Programming based approaches to reasoning tasks have substantially expanded the types of questions models can answer about visual scenes. Yet on benchmark visual reasoning data, when models answer correctly, they produce incorrect programs 33% of the time. These models are often right for the wrong reasons and risk unexpected failures on new data. Unit tests play a foundational role in ensuring code correctness and could be used to repair such failures. We propose Visual Unit Testing (ViUniT), a framework to improve the reliability of visual programs by automatically generating unit tests. In our framework, a unit test is represented as a novel image and answer pair meant to verify the logical correctness of a program produced for a given query. Our method leverages a language model to create unit tests in the form of image descriptions and expected answers, followed by image synthesis to produce corresponding images. We conduct a comprehensive analysis of what constitutes an effective visual unit test suite, exploring unit test generation, sampling strategies, image generation methods, and varying the number of programs and unit tests. Additionally, we introduce four applications of visual unit tests: best program selection, answer refusal, re-prompting, and unsupervised reward formulations for reinforcement learning. Experiments with two models across three datasets in visual question answering and image-text matching demonstrate that ViUniT improves model performance by 11.4 points in accuracy. Notably, it enables 7B open-source language models to outperform gpt-4o-mini in visual program generation by an average of 7.7 points and reduces the occurrence of programs that are correct for the wrong reasons by 40%.

Artemis Panagopoulou, Le Xue, Honglu Zhou, Silvio Savarese, Ran Xu, Caiming Xiong, Chris Callison-Burch, Mark Yatskar, Juan Carlos Niebles. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Recent advances in vision backbones have yielded powerful and diverse visual and video encoders. Yet, current Video Large Language Models encode visual inputs using an encoder from a single backbone family, limiting the amount and type of visual information they can process. We propose MERV, a Multi-Encoder Video Representation, which utilizes multiple encoders for a comprehensive video representation. To optimize heterogeneous features from a broad spectrum of encoders and ensure efficient and coherent feature integration, MERV first aligns encoder features spatio-temporally, then projects them into a unified structure, and finally fuses them through cross-attention. Under fair comparison, MERV achieves up to 4.62% higher accuracy than its base model, while introducing minimal extra parameters and training faster than equivalent single-encoder methods after parallelizing visual processing. Qualitative analysis shows MERV successfully captures and integrates domain knowledge from each encoder, opening new possibilities for scaling enhanced video understanding.

Unarguably, deep learning models capable of generalizing to unseen domain data while leveraging a few labels are of great practical significance due to low developmental costs. In search of this endeavor, we study the challenging problem of semi-supervised domain generalization (SSDG), where the goal is to learn a domain-generalizable model while using only a small fraction of labeled data and a relatively large fraction of unlabeled data. Domain generalization (DG) methods show subpar performance under the SSDG setting, whereas semi-supervised learning (SSL) methods demonstrate relatively better performance, however, they are considerably poor compared to the fully-supervised DG methods. Towards handling this new, but challenging problem of SSDG, we propose a novel method that can facilitate the generation of accurate pseudo-labels under various domain shifts. This is accomplished by retaining the domain-level specialism in the classifier during training corresponding to each source domain. Specifically, we first create domain-level information vectors on the fly which are then utilized to learn a domain-aware mask for modulating the classifier's weights. We provide a mathematical interpretation for the effect of this modulation procedure on both pseudo-labeling and model training. Our method is plug-and-play and can be readily applied to different SSL baselines for SSDG. Extensive experiments on six challenging datasets in two different SSDG settings show that our method provides visible gains over the various strong SSL-based SSDG baselines. Our code is available at github.com/DGWM.

The study of complex human interactions and group activities has become a focal point in human-centric computer vision. However, progress in related tasks is often hindered by the challenges of obtaining large-scale labeled datasets from real-world scenarios. To address the limitation, we introduce M3 Act, a synthetic data generator for multi-view multi-group multi-person human atomic actions and group activities. Powered by Unity Engine, M3 Act features mul-tiple semantic groups, highly diverse and photorealistic images, and a comprehensive set of annotations, which facilitates the learning of human-centered tasks across single-person, multi-person, and multi-group conditions. We demonstrate the advantages of M3 Act across three core experiments. The results suggest our synthetic dataset can significantly improve the performance of several downstream methods and replace real-world datasets to reduce cost. Notably, M3 Act improves the state-of-the-art MOTRv2 on DanceTrack dataset, leading to a hop on the leaderboard from 10thto 2ndplace. Moreover, M3 Act opens new research for controllable 3D group activity generation. We define multiple metrics and propose a competitive baseline for the novel task. Our code and data are available at our project page: http://cjerry1243.github.io/M3Act.

In this paper, we explore the capability of an agent to construct a logical sequence of action steps, thereby as-sembling a strategic procedural plan. This plan is cru-cial for navigating from an initial visual observation to a target visual outcome, as depicted in real-life instructional videos. Existing works have attained partial suc-cess by extensively leveraging various sources of information available in the datasets, such as heavy intermediate visual observations, procedural names, or natural language step-by-step instructions, for features or supervision signals. However, the task remains formidable due to the implicit causal constraints in the sequencing of steps and the variability inherent in multiple feasible plans. To tackle these intricacies that previous efforts have over-looked, we propose to enhance the agent's capabilities by infusing it with procedural knowledge. This knowledge, sourced from training procedure plans and structured as a directed weighted graph, equips the agent to better nav-igate the complexities of step sequencing and its poten-tial variations. We coin our approach KEPP, a novel Knowledge-Enhanced Procedure Planning system, which harnesses a probabilistic procedural knowledge graph ex-tracted from training data, effectively acting as a compre-hensive textbook for the training domain. Experimental evaluations across three widely-used datasets under set-tings of varying complexity reveal that KEPP attains su-perior, state-of-the-art results while requiring only mini-mal supervision. Code and trained model are available at https://github.com/Ravindu-Yasas-Nagasinghe/KEPP

Our goal is to learn a video representation that is useful for downstream procedure understanding tasks in instructional videos. Due to the small amount of available annotations, a key challenge in procedure understanding is to be able to extract from unlabeled videos the procedural knowledge such as the identity of the task (e.g., ‘make latte’), its steps (e.g., ‘pour milk’), or the potential next steps given partial progress in its execution. Our main insight is that instructional videos depict sequences of steps that repeat between instances of the same or different tasks, and that this structure can be well represented by a Procedural Knowledge Graph (PKG), where nodes are discrete steps and edges connect steps that occur sequentially in the instructional activities. This graph can then be used to generate pseudo labels to train a video representation that encodes the procedural knowledge in a more accessible form to generalize to multiple procedure understanding tasks. We build a PKG by combining information from a text-based procedural knowledge database and an unlabeled instructional video corpus and then use it to generate training pseudo labels with four novel pre-training objectives. We call this PKG-based pre-training procedure and the resulting model Paprika, Procedure-Aware PRe-training for Instructional Knowledge Acquisition. We evaluate Paprika on COIN and CrossTask for procedure understanding tasks such as task recognition, step recognition, and step forecasting. Paprika yields a video representation that improves over the state of the art: up to 11.23% gains in accuracy in 12 evaluation settings. Implementation is available at https://github.com/salesforce/paprika.

FSS (Few-shot segmentation) aims to segment a target class using a small number of labeled images (support set). To extract information relevant to the target class, a dominant approach in best performing FSS methods removes background features using a support mask. We observe that this feature excision through a limiting support mask introduces an information bottleneck in several challenging FSS cases, e.g., for small targets and/or inaccurate target boundaries. To this end, we present a novel method (MSI), which maximizes the support-set information by exploiting two complementary sources of features to generate super correlation maps. We validate the effectiveness of our approach by instantiating it into three recent and strong FSS methods. Experimental results on several publicly available FSS benchmarks show that our proposed method consistently improves performance by visible margins and leads to faster convergence. Our code and trained models are available at: https://github.com/moonsh/MSI-Maximize-Support-Set-Information

Digraph Representation Learning aims to learn representations for directed homogeneous graphs (digraphs). Prior work is largely constrained or has poor generalizability across tasks. Most Graph Neural Networks exhibit poor performance on digraphs due to the neglect of modeling neighborhoods and preserving asymmetry. In this paper, we address these notable challenges by leveraging hyperbolic collaborative learning from multi-ordered partitioned neighborhoods and asymmetry-preserving regularizers. Our resulting formalism, Digraph Hyperbolic Networks (D-HYPR), is versatile for multiple tasks including node classification, link presence prediction, and link property prediction. The efficacy of D-HYPR was meticulously examined against 21 previous techniques, using 8 real-world digraph datasets. D-HYPR statistically significantly outperforms the current state of the art. We release our code at https://github. com/hongluzhou/dhypr.

Digraph Representation Learning (DRL) aims to learn representations for directed homogeneous graphs (digraphs). Prior work in DRL is largely constrained (e.g., limited to directed acyclic graphs), or has poor generalizability across tasks (e.g., evaluated solely on one task). Most Graph Neural Networks (GNNs) exhibit poor performance on digraphs due to the neglect of modeling neighborhoods and preserving asymmetry. In this paper, we address these notable challenges by leveraging hyperbolic collaborative learning from multi-ordered and partitioned neighborhoods, and regularizers inspired by socio-psychological factors. Our resulting formalism, Digraph Hyperbolic Networks (D-HYPR) -- albeit conceptually simple -- generalizes to digraphs where cycles and non-transitive relations are common, and is applicable to multiple downstream tasks including node classification, link presence prediction, and link property prediction. In order to assess the effectiveness of D-HYPR, extensive evaluations were performed across 8 real-world digraph datasets involving 21 prior techniques. D-HYPR statistically significantly outperforms the current state of the art. We release our code at https://github.com/hongluzhou/dhypr

With the rise in popularity of short-term Human Trajectory Prediction (HTP), Long-Term Crowd Flow Prediction (LTCFP) has been proposed to forecast crowd movement in large and complex environments. However, the input representations, models, and datasets for LTCFP are currently limited. To this end, we propose Fourier Isovists, a novel input representation based on egocentric visibility, which consistently improves all existing models. We also propose GeoInteractNet (GINet), which couples the layers between a multi-scale attention network (M-SCAN) and a convolutional encoder-decoder network (CED). M-SCAN approximates a super-resolution map of where humans are likely to interact on the way to their goals and produces multi-scale attention maps. The CED then uses these maps in either its encoder's inputs or its decoder's attention gates, which allows GINet to produce super-resolution predictions with substantially higher accuracy than existing models even with Fourier Isovists. In order to evaluate the scalability of models to large and complex environments, which the only existing LTCFP dataset is unsuitable for, a new synthetic crowd dataset with both real and synthetic environments has been generated. In its nascent state, LTCFP has much to gain from our key contributions. The Supplementary Materials, dataset, and code are available at sssohn.github.io/GeoInteractNet.

Multiscale modeling has yielded immense success on various machine learning tasks. However, it has not been properly explored for the prominent task of information diffusion, which aims to understand how information propagates along users in online social networks. For a specific user, whether and when to adopt a piece of information propagated from another user is affected by complex interactions, and thus, is very challenging to model. Current state-of-the-art techniques invoke deep neural models with vector representations of users. In this paper, we present a Hierarchical Information Diffusion (HID) framework by integrating user representation learning and multiscale modeling. The proposed framework can be layered on top of all information diffusion techniques that leverage user representations, so as to boost the predictive power and learning efficiency of the original technique. Extensive experiments on three real-world datasets showcase the superiority of our method.

Personalized recommendation benefits users in accessing contents of interests effectively. Current research on recommender systems mostly focuses on matching users with proper items based on user interests. However, significant efforts are missing to understand how the recommendations influence user preferences and behaviors, e.g., if and how recommendations result in echo chambers. Extensive efforts have been made in examining the phenomenon in online media and social network systems. Meanwhile, there are growing concerns that recommender systems might lead to the self-reinforcing of user's interests due to narrowed exposure of items, which may be the potential cause of echo chamber. In this paper, we aim to analyze the echo chamber phenomenon in Alibaba Taobao --- one of the largest e-commerce platforms in the world.

Crowd simulation, the study of the movement of multiple agents in complex environments, presents a unique application domain for machine learning. One challenge in crowd simulation is to imitate the movement of expert agents in highly dense crowds. An imitation model could substitute an expert agent if the model behaves as good as the expert. This will bring many exciting applications. However, we believe no prior studies have considered the critical question of how training data and training methods affect imitators when these models are applied to novel scenarios. In this work, a general imitation model is represented by applying either the Behavior Cloning (BC) training method or a more sophisticated Generative Adversarial Imitation Learning (GAIL) method, on three typical types of data domains: standard benchmarks for evaluating crowd models, random sampling of state-action pairs, and egocentric scenarios that capture local interactions. Simulated results suggest that (i) simpler training methods are overall better than more complex training methods, (ii) training samples with diverse agent-agent and agent-obstacle interactions are beneficial for reducing collisions when the trained models are applied to new scenarios. We additionally evaluated our models in their ability to imitate real world crowd trajectories observed from surveillance videos. Our findings indicate that models trained on representative scenarios generalize to new, unseen situations observed in real human crowds.

Traveling salesman problem is a typical combinatorial optimization problem which can be effectively solved by genetic algorithm. This paper proposes an improved partheno-genetic algorithm to solve TSP problem. A comparison between our algorithm and the conventional partheno-genetic algorithm shows its effectiveness.