments on six benchmarks for classification of images and microblog threads demonstrate the effectiveness of our proposal.