Knowledge Tracing Machines (KTMs) generalize several existing models, including Item Response Theory (IRT), the Additive Factor Model (AFM), Performance Factor Analysis (PFA), and Multidimensional Item Response Theory (MIRT).

Relation to IRT: When $d = 0$ (without embeddings), if the features are only users and items represented as one-hot encodings, the KTM expression becomes: $\log \frac{p(x)}{1 - p(x)} = \mu + w_i + w_{n+j} = \theta_i - d_i$ This means the KTM simplifies to the Rasch Model (1-PL IRT model).

Relation to AFM and PFA: If the features are skills, along with wins and fails at the skill level, and the Q-matrix between items and skills is known, encoding the weights as (beta_1, dots, beta_s, lambda_1, dots, lambda_s, delta_1, dots, delta_s) and features as (q_{j1}, dots, q_{js}, W_{i1}, dots, W_{is}, F_{i1}, dots, F_{is}) (where $W$ and $F$ are counters for successful and unsuccessful attempts) makes the KTM behave exactly like AFM and PFA.

Relation to MIRT: When $d > 0$, the KTM acts like MIRT.