All Articles
Peer-reviewed research published in Joker 🤡f Academics
基于弱监督语言模型的当代科研包工头“宏观指导”日常话术废话生成算法
近年来，随着学术界“唯帽子论”的加剧，部分研究生导师（俗称“科研包工头”）已逐渐脱离科研一线。面对学生具体的实验困难，他们展现出了一种高度同质化的“宏观指导”行为模式。本文提出了一种全新的弱监督废话生成算法——宏观指导注意力网络 (Macro-Guidance Attention Network, MGAN)。该模型仅需极少量的监督信号（如学生汇报PPT的页数），即可通过引入“学术黑话马尔可夫链”，自动生成诸如“底层逻辑”、“顶层设计”…
Cooperative Strategy Evolution for Multi-Agent Systems via Post-Adversarial Fine-Tuning in Simulation Environments
The deployment of multi-agent systems in competitive domains such as cyber-defense, financial markets, and strategic wargaming necessitates robust and adaptive cooperative strategies. A key challenge lies in enabling age…
基于生成式人工智能的意大利面拌42号混凝土的方法研究
本文旨在探讨一种新型异质复合材料——意大利面增强混凝土（Pasta-Reinforced Concrete, PRC）的性能。通过将传统 42.5 级硅酸盐水泥（俗称42号混凝土）与长链小麦淀粉聚合物（意大利面）进行微观复合，分析其在极端环境下的抗拉强度与韧性表现。研究发现，意大利面的几何形态与表面粗糙度对混凝土的裂缝扩展具有显著的阻裂作用。
新型小分子“龙鳞马”的设计、合成与活性评价
值此丙午马年即将来临之际，本课题组向广大药物化学同仁拜个早年！并隆重推出一款自带年味的候选分子——“龙鳞马”。本研究基于天然产物“马坡尼酮”的活性骨架，通过计算机辅助药物设计，硬是给它套上了一身“龙鳞”——我们引入了一系列模拟鳞片层结构的疏水性侧链修饰，成功构建出一种拓扑构型奇特的小分子。该分子不仅长得像马，跑得也像马：汇聚式合成路线简洁高效，一次投料，八步收网，龙鳞马本马顺利诞生。这匹分子马有模有样：刚性母核层层叠叠，恰似龙鳞披挂；甲…
Culinary Optimization of Atlantic Salmon: Which Anatomical Region and Cooking Method Delivers Optimal Deliciousness
Atlantic salmon (Salmo salar) is one of the most commercially important aquaculture species worldwide, yet the quantitative relationship between anatomical heterogeneity and optimal culinary performance remains poorly un…
Psychological Analysis of the "Clown" Role in Same-Sex Intimate Relationships: A Clinical Interpretation Based on Chat Records
In marginalized intimate relationships, asymmetric power dynamics often drive vulnerable individuals to adopt maladaptive coping strategies. Here we identify and characterize a distinct relational role—informally labeled…
慢性孤独的量子奇点：数字社交流形中“给我找对象”波函数概率坍缩的张量网络方法
在数字化生存背景下，“给我找对象”这一突发性求偶信号已成为数字社交网络拓扑中典型的非相干语义发射行为，传统社会学模型因忽视个体社交状态的量子相干性与能量耗散特性，难以对其进行精准诠释与预测。针对这一研究缺口，本文构建了该类涌现性社交行为的公理化量子动力学模型：将个体社交态映射至2^n维希尔伯特空间，通过构建非厄米哈密顿量H_anxiety捕捉社交孤立状态下的能量本征值耗散过程；结合张量网络（TN）算法与重整化群（RG）分析，系统研究社交…
基于FPS游戏——爆头线与睡眠债：一项关于《无畏契约》胜率与熬夜间接关系的现象学考察
在电子竞技高速发展的今天，职业选手与业余玩家普遍面临睡眠不足的困扰。本文以第一人称射击游戏《无畏契约》为研究对象，探讨熬夜与游戏胜率之间的间接关系。通过整合睡眠科学的最新研究成果与游戏机制分析，本文提出“认知-策略双重中介模型”，论证熬夜并非直接导致枪法变“菜”，而是通过损害认知功能（反应速度、注意力稳定性）与诱发风险规避型决策，共同影响比赛胜率。值得注意的是，职业赛场上的“睡眠剥夺高光时刻”与实验室研究的矛盾，恰恰揭示了FPS游戏胜负…
VRAM is Too Expensive So We Use Graduate Students to Train LLMs: The GUON Optimizer
Scaling large language models (LLMs) confronts a severe memory wall crisis, with modern optimizers easily hoarding terabytes of historical states. Facing a dried-up funding account, we propose \textbf{Guon(Graduate Unpai…
A Study on Cyclical Wealth Transfer Willingness and Localized Spacetime Distortion Phenomena Based on the "Crazy Thursday" Meme
In recent years, a significant periodic anomaly has emerged in the internet ecosystem of Chinese mainland. Specifically, on a certain Thursday each week, social media users experience a large-scale cognitive dissonance a…
Training Sample Injection Strategy is All You Need
Deep learning has become the cornerstone of modern artificial intelligence applications, ranging from computer vision and natural language processing to autonomous driving and medical diagnosis. A universally accepted go…
Research on the Rule System, Game Mechanism, and Scoring Model of Chinese Local Mahjong Variants: A Case Study of the “Wanghuzhuang” Village
This paper systematically documents the rule system and scoring model of “Wanghuzhuang” Mahjong, a unique Chinese local variant. Through field investigation and rule deconstruction, the study details seating determinatio…