Computational Information Geometry for Binary Classification of High-Dimensional Random Tensors - Archive ouverte HAL Access content directly
Journal Articles Entropy Year : 2018

## Computational Information Geometry for Binary Classification of High-Dimensional Random Tensors

Gia-Thuy Pham
Remy Boyer
Frank Nielsen
• Function : Author
• PersonId : 953872

#### Abstract

Evaluating the performance of Bayesian classification in a high-dimensional random tensor is a fundamental problem, usually difficult and under-studied. In this work, we consider two Signal to Noise Ratio (SNR)-based binary classification problems of interest. Under the alternative hypothesis, i.e., for a non-zero SNR, the observed signals are either a noisy rank-R tensor admitting a Q-order Canonical Polyadic Decomposition (CPD) with large factors of size N q × R, i.e., for 1 ≤ q ≤ Q, where R, N q → ∞ with R 1/q /N q converge towards a finite constant or a noisy tensor admitting TucKer Decomposition (TKD) of multilinear (M 1 ,. .. , M Q)-rank with large factors of size N q × M q , i.e., for 1 ≤ q ≤ Q, where N q , M q → ∞ with M q /N q converge towards a finite constant. The classification of the random entries (coefficients) of the core tensor in the CPD/TKD is hard to study since the exact derivation of the minimal Bayes' error probability is mathematically intractable. To circumvent this difficulty, the Chernoff Upper Bound (CUB) for larger SNR and the Fisher information at low SNR are derived and studied, based on information geometry theory. The tightest CUB is reached for the value minimizing the error exponent, denoted by s. In general, due to the asymmetry of the s-divergence, the Bhattacharyya Upper Bound (BUB) (that is, the Chernoff Information calculated at s = 1/2) cannot solve this problem effectively. As a consequence, we rely on a costly numerical optimization strategy to find s. However, thanks to powerful random matrix theory tools, a simple analytical expression of s is provided with respect to the Signal to Noise Ratio (SNR) in the two schemes considered. This work shows that the BUB is the tightest bound at low SNRs. However, for higher SNRs, the latest property is no longer true.

#### Domains

Statistics [stat] Other Statistics [stat.ML]

### Dates and versions

hal-01758637 , version 1 (04-04-2018)

### Identifiers

• HAL Id : hal-01758637 , version 1
• DOI :

### Cite

Gia-Thuy Pham, Remy Boyer, Frank Nielsen. Computational Information Geometry for Binary Classification of High-Dimensional Random Tensors. Entropy, 2018, 20 (3), pp.203. ⟨10.3390/e20030203⟩. ⟨hal-01758637⟩

### Export

BibTeX TEI Dublin Core DC Terms EndNote Datacite

169 View