Multiple instance learning (MIL) has been increasingly applied to the classification of histopathology whole slide images (WSI). While many classic attention-based MIL algorithms have achieved good classification performance, they do not account for the relationships between instances. Using transformers in MIL can effectively capture these relationships but requires substantial computational resources. To address the trade-off between performance and computational cost, we propose a new simplified attention-based MIL model—TKA-MIL, which only capture the relationships between all instances and critical instances. Additionally, we derive the probability of an instance being positive based on the attention weights of TKA-MIL. Experiments on the Camelyon16 and TCGA Lung Cancer datasets demonstrate that our approach achieves an optimal balance between classification performance and model complexity, and the derived instance probabilities are more accurate compared to directly using normalized attention weights.

3D Gaussian Splatting (3DGS) has become a popular solution in SLAM due to its high-fidelity and real-time novel view synthesis performance. However, some previous 3DGS SLAM methods employ a differentiable rendering pipeline for tracking, lack geometric priors in outdoor scenes. Other approaches introduce separate tracking modules, but they accumulate errors with significant camera movement, leading to scale drift. To address these challenges, we propose a robust RGB-only outdoor 3DGS SLAM method: S3PO-GS. Technically, we establish a self-consistent tracking module anchored in the 3DGS pointmap, which avoids cumulative scale drift and achieves more precise and robust tracking with fewer iterations. Additionally, we design a patch-based pointmap dynamic mapping module, which introduces geometric priors while avoiding scale ambiguity. This significantly enhances tracking accuracy and the quality of scene reconstruction, making it particularly suitable for complex outdoor environments. Our experiments on the Waymo, KITTI, and DL3DV datasets demonstrate that S3PO-GS achieves state-of-the-art results in novel view synthesis and outperforms other 3DGS SLAM methods in tracking accuracy. Project page: https://3dagentworld.github.io/S3PO-GS/.

Video Large Language Models (Video-LLMs) have made remarkable progress in video understanding tasks. However, they are constrained by the maximum length of input tokens, making it impractical to input entire videos. Existing frame selection approaches, such as uniform frame sampling and text-frame retrieval, fail to account for the information density variations in the videos or the complex instructions in the tasks, leading to sub-optimal performance. In this paper, we propose Frame-Voyager that learns to query informative frame combinations, based on the given textual queries in the task. To train Frame-Voyager, we introduce a new data collection and labeling pipeline, by ranking frame combinations using a pre-trained Video-LLM. Given a video of M frames, we traverse its T-frame combinations, feed them into a Video-LLM, and rank them based on Video-LLM's prediction losses. Using this ranking as supervision, we train Frame-Voyager to query the frame combinations with lower losses. In experiments, we evaluate Frame-Voyager on four Video Question Answering benchmarks by plugging it into two different Video-LLMs. The experimental results demonstrate that Frame-Voyager achieves impressive results in all settings, highlighting its potential as a plug-and-play solution for Video-LLMs.

Large vision-language models (LVLMs) have significantly advanced numerous fields. In this work, we explore how to harness their potential to address 3D scene understanding tasks, using 3D question answering (3D-QA) as a representative example. Due to the limited training data in 3D, we do not train LVLMs but infer in a zero-shot manner. Specifically, we sample 2D views from a 3D point cloud and feed them into 2D models to answer a given question. When the 2D model is chosen, e.g., LLAVA-OV, the quality of sampled views matters the most. We propose cdViews, a novel approach to automatically selecting critical and diverse Views for 3D-QA. cdViews consists of two key components: viewSelector prioritizing critical views based on their potential to provide answer-specific information, and viewNMS enhancing diversity by removing redundant views based on spatial overlap. We evaluate cdViews on the widely-used ScanQA and SQA benchmarks, demonstrating that it achieves state-of-the-art performance in 3D-QA while relying solely on 2D models without fine-tuning. These findings support our belief that 2D LVLMs are currently the most effective alternative (of the resource-intensive 3D LVLMs) for addressing 3D tasks.

3D Gaussian Splatting (3DGS) has become a popular solution in SLAM, as it can produce high-fidelity novel views. However, previous GS-based methods primarily target indoor scenes and rely on RGB-D sensors or pretrained depth estimation models, hence underperforming in outdoor scenarios. To address this issue, we propose a RGB-only gaussian splatting SLAM method for unbounded out-door scenes—OpenGS-SLAM. Technically, we first employ a pointmap regression network to generate consistent pointmaps between frames for pose estimation. Compared to commonly used depth maps, pointmaps include spatial relationships and scene geometry across multiple views, enabling robust camera pose estimation. Then, we propose integrating the estimated camera poses with 3DGS rendering as an end-to-end differentiable pipeline. Our method achieves simultaneous optimization of camera poses and 3DGS scene parameters, significantly enhancing system tracking accuracy. Specifically, we also design an adaptive scale mapper for the pointmap regression network, which provides more accurate pointmap mapping to the 3DGS map representation. Our experiments on the Waymo dataset demonstrate that OpenGS-SLAM reduces tracking error to 9.8% of previous 3DGS methods, and achieves state-of-the-art results in novel view synthesis. Project page: https://3dagentworld.github.io/pengs-slam/.

3D Gaussian Splatting (3DGS) has revolutionized 3D scene reconstruction, which effectively balances rendering quality, efficiency, and speed. However, existing 3DGS approaches usually generate plausible outputs and face significant challenges in complex scene reconstruction, manifesting as incomplete holistic structural outlines and unclear local lighting effects. To address these issues simultaneously, we propose a novel decoupled optimization framework, which integrates wavelet decomposition into 3D Gaussian Splatting and 2D sampling. Technically, through 3D wavelet decomposition, our approach divides point clouds into high-frequency and low-frequency components, enabling targeted optimization for each. The low-frequency component captures global structural outlines and manages the distribution of Gaussians through voxelization. In contrast, the high-frequency component restores intricate geometric and textural details while incorporating a relight module to mitigate lighting artifacts and enhance photorealistic rendering. Additionally, a 2D wavelet decomposition is applied to the training images, simulating radiance variations. This provides critical guidance for high-frequency detail reconstruction, ensuring seamless integration of details with the global structure. Extensive experiments on challenging datasets demonstrate our method achieves state-of-the-art performance across various metrics, surpassing existing approaches and advancing the field of 3D scene reconstruction.

This paper presents JavisGPT, the first unified multimodal large language model (MLLM) for joint audio-video (JAV) comprehension and generation. JavisGPT has a concise encoder-LLM-decoder architecture, which has a SyncFusion module for spatio-temporal audio-video fusion and synchrony-aware learnable queries to bridge a pretrained JAV-DiT generator. This design enables temporally coherent video-audio understanding and generation from multimodal instructions. We design an effective three-stage training pipeline consisting of multimodal pretraining, audio-video fine-tuning, and large-scale instruction-tuning, to progressively build multimodal comprehension and generation from existing vision-language models. For instruction tuning, we construct JavisInst-Omni, a high-quality instruction dataset with over 200K GPT-4o-curated audio-video-text dialogues that cover diverse and multi-level comprehension and generation scenarios. On JAV comprehension and generation benchmarks, our experiments show that JavisGPT outperforms existing MLLMs, particularly in complex and temporally synchronized settings.

In the realm of food computing, segmenting ingredients from images poses substantial challenges due to the large intra-class variance among the same ingredients, the emergence of new ingredients, and the high annotation costs as-sociated with large food segmentation datasets. Existing approaches primarily utilize a closed-vocabulary and static text embeddings setting. These methods often fall short in effectively handling the ingredients, particularly new and diverse ones. In response to these limitations, we introduce OVFoodSeg, a framework that adopts an open-vocabulary setting and enhances text embeddings with visual context. By integrating vision-language models (VLMs), our approach enriches text embedding with image-specific infor-mation through two innovative modules, e.g., an image-to-text learner FoodLearner and an Image-Informed Text Encoder. The training process of OVFoodSeg is divided into two stages: the pre-training of FoodLearner and the sub-sequent learning phase for segmentation. The pre-training phase equips FoodLearner with the capability to align visual information with corresponding textual representations that are specifically related to food, while the second phase adapts both the FoodLearner and the Image-Informed Text Encoder for the segmentation task. By addressing the de-ficiencies of previous models, OVFoodSeg demonstrates a significant improvement, achieving an 4.9% increase in mean Intersection over Union (mIoU) on the FoodSeg103 dataset, setting a new milestone for food image segmentation.

Speculative decoding is a relatively new decoding framework that leverages small and efficient draft models to reduce the latency of LLMs. In this study, we introduce GliDe and CaPE, two low-hassle modifications to vanilla speculative decoding to further improve the decoding speed of a frozen LLM. Specifically, GliDe is a modified draft model architecture that reuses the cached keys and values from the target LLM, while CaPE is a proposal expansion method that uses the draft model’s confidence scores to help select additional candidate tokens for verification. Extensive experiments on different benchmarks demonstrate that our proposed GliDe draft model significantly reduces the expected decoding latency. Additional evaluation using walltime reveals that GliDe can accelerate Vicuna models up to 2.17x and further extend the improvement to 2.61x with CaPE. We will release our code, data, and the trained draft models.

Out-of-distribution (OOD) settings are used to measure a model's performance when the distribution of the test data is different from that of the training data.NLU models are known to suffer in OOD settings (Utama et al., 2020b).We study this issue from the perspective of causality, which sees confounding bias as the reason for models to learn spurious correlations.While a common solution is to perform intervention, existing methods handle only known and single confounder (Pearl and Mackenzie, 2018), but in many NLU tasks the confounders can be both unknown and multifactorial.In this paper, we propose a novel interventional training method called Bottom-up Automatic Intervention (BAI) that performs multi-granular intervention with identified multifactorial confounders.Our experiments on three NLU tasks, namely, natural language inference, fact verification and paraphrase identification, show the effectiveness of BAI for tackling different OOD settings.1

Multiple-Choice Question Answering (MCQA) is one of the challenging tasks in machine reading comprehension. The main challenge in MCQA is to extract "evidence" from the given context that supports the correct answer. In OpenbookQA dataset [1], the requirement of extracting "evidence" is particularly important due to the mutual independence of sentences in the context. Existing work tackles this problem by annotated evidence or distant supervision with rules which overly rely on human efforts. To address the challenge, we propose a simple yet effective approach termed evidence filtering to model the relationships between the encoded contexts with respect to different options collectively, and to potentially highlight the evidence sentences and filter out unrelated sentences. In addition to the effective reduction of human efforts of our approach compared, through extensive experiments on OpenbookQA, we show that the proposed approach outperforms the models that use the same backbone and more training data; and our parameter analysis also demonstrates the interpretability of our approach.

Sicheng Yu, Hao Zhang, Yulei Niu, Qianru Sun, Jing Jiang. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Historically, the harm caused by insiders has proven to be one of the greatest concerns for any organization. As such, it has received considerable attention from both the industrial and research communities. Existing works mainly focused on modeling the employees’ normal biometric behavior (e.g., human to device interaction pattern) to detect anomalous behavior which corresponds to the insider activity. However, it is unattainable to stop the insider at the final moment when the malicious act is being carried out. In this paper, we propose a novel framework which performs employee profiling based on aspect-based sentiments and social network information and examine its applicability for early detection of potential insider threats. On the contrary to the traditional sentiment analysis , aspect-based sentiment analysis provides more fine-grained information on the employee. Our framework employs a combination of deep learning techniques such as Gated Recurrent Unit (GRU) and skipgram to build temporal sentiment profiles for the employees. It then performs anomaly detection on the profiles and ranks the employees based on their respective anomaly score. Due to the absence of relevant benchmark dataset, we augmented the publicly available real-world Enron email corpus with an insider threat scenario to evaluate our framework. The evaluation results demonstrate that the augmentation is indeed reflected in the augmented employee’s anomaly ranking (i.e., from normal to abnormal) and her close associates are indeed placed closely to her when the profiles are visualized in the 2D space. The profiles obtained from our framework can also be used to complement any existing expert and intelligent systems with additional capabilities in handling textual information such as, integration with profiles obtained from biometric behavior to form a more comprehensive threat detection system.