The number of walks of length d between vertices i and j is equal to the (i, j)-th element of Ad.
Spectral graph theory is about how eigenvalues, eigenvectors, and other linear-algebraic quantities give us useful information about a graph, for example about how well-connected it is, how well we can cluster or color the nodes, and how quickly random walks converge to a limiting distribution.
[2] In the context of Spectral graph theory the eigenvectors and the eigenvalues of graph's Adjacency matrix provide valid and essential insights into several structural properties such as connectivity, clustering, coloring, and the behavior of random walks, these insights are tightly tied to the adjacency algebra generated by the Adjacency matrix, including all its powers and linear combinations.
Both concepts are concerned with the adjacency matrix but approach it differently and are looked with different perspectives, Spectral graph theory focuses more on the specific spectral properties (eigenvalues and eigenvectors) to extract information about the graph's connectivity.
Adjacency algebra can be utilized in network analysis and graph theory, where the powers of the adjacency matrix can help to determine how connected a specific graph is by tracking paths of length k between vertices.