Answering questions with Chain-of-Thought (CoT) has significantly enhanced the reasoning capabilities of Large Language Models (LLMs), yet its impact on Large Multimodal Models (LMMs) still lacks a systematic assessment and in-depth investigation. In this paper, we introduce MME-CoT, a specialized benchmark evaluating the CoT reasoning performance of LMMs, spanning six domains: math, science, OCR, logic, space-time, and general scenes. As the first comprehensive study in this area, we propose a thorough evaluation suite incorporating three novel metrics that assess the reasoning quality, robustness, and efficiency at a fine-grained level. Leveraging curated high-quality data and a unique evaluation strategy, we conduct an in-depth analysis of state-of-the-art LMMs, uncovering several key insights: 1) Models with reflection mechanism demonstrate a superior CoT quality, with Kimi k1.5 outperforming GPT-4o and demonstrating the highest quality results; 2) CoT prompting often degrades LMM performance on perception-heavy tasks, suggesting a potentially harmful overthinking behavior; and 3) Although the CoT quality is high, LMMs with reflection exhibit significant inefficiency in both normal response and self-correction phases. We hope MME-CoT serves as a foundation for advancing multimodal reasoning in LMMs. Project Page: https://mmecot.github.io/

The field of robotic grasping has seen significant advancements in recent years, with deep learning and vision-language models driving progress towards more intelligent and adaptable grasping systems [1, 2, 3]. However, robotic grasping in highly cluttered environments remains a major challenge, as target objects are often severely occluded or completely hidden [4, 5, 6]. Even stateof-the-art methods struggle to accurately identify and grasp objects in such scenarios. To address this challenge, we propose ThinkGrasp, which combines the strength of large-scale pretrained vision-language models with an occlusion handling system. ThinkGrasp leverages the advanced reasoning capabilities of models like GPT-4o [7] to gain a visual understanding of environmental and object properties such as sharpness and material composition. By integrating this knowledge through a structured prompt-based chain of thought, ThinkGrasp can significantly enhance success rates and ensure the safety of grasp poses by strategically eliminating obstructing objects. For instance, it prioritizes larger and centrally located objects to maximize visibility and access and focuses on grasping the safest and most advantageous parts, such as handles or flat surfaces. Unlike VL-Grasp[8], which relies on the RoboRefIt dataset for robotic perception and reasoning, ThinkGrasp benefits from GPT-4o's reasoning and generalization capabilities. This allows ThinkGrasp to intuitively select the right objects and achieve higher performance in complex environments, as demonstrated by our comparative experiments.

Decoding of seen visual contents with non-invasive brain recordings has important scientific and practical values. Efforts have been made to recover the seen images from brain signals. However, most existing approaches cannot faithfully reflect the visual contents due to insufficient image quality or semantic mismatches. Compared with reconstructing pixel-level visual images, speaking is a more efficient and effective way to explain visual information. Here we introduce a non-invasive neural decoder, termed as MindGPT, which interprets perceived visual stimuli into natural languages from fMRI signals. Specifically, our model builds upon a visually guided neural encoder with a cross-attention mechanism, which permits us to guide latent neural representations towards a desired language semantic direction in an end-to-end manner by the collaborative use of the large language model GPT. By doing so, we found that the neural representations of the MindGPT are explainable, which can be used to evaluate the contributions of visual properties to language semantics. Our experiments show that the generated word sequences truthfully represented the visual information (with essential details) conveyed in the seen stimuli. The results also suggested that with respect to language decoding tasks, the higher visual cortex (HVC) is more semantically informative than the lower visual cortex (LVC), and using only the HVC can recover most of the semantic information. The code of the MindGPT model will be publicly available at https://github.com/JxuanC/MindGPT.

Brain-computer interfaces (BCIs) provide a direct pathway from the brain to external devices and have demonstrated great potential for assistive and rehabilitation technologies. Endogenous BCIs based on electroencephalogram (EEG) signals, such as motor imagery (MI) BCIs, can provide some level of control. However, mastering spontaneous BCI control requires the users to generate discriminative and stable brain signal patterns by imagery, which is challenging and is usually achieved over a very long training time (weeks/months). Here, we propose a human-machine joint learning framework to boost the learning process in endogenous BCIs, by guiding the user to generate brain signals towards an optimal distribution estimated by the decoder, given the historical brain signals of the user. To this end, we firstly model the human-machine joint learning process in a uniform formulation. Then a human-machine joint learning framework is proposed: 1) for the human side, we model the learning process in a sequential trial-and-error scenario and propose a novel ``copy/new'' feedback paradigm to help shape the signal generation of the subject toward the optimal distribution; 2) for the machine side, we propose a novel adaptive learning algorithm to learn an optimal signal distribution along with the subject's learning process. Specifically, the decoder reweighs the brain signals generated by the subject to focus more on ``good'' samples to cope with the learning process of the subject. Online and psuedo-online BCI experiments with 18 healthy subjects demonstrated the advantages of the proposed joint learning process over co-adaptive approaches in both learning efficiency and effectiveness.

Spiking Neural Networks (SNNs) are biologically realistic and practically promising in low-power computation because of their event-driven mechanism. Usually, the training of SNNs suffers accuracy loss on various tasks, yielding an inferior performance compared with ANNs. A conversion scheme is proposed to obtain competitive accuracy by mapping trained ANNs' parameters to SNNs with the same structures. However, an enormous number of time steps are required for these converted SNNs, thus losing the energy-efficient benefit. Utilizing both the accuracy advantages of ANNs and the computing efficiency of SNNs, a novel SNN training framework is proposed, namely layer-wise ANN-to-SNN knowledge distillation (LaSNN). In order to achieve competitive accuracy and reduced inference latency, LaSNN transfers the learning from a well-trained ANN to a small SNN by distilling the knowledge other than converting the parameters of ANN. The information gap between heterogeneous ANN and SNN is bridged by introducing the attention scheme, the knowledge in an ANN is effectively compressed and then efficiently transferred by utilizing our layer-wise distillation paradigm. We conduct detailed experiments to demonstrate the effectiveness, efficacy, and scalability of LaSNN on three benchmark data sets (CIFAR-10, CIFAR-100, and Tiny ImageNet). We achieve competitive top-1 accuracy compared to ANNs and 20x faster inference than converted SNNs with similar performance. More importantly, LaSNN is dexterous and extensible that can be effortlessly developed for SNNs with different architectures/depths and input encoding methods, contributing to their potential development.

Autoregressive language modeling (ALM) have been successfully used in self-supervised pre-training in Natural language processing (NLP). However, this paradigm has not achieved comparable results with other self-supervised approach in computer vision (e.g., contrastive learning, mask image modeling). In this paper, we try to find the reason why autoregressive modeling does not work well on vision tasks. To tackle this problem, we fully analyze the limitation of visual autoregressive methods and proposed a novel stochastic autoregressive image modeling (named SAIM) by the two simple designs. First, we employ stochastic permutation strategy to generate effective and robust image context which is critical for vision tasks. Second, we create a parallel encoder-decoder training process in which the encoder serves a similar role to the standard vision transformer focus on learning the whole contextual information, and meanwhile the decoder predicts the content of the current position, so that the encoder and decoder can reinforce each other. By introducing stochastic prediction and the parallel encoder-decoder, SAIM significantly improve the performance of autoregressive image modeling. Our method achieves the best accuracy (83.9%) on the vanilla ViT-Base model among methods using only ImageNet-1K data. Transfer performance in downstream tasks also show that our model achieves competitive performance.

In this paper, we consider sequential online prediction (SOP) for streaming data in the presence of outliers and change points. We propose an INstant TEmporal structure Learning (INTEL) algorithm to address this problem.Our INTEL algorithm is developed based on a full consideration to the duality between online prediction and anomaly detection. We first employ a mixture of weighted GP models (WGPs) to cover the expected possible temporal structures of the data. Then, on the basis of the rich modeling capacity of this WGP mixture, we develop an efficient technique to instantly learn (capture) the temporal structure of the data that follows a regime shift. This instant learning is achieved only by adjusting one hyper-parameter value of the mixture model. A weighted generalization of the product of experts (POE) model is used for fusing predictions yielded from multiple GP models. An outlier is declared once a real observation seriously deviates from the fused prediction. If a certain number of outliers are consecutively declared, then a change point is declared. Extensive experiments are performed using a diverse of real datasets. Results show that the proposed algorithm is significantly better than benchmark methods for SOP in the presence of outliers and change points.