As a foundation of large language models, fine-tuning drives rapid progress, broad applicability, and profound impacts on human–AI collaboration, surpassing earlier technological advancements. This paper provides a comprehensive overview of large language model (LLM) fine-tuning by integrating hermeneutic theories of human comprehension, with a focus on the essential cognitive conditions that underpin this process. Drawing on Gadamer’s concepts of Vorverständnis, Distanciation, and the Hermeneutic Circle, the paper explores how LLM fine-tuning evolves from initial learning to deeper comprehension, ultimately advancing toward self-awareness. It examines the core principles, development, and applications of fine-tuning techniques, emphasizing its growing significance across diverse field and industries. The paper introduces a new term, “Tutorial Fine-Tuning (TFT)”, which annotates a process of intensive tuition given by a “tutor” to a small number of “students”, to define the latest round of LLM fine-tuning advancements. By addressing key challenges associated with fine-tuning, including ensuring adaptability, precision, credibility and reliability, this paper explores potential future directions for the co-evolution of humans and AI. By bridging theoretical perspectives with practical implications, this work provides valuable insights into the ongoing development of LLMs, emphasizing their potential to achieve higher levels of cognitive and operational intelligence.

Depression is the result of a complex interaction of social, psychological and physiological elements. Research into the brain disorders of patients suffering from depression can help doctors to understand the pathogenesis of depression and facilitate its diagnosis and treatment. Functional near-infrared spectroscopy (fNIRS) is a non-invasive approach to the detection of brain functions and activities. In this paper, a comprehensive fNIRS-based depression-processing architecture, including the layers of source, feature and model, is first established to guide the deep modeling for fNIRS. In view of the complexity of depression, we propose a methodology in the time and frequency domains for feature extraction and deep neural networks for depression recognition combined with current research. It is found that compared to non-depression people, patients with depression have a weaker encephalic area connectivity and lower level of activation in the prefrontal lobe during brain activity. Finally, based on raw data, manual features and channel correlations, the AlexNet model shows the best performance, especially in terms of the correlation features and presents an accuracy rate of 0.90 and a precision rate of 0.91, which is higher than ResNet18 and machine-learning algorithms on other data. Therefore, the correlation of brain regions can effectively recognize depression (from cases of non-depression), making it significant for the recognition of brain functions in the clinical diagnosis and treatment of depression.

In recent years, depression has become an increasingly serious problem globally. Previous studies of automatic depression recognition based on functional near-Infrared spectroscopy (fNIRS) or other brain imaging techniques have shown potential to serve as auxiliary diagnosis methods that provide assistance to medical professionals. Recently, some studies have found that, besides directly using the data themselves (temporal data), the use of functional connectivity among channels (spatial data) also can be effective. In this paper, we propose a method based on Graph Neural Network (GNN) that combines both temporal and spatial features of fNIRS data for automatic depression recognition. Specifically, fNIRS data of 96 subjects were collected and pre-processed. Basic statistical metrics of each channel were extracted as temporal features, and channel connectivity (coherence and correlation) were calculated as spatial features. Point-biserial analysis was conducted on these features and depression labels as a data-driven motivation. For classification, we considered data of each subject as a graph, with temporal features as node features and spatial features as edge weights. The graphs were fed into GNNs for training and testing. Experimental results showed that our GNN-based methods realized the best depression recognition performance compared with classical machine-learning methods regarding accuracy, F1 score, and precision, especially in F1 score for over 10%.

Objective: To investigate the volumetric and functional connectivity of the habenular nucleus in treatment‐resistant depression (TRD) patients using the resting‐state functional magnetic resonance imaging (rs‐fMRI) approach.Methods: A total of 15 TRD patients, who visited the Mental Health Institute of the First Hospital Affiliated with Jilin University between August 2014 and March 2015, along with 15 normal subjects, were enrolled into this study for structural and functional imaging. Functional connectivity analysis was performed using bilateral habenular nuclei as the region of interest in contrast to whole‐brain voxels.Results: No significant difference of absolute volume was found in bilateral habenular nuclei between TRD patients and healthy controls, or after controlling for individual total intracranial volume. However, functional connectivity analysis showed increased connectivity between the right habenular nucleus with the medial superior frontal gyrus, anterior cingulate cortex and medial orbitofrontal gyrus, and decreased connectivity with the corpus callosum in the TRD group. For the left habenular nucleus seed, the brain region with increased functional connectivity in the inferior temporal gyrus and decreased functional connectivity in the insular was found in the TRD patients.Conclusion: Abnormal functional connectivity was present between the habenular nucleus and the default mode network in TRD patients. Dysfunction in habenular nucleus‐related circuitry for processing negative emotion might form the pathological basis for TRD. Significant asymmetric functional connectivity was also found between bilateral habenular nuclei in TRD patients. Such asymmetry suggests potentially divergent strategy for intervention on bilateral habenular nucleus regions in the future management of depression.

Most researchers focus on designing accurate crowd counting models with heavy parameters and computations but ignore the resource burden during the model deployment. A real-world scenario demands an efficient counting model with low-latency and high-performance. Knowledge distillation provides an elegant way to transfer knowledge from a complicated teacher model to a compact student model while maintaining accuracy. However, the student model receives the wrong guidance with the supervision of the teacher model due to the inaccurate information understood by the teacher in some cases. In this paper, we propose a dual-knowledge distillation (DKD) framework, which aims to reduce the side effects of the teacher model and transfer hierarchical knowledge to obtain a more efficient counting model. First, the student model is initialized with global information transferred by the teacher model via adaptive perspectives. Then, the self-knowledge distillation forces the student model to learn the knowledge by itself, based on intermediate feature maps and target map. Specifically, the optimal transport distance is utilized to measure the difference of feature maps between the teacher and the student to perform the distribution alignment of the counting area. Extensive experiments are conducted on four challenging datasets, demonstrating the superiority of DKD. When there are only approximately 6% of the parameters and computations from the original models, the student model achieves a faster and more accurate counting performance as the teacher model even surpasses it.

Due to the expensive and laborious annotations of labeled data required by fully-supervised learning in the crowd counting task, it is desirable to explore a method to reduce the labeling burden. There exists a large number of unlabeled images in the wild that can be easily obtained compared to labeled datasets. Based on the characteristics of consistent spatial transformation with the annotations of heads and image, this paper proposes a self-supervised learning framework with unlabeled and limited labeled data for pre-training and fine-tuning crowd counting model (SSL-FT). It includes an online network and a target network that receive the same image but are randomly processed by two defined augmentation transformations. We leverage unlabeled data to pre-train the online network based on a self-supervised loss and small-scale labeled data to transfer the model to a specific domain based on a fully-supervised loss. We demonstrate the effectiveness of the SSL-FT on four public datasets including ShanghaiTech PartA, PartB, UCF-QNRF and WorldExpo'10 utilizing a classical counting model. Experimental results show that our approach performs better than state-of-art semi-supervised methods.