Supervised Learning

Machine learning is typically divided into three major paradigms: supervised learning, unsupervised learning, and reinforcement learning.

Machine learning is divided into three major categories: supervised learning, unsupervised learning, and semi-supervised learning.

Learning-based methods have stronger performance when data is sufficient, but at the cost of robustness under domain shift.

Unlike supervised learning, which relies on labeled data, reinforcement learning does not require humans to provide correct answers; instead, it allows agents to autonomously discover optimal behavioral paths through trial and error.

Supervised learning training uses labeled data consisting of sample pairs of known inputs and their corresponding outputs.

仅用成功轨迹训练的模型往往存在分布外泛化的严重缺陷，遇到工具返回错误时便丧失纠偏能力

强化学习中的智能体不依赖标注数据，而是通过与环境的持续交互，依据奖励信号不断调整策略

The fundamental distinction between supervised learning and unsupervised learning lies in whether labels are present in the training data.

The capability of supervised learning models is bounded by the ceiling of data quality and cannot surpass the level of the teacher algorithm.

Regression is used to predict continuous numerical values and is the most fundamental part of supervised learning.

Traditional structured information retrieval methods include rule-based pattern matching, query rewriting that relies on manual features, and supervised learning that requires large amounts of labeled data.

Examples of technical-route debates include: in model architecture, the debate between Transformer and hybrid architectures that alternate convolution and attention; in training methods, the trade-off between self-supervised learning and supervised learning; in deployment strategy, the weighing of cloud large models against on-device small models.

Exploration vs. exploitation is the core challenge that distinguishes RL from supervised learning: the agent needs to trade off between trying new actions to gain information and exploiting the known best action to obtain rewards.

Common technical approaches to making engines "human-like" include supervised learning on human games, behavior cloning, and sampling based on human mistake distributions.

The fundamental difference between reinforcement learning and supervised learning is that the data distribution changes with the agent's behavior, forcing it to deal with the consequences of its own actions.

The optimization landscape of RLVR is fundamentally different from supervised learning: the reward signal is sparse and discrete, gradient estimates come from sampled responses, and the variance is significantly higher than in traditional training settings.

This study transforms LLM post-editing from an inference-time prompt engineering problem into a supervised learning problem, using oracle information to guide the model in learning when to select and when to rewrite.

Supervised learning requires humans to provide labeled data categories, unsupervised learning automatically categorizes and clusters data objects, and reinforcement learning guides model learning through a reward mechanism.

Earnings forecasting tasks have clear objective standards—whether a prediction is accurate becomes immediately apparent once the actual earnings are released—making them a supervised learning scenario well suited to machine learning

The mainstream approaches to depth estimation fall into two main categories: geometry-based multi-view stereo (MVS) and deep-learning-based monocular depth estimation.