The low-rank matrix approximation problem with respect to the component-wise ℓ 1 -norm (ℓ 1 -LRA), which is closely related to robust principal component analysis (PCA), has become a very popular tool in data mining and machine learning. Robust PCA aims at recovering a low-rank matrix that was perturbed with sparse noise, with applications for example in foreground-background video separation. Although ℓ 1 -LRA is strongly believed to be NP-hard, there is, to the best of our knowledge, no formal proof of this fact. In this paper, we prove that ℓ 1 -LRA is NP-hard, already in the rank-one case, using a reduction from MAX CUT. Our derivations draw interesting connections between ℓ 1 -LRA and several other well-known problems, namely, robust PCA, ℓ 0 -LRA, binary matrix factorization, a particular densest bipartite subgraph problem, the computation of the cut norm of {−1, +1} matrices, and the discrete basis problem, which we all prove to be NP-hard. ., √ p > |E||V |. Therefore, we choose p > |E| 2 |V | 2 and also p a power of 2.Corollary 2. It is NP-hard to compute the cut norm (3.1) of {−1, +1}-matrices.Corollary 3. Rank-one ℓ 0 -LRA, that is, problem (1.3) with r = 1, is NP-hard.Corollary 4. Rank-one BMF (2.1) is NP-hard.Remark 1 (The discrete basis problem and boolean matrix factorization). The discrete basis problem (DSP) [28], also known as Boolean matrix factorization, is similar to BMF and can be formulated a follows: given a binary matrix M ∈ {0, 1} m×n and a rank r, solvewhere 0 0 = 0, 0 1 = 1 and 1 1 = 1. For r = 1, BMF and DSP coincide, therefore our result also implies that rank-one DSP is NP-hard. Note that DSP is closely related to the rectangle covering problem which is equivalent to the minimum biclique cover problem in bipartite graph; see Fiorini et al. [13].