Tata Institute of Fundamental Research
A Chernoff-Stein Lemma for adversarial hypothesis testing
STCS Student Seminar
Speaker:
Eeshan Modak
Organiser:
Agniv Bandyopadhyay
Date:
Friday, 17 Feb 2023, 16:00 to 17:00
Venue:
A201
(Scan to add to calendar)
Abstract:
In the composite hypothesis testing setting, the detector receives n i.i.d. samples either from a distribution p∈P or from a distribution q∈Q. It then decides the correct set from which the samples were drawn.
Brandao, Harrow, Lee and Peres [BHLP] considered an adaptive generalization of this problem where the choice of p∈P and q∈Q can change in each sample in some way that depends arbitrarily on the previous samples. Initially, one might think that this might confuse the detector and worsen the optimal error exponent. But, [BHLP] showed that the exponent remains the same and a simple maximum likelihood ratio test achieves it.
Link to the paper: https://arxiv.org/abs/1308.6702
| Speaker: | Eeshan Modak |
| Organiser: | Agniv Bandyopadhyay |
| Date: | Friday, 17 Feb 2023, 16:00 to 17:00 |
| Venue: | A201 |
(Scan to add to calendar)
Abstract:
In the composite hypothesis testing setting, the detector receives n i.i.d. samples either from a distribution p∈P or from a distribution q∈Q. It then decides the correct set from which the samples were drawn.
Brandao, Harrow, Lee and Peres [BHLP] considered an adaptive generalization of this problem where the choice of p∈P and q∈Q can change in each sample in some way that depends arbitrarily on the previous samples. Initially, one might think that this might confuse the detector and worsen the optimal error exponent. But, [BHLP] showed that the exponent remains the same and a simple maximum likelihood ratio test achieves it.
Link to the paper: https://arxiv.org/abs/1308.6702
Brandao, Harrow, Lee and Peres [BHLP] considered an adaptive generalization of this problem where the choice of p∈P and q∈Q can change in each sample in some way that depends arbitrarily on the previous samples. Initially, one might think that this might confuse the detector and worsen the optimal error exponent. But, [BHLP] showed that the exponent remains the same and a simple maximum likelihood ratio test achieves it.
Link to the paper: https://arxiv.org/abs/1308.6702