Achieving human-level intelligence requires refining the transition from the fast, intuitive System 1 to the slower, more deliberate System 2 reasoning. While System 1 excels in quick, heuristic decisions, System 2 relies on logical reasoning for more accurate judgments and reduced biases. Foundational Large Language Models (LLMs) excel at fast decision-making but lack the depth for complex reasoning, as they have not yet fully embraced the step-by-step analysis characteristic of true System 2 thinking. Recently, reasoning LLMs like OpenAI’s o1/o3 and DeepSeek’s R1 have demonstrated expert-level performance in fields such as mathematics and coding, closely mimicking the deliberate reasoning of System 2 and showcasing human-like cognitive abilities. This survey begins with a brief overview of the progress in foundational LLMs and the early development of System 2 technologies, exploring how their combination has paved the way for reasoning LLMs. Next, we discuss how to construct reasoning LLMs, trace the evolution of various reasoning models, and examine the core methods that enable advanced reasoning behind them. Additionally, we provide an overview of reasoning benchmarks, offering an in-depth comparison of the performance of representative reasoning LLMs. Finally, we explore promising directions for advancing reasoning LLMs and maintain a real-time GitHub Repository to track the latest developments. We hope this survey will serve as a valuable resource to inspire innovation and drive progress in this rapidly evolving field.

Despite the success of convolutional neural network (CNN) in conventional closed-set recognition (CSR), it still lacks robustness for dealing with unknowns (those out of known classes) in open environment. To improve the robustness of CNN in open-set recognition (OSR) and meanwhile maintain its high accuracy in CSR, we propose an alternative deep framework called convolutional prototype network (CPN), which keeps CNN for representation learning but replaces the closed-world assumed softmax with an open-world oriented and human-like prototype model. To equip CPN with discriminative ability for classifying known samples, we design several discriminative losses for training. Moreover, to increase the robustness of CPN for unknowns, we interpret CPN from the perspective of generative model and further propose a generative loss, which is essentially maximizing the log-likelihood of known samples and serves as a latent regularization for discriminative learning. The combination of discriminative and generative losses makes CPN a hybrid model with advantages for both CSR and OSR. Under the designed losses, the CPN is trained end-to-end for learning the convolutional network and prototypes jointly. For application of CPN in OSR, we propose two rejection rules for detecting different types of unknowns. Experiments on several datasets demonstrate the efficiency and effectiveness of CPN for both CSR and OSR tasks.

Background: Spinal cord ischemia-reperfusion injury (SCIRI) often leads to neurological damage and mortality. In this regard, understanding the pathology of SCIRI and preventing its development are of great clinic value. Methods: Herein, we analyzed the role of bone marrow mesenchymal stem cell (BMMSC)-derived exosomal microRNA (miR)-124-3p in SCIRI. A SCIRI rat model was established, and the expression of Ern1 and M2 macrophage polarization markers (Arg1, Ym1, and Fizz) was determined using immunohistochemistry, immunofluorescence assay, RT-qPCR, and western blot analysis. Targeting relationship between miR-124-3p and Ern1 was predicted using bioinformatic analysis and verified by dual-luciferase reporter assay. Macrophages were co-cultured with miR-124-3p-containing BMMSC-derived exosomes. M2 macrophages were identified using flow cytometry, and the expression of Arg1, Ym1, and Fizz was determined. In addition, SCIRI rats were injected with miR-124-3p-containing exosomes, spinal cord cell apoptosis was observed using TUNEL assay, and the pathological condition was evaluated with H&E staining. Results: In SCIRI, Ern1 was highly expressed and M2 polarization markers were poorly expressed. Silencing Ern1 led to elevated expression of M2 polarization markers. MiR-124-3p targeted and negatively regulated Ern1. Exosomal miR-124-3p enhanced M2 polarization. Highly expressed exosomal miR-124-3p impeded cell apoptosis and attenuated SCIRI-induced tissue impairment and nerve injury. miR-124-3p from BMMSC-derived exosomes ameliorated SCIRI and its associated nerve injury through inhibiting Ern1 and promoting M2 polarization. Conclusion: In summary, exosomal miR-124-3p derived from BMMSCs attenuated nerve injury induced by SCIRI by regulating Ern1 and M2 macrophage polarization.

Geometry problem solving (GPS) is a high-level mathematical reasoning requiring the capacities of multi-modal fusion and geometric knowledge application. Recently, neural solvers have shown great potential in GPS but still be short in diagram presentation and modal fusion. In this work, we convert diagrams into basic textual clauses to describe diagram features effectively, and propose a new neural solver called PGPSNet to fuse multi-modal information efficiently. Combining structural and semantic pre-training, data augmentation and self-limited decoding, PGPSNet is endowed with rich knowledge of geometry theorems and geometric representation, and therefore promotes geometric understanding and reasoning. In addition, to facilitate the research of GPS, we build a new large-scale and fine-annotated GPS dataset named PGPS9K, labeled with both fine-grained diagram annotation and interpretable solution program. Experiments on PGPS9K and an existing dataset Geometry3K validate the superiority of our method over the state-of-the-art neural solvers. Our code, dataset and appendix material are available at \url{https://github.com/mingliangzhang2018/PGPS}.

It has been widely observed that deep neural networks are highly vulnerable to adversarial examples. Decision-based attacks could generate adversarial examples based solely on top-1 labels returned by the target model. However, they typically make excessive queries and could not bypass detection effectively. To comprehensively assess a decision-based attack, besides its query efficiency, the performance against detection is also a concern. Considering that previous detections consume massive resources and always mistakenly recognize benign video frames as malicious attacks, we design a lightweight detection called boundary detection to overcome the above limitations, whose success reveals serious limitations of existing decision-based attacks. To develop more powerful attacks, we first present f-mixup as a basic method to produce candidate adversarial examples in the frequency domain. Using f-mixup as the building block, we propose f-attack as a complete decision-based attack. With the help of several natural images, f-attack could both work well with limited (hundreds of) queries and bypass detection effectively. Nevertheless, if the attacker could make relatively adequate (thousands of) queries and the target model is not equipped with detection, f-attack will lag behind existing decision-based attacks. We additionally introduce frequency binary search based on f-mixup , which serves as a plug-and-play module for existing decision-based attacks to further improve their query efficiency. Experimental results verify the effectiveness of our proposed methods.