Sequential Tests of Statistical Hypotheses.

Multivariate time series (MTS) datasets are common in various multimedia, medical and financial applications. In order to efficiently perform k nearest neighbor searches for MTS datasets, we present a similarity measure, Eros (extended frobenius norm), an index structure, Muse (multilevel distance-based index structure for Eros), and a feature subset selection technique, Ropes (recursive feature elimination on common principal components for Eros). Eros is based on principal component analysis, and computes the similarity between two MTS items by measuring how close the corresponding principal components are using the eigenvalues as weights. Muse constructs each level as a distance-based index structure without using the weights, up to z levels, which are combined at the query time with the weights. Ropes utilizes both the common principal components and the weights recursively in order to select a subset of features for Eros. The experimental results show the superiority of our techniques as compared to earlier approaches.

An efficient k nearest neighbor search for multivariate time series

It is well-known that the most existing machine learning (ML)-based safety-critical applications are vulnerable to carefully crafted input instances called adversarial examples (AXs). An adversary can conveniently attack these target systems from digital as well as physical worlds. This paper aims to the generation of robust physical AXs against face recognition systems. We present a novel smoothness loss function and a patch-noise combo attack for realizing powerful physical AXs. The smoothness loss interjects the concept of delayed constraints during the attack generation process, thereby causing better handling of optimization complexity and smoother AXs for the physical domain. The patch-noise combo attack combines patch noise and imperceptibly small noises from different distributions to generate powerful registration-based physical AXs. An extensive experimental analysis found that our smoothness loss results in robust and more transferable digital and physical AXs than the conventional techniques. Notably, our smoothness loss results in a 1.17 and 1.97 times better mean attack success rate (ASR) in physical white-box and black-box attacks, respectively. Our patch-noise combo attack furthers the performance gains and results in 2.39 and 4.74 times higher mean ASR than conventional technique in physical world white-box and black-box attacks, respectively.

Powerful Physical Adversarial Examples Against Practical Face Recognition Systems

It is well-known that the most existing machine learning (ML)-based safety-critical applications are vulnerable to carefully crafted input instances called adversarial examples (AXs). An adversary can conveniently attack these target systems from digital as well as physical worlds. This paper aims to the generation of robust physical AXs against face recognition systems. We present a novel smoothness loss function and a patch-noise combo attack for realizing powerful physical AXs. The smoothness loss interjects the concept of delayed constraints during the attack generation process, thereby causing better handling of optimization complexity and smoother AXs for the physical domain. The patch-noise combo attack combines patch noise and imperceptibly small noises from different distributions to generate powerful registration-based physical AXs. An extensive experimental analysis found that our smoothness loss results in robust and more transferable digital and physical AXs than the conventional techniques. Notably, our smoothness loss results in a 1.17 and 1.97 times better mean attack success rate (ASR) in physical white-box and black-box attacks, respectively. Our patch-noise combo attack furthers the performance gains and results in 2.39 and 4.74 times higher mean ASR than conventional technique in physical world white-box and black-box attacks, respectively. 

/pdf/powerful-physical-adversarial-examples-against-practical-16i8at6x.pdf

We propose a model for multiclass classification of time series to make a prediction as early and as accurate as possible. The matrix sequential probability ratio test (MSPRT) is known to be asymptotically optimal for this setting, but contains a critical assumption that hinders broad real-world applications; the MSPRT requires the underlying probability density. To address this problem, we propose to solve density ratio matrix estimation (DRME), a novel type of density ratio estimation that consists of estimating matrices of multiple density ratios with constraints and thus is more challenging than the conventional density ratio estimation. We propose a log-sum-exp-type loss function (LSEL) for solving DRME and prove the following: (i) the LSEL provides the true density ratio matrix as the sample size of the training set increases (consistency); (ii) it assigns larger gradients to harder classes (hard class weighting effect); and (iii) it provides discriminative scores even on class-imbalanced datasets (guess-aversion). Our overall architecture for early classification, MSPRT-TANDEM, statistically significantly outperforms baseline models on four datasets including action recognition, especially in the early stage of sequential observations. Our code and datasets are publicly available at: this https URL.

The Power of Log-Sum-Exp: Sequential Density Ratio Matrix Estimation for Speed-Accuracy Optimization

DNN-based face verification systems are vulnerable to adversarial examples. The previous paper’s evaluation protocol (scenario), which we called the probe-dependent attack scenario, was unrealistic. We define a more practical attack scenario, the probe-agnostic attack. We empirically show that these attacks are more challenging than probe-dependent ones. We propose a simple and effective method, PAMTAM, to improve the attack success rate for probe-agnostic attacks. We show that PAMTAM successfully improves the attack success rate in a wide variety of experimental settings.

Toward Practical Adversarial Attacks on Face Verification Systems

Classifying sequential data as early as and as accurately as possible is a challenging yet critical problem, especially when a sampling cost is high. One algorithm that achieves this goal is the sequential probability ratio test (SPRT), which is known as Bayes-optimal: it can keep the expected number of data samples as small as possible, given the desired error upper-bound. The SPRT has recently been found to be the best model that explains the activities of the neurons in the primate parietal cortex that are thought to mediate our complex decision-making processes. However, the original SPRT makes two critical assumptions that limit its application in real-world scenarios: (i) samples are independently and identically distributed, and (ii) the likelihood of the data being derived from each class can be calculated precisely. Here, we propose the SPRT-TANDEM, a deep neural network-based SPRT algorithm that overcomes the above two obstacles. The SPRT-TANDEM estimates the log-likelihood ratio of two alternative hypotheses by leveraging a novel Loss function for Log-Likelihood Ratio estimation (LLLR), while allowing for correlations up to $N (\in \mathbb{N})$ preceding samples. In tests on one original and two public video databases, Nosaic MNIST, UCF101, and SiW, the SPRT-TANDEM achieves statistically significantly better classification accuracy than other baseline classifiers, with a smaller number of data samples. The code and Nosaic MNIST are publicly available at this https URL.

Deep Neural Networks for the Sequential Probability Ratio Test on Non-i.i.d. Data Series.

Classifying sequential data as early and as accurately as possible is a challenging yet critical problem, especially when a sampling cost is high One algorithm that achieves this goal is the sequential probability ratio test (SPRT), which is known as Bayes-optimal: it can keep the expected number of data samples as small as possible, given the desired error upper-bound However, the original SPRT makes two critical assumptions that limit its application in real-world scenarios: (i) samples are independently and identically distributed, and (ii) the likelihood of the data being derived from each class can be calculated precisely Here, we propose the SPRT-TANDEM, a deep neural network-based SPRT algorithm that overcomes the above two obstacles The SPRT-TANDEM sequentially estimates the log-likelihood ratio of two alternative hypotheses by leveraging a novel Loss function for Log-Likelihood Ratio estimation (LLLR) while allowing correlations up to N(∈N) preceding samples In tests on one original and two public video databases, Nosaic MNIST, UCF101, and SiW, the SPRT-TANDEM achieves statistically significantly better classification accuracy than other baseline classifiers, with a smaller number of data samples The code and Nosaic MNIST are publicly available at https://githubcom/TaikiMiyagawa/SPRT-TANDEM

/pdf/sequential-density-ratio-estimation-for-simultaneous-enj0cnwi3o.pdf

Sequential Density Ratio Estimation for Simultaneous Optimization of Speed and Accuracy

Classifying sequential data as early and as accurately as possible is a challenging yet critical problem, especially when a sampling cost is high. One algorithm that achieves this goal is the sequential probability ratio test (SPRT), which is known as Bayes-optimal: it can keep the expected number of data samples as small as possible, given the desired error upper-bound. However, the original SPRT makes two critical assumptions that limit its application in real-world scenarios: (i) samples are independently and identically distributed, and (ii) the likelihood of the data being derived from each class can be calculated precisely. Here, we propose the SPRT-TANDEM, a deep neural network-based SPRT algorithm that overcomes the above two obstacles. The SPRT-TANDEM sequentially estimates the log-likelihood ratio of two alternative hypotheses by leveraging a novel Loss function for Log-Likelihood Ratio estimation (LLLR) while allowing correlations up to $N (\in \mathbb{N})$ preceding samples. In tests on one original and two public video databases, Nosaic MNIST, UCF101, and SiW, the SPRT-TANDEM achieves statistically significantly better classification accuracy than other baseline classifiers, with a smaller number of data samples. The code and Nosaic MNIST are publicly available at this https URL

Taiki Miyagawa

Papers

Toward Practical Adversarial Attacks on Face Verification Systems

Deep Neural Networks for the Sequential Probability Ratio Test on Non-i.i.d. Data Series.

Sequential Density Ratio Estimation for Simultaneous Optimization of Speed and Accuracy

The Power of Log-Sum-Exp: Sequential Density Ratio Matrix Estimation for Speed-Accuracy Optimization

Sequential Density Ratio Estimation for Simultaneous Optimization of Speed and Accuracy.