We introduce a novel approach to feed-forward neural network interpretation based on partitioning the space of sequences of neuron activations. In line with this approach, we propose a model-specific interpretation method, called YASENN. Our method inherits many advantages of model-agnostic distillation, such as an ability to focus on the particular input region and to express an explanation in terms of features different from those observed by a neural network. Moreover, examination of distillation error makes the method applicable to the problems with low tolerance to interpretation mistakes. Technically, YASENN distills the network with an ensemble of layer-wise gradient boosting decision trees and encodes the sequences of neuron activations with leaf indices. The finite number of unique codes induces a partitioning of the input space. Each partition may be described in a variety of ways, including examination of an interpretable model (e.g. a logistic regression or a decision tree) trained to discriminate between objects of those partitions. Our experiments provide an intuition behind the method and demonstrate revealed artifacts in neural network decision making.

Theatrical improvisation (impro or improv) is a demanding form of live, collaborative performance. Improv is a humorous and playful artform built on an open-ended narrative structure which simultaneously celebrates effort and failure. It is thus an ideal test bed for the development and deployment of interactive artificial intelligence (AI)-based conversational agents, or artificial improvisors. This case study introduces an improv show experiment featuring human actors and artificial improvisors. We have previously developed a deep-learning-based artificial improvisor, trained on movie subtitles, that can generate plausible, context-based, lines of dialogue suitable for theatre. In this work, we have employed it to control what a subset of human actors say during an improv performance. We also give human-generated lines to a different subset of performers. All lines are provided to actors with headphones and all performers are wearing headphones. This paper describes a Turing test, or imitation game, taking place in a theatre, with both the audience members and the performers left to guess who is a human and who is a machine. In order to test scientific hypotheses about the perception of humans versus machines we collect anonymous feedback from volunteer performers and audience members. Our results suggest that rehearsal increases proficiency and possibility to control events in the performance. That said, consistency with real world experience is limited by the interface and the mechanisms used to perform the show. We also show that human-generated lines are shorter, more positive, and have less difficult words with more grammar and spelling mistakes than the artificial improvisor generated lines.

Abstract--The superior temporal gyrus (STG) region of cortex critically contributes to speech recognition. In this work, we show that a proposed deep network inspired by WaveNet, trained with limited available data, is able to reconstruct speech stimuli from STG intracranial recordings. We further investigate the impulse response of the fitted model for each recording electrode and observe phoneme level temporospectral tuning properties in some recorded area. This discovery is consistent with previous studies implicating the posterior STG (pSTG) in a phonetic representation of speech and provides detailed acoustic features that certain electrode sites possibly extract during speech recognition. Research studies on the superior temporal gyrus (STG) cortex area have shown that this area plays an important role in words and sentence recognition on a phonetic and prelexical stage [1]-[9].

Perceptual ad-blocking is a novel approach that uses visual cues to detect online advertisements. Compared to classical filter lists, perceptual ad-blocking is believed to be less prone to an arms race with web publishers and ad-networks. In this work we use techniques from adversarial machine learning to demonstrate that this may not be the case. We show that perceptual ad-blocking engenders a new arms race that likely disfavors ad-blockers. Unexpectedly, perceptual ad-blocking can also introduce new vulnerabilities that let an attacker bypass web security boundaries and mount DDoS attacks. We first analyze the design space of perceptual ad-blockers and present a unified architecture that incorporates prior academic and commercial work. We then explore a variety of attacks on the ad-blocker's full visual-detection pipeline, that enable publishers or ad-networks to evade or detect ad-blocking, and at times even abuse its high privilege level to bypass web security boundaries. Our attacks exploit the unreasonably strong threat model that perceptual ad-blockers must survive. Finally, we evaluate a concrete set of attacks on an ad-blocker's internal ad-classifier by instantiating adversarial examples for visual systems in a real web-security context. For six ad-detection techniques, we create perturbed ads, ad-disclosures, and native web content that misleads perceptual ad-blocking with 100% success rates. For example, we demonstrate how a malicious user can upload adversarial content (e.g., a perturbed image in a Facebook post) that fools the ad-blocker into removing other users' non-ad content.

The performance of deep neural networks crucially depends on good hyperparameter configurations. Bayesian optimization is a powerful framework for optimizing the hyperparameters of DNNs. These methods need sufficient evaluation data to approximate and minimize the validation error function of hyperparameters. However, the expensive evaluation cost of DNNs leads to very few evaluation data within a limited time, which greatly reduces the efficiency of Bayesian optimization. Besides, the previous researches focus on using the complete evaluation data to conduct Bayesian optimization, and ignore the intermediate evaluation data generated by early stopping methods. To alleviate the insufficient evaluation data problem, we propose a fast hyperparameter optimization method, HOIST, that utilizes both the complete and intermediate evaluation data to accelerate the hyperparameter optimization of DNNs. Specifically, we train multiple basic surrogates to gather information from the mixed evaluation data, and then combine these basic surrogates using weighted bagging to provide an accurate ensemble surrogate. Our empirical studies show that HOIST outperforms the state-of-theart approaches on a wide range of DNNs, including feed forward neural networks, convolutional neural networks, recurrent neural networks, and variational autoencoder.

This paper describes a novel approach of packing sparse convolutional neural networks for their efficient systolic array implementations. By combining subsets of columns in the original filter matrix associated with a convolutional layer, we increase the utilization efficiency of the systolic array substantially (e.g., ~4x) due to the increased density of nonzeros in the resulting packed filter matrix. In combining columns, for each row, all filter weights but one with the largest magnitude are pruned. We retrain the remaining weights to preserve high accuracy. We demonstrate that in mitigating data privacy concerns the retraining can be accomplished with only fractions of the original dataset (e.g., 10\% for CIFAR-10). We study the effectiveness of this joint optimization for both high utilization and classification accuracy with ASIC and FPGA designs based on efficient bit-serial implementations of multiplier-accumulators. We present analysis and empirical evidence on the superior performance of our column combining approach against prior arts under metrics such as energy efficiency (3x) and inference latency (12x).

Understanding and interpreting how machine learning (ML) models make decisions have been a big challenge. While recent research has proposed various technical approaches to provide some clues as to how an ML model makes individual predictions, they cannot provide users with an ability to inspect a model as a complete entity. In this work, we propose a novel technical approach that augments a Bayesian non-parametric regression mixture model with multiple elastic nets. Using the enhanced mixture model, we can extract generalizable insights for a target model through a global approximation. To demonstrate the utility of our approach, we evaluate it on different ML models in the context of image recognition. The empirical results indicate that our proposed approach not only outperforms the state-of-the-art techniques in explaining individual decisions but also provides users with an ability to discover the vulnerabilities of the target ML models.

In high dimensional settings, density estimation algorithms rely crucially on their inductive bias. Despite recent empirical success, the inductive bias of deep generative models is not well understood. In this paper we propose a framework to systematically investigate bias and generalization in deep generative models of images. Inspired by experimental methods from cognitive psychology, we probe each learning algorithm with carefully designed training datasets to characterize when and how existing models generate novel attributes and their combinations. We identify similarities to human psychology and verify that these patterns are consistent across commonly used models and architectures.

An activation boundary for a neuron refers to a separating hyperplane that determines whether the neuron is activated or deactivated. It has been long considered in neural networks that the activations of neurons, rather than their exact output values, play the most important role in forming classification friendly partitions of the hidden feature space. However, as far as we know, this aspect of neural networks has not been considered in the literature of knowledge transfer. In this paper, we propose a knowledge transfer method via distillation of activation boundaries formed by hidden neurons. For the distillation, we propose an activation transfer loss that has the minimum value when the boundaries generated by the student coincide with those by the teacher. Since the activation transfer loss is not differentiable, we design a piecewise differentiable loss approximating the activation transfer loss. By the proposed method, the student learns a separating boundary between activation region and deactivation region formed by each neuron in the teacher. Through the experiments in various aspects of knowledge transfer, it is verified that the proposed method outperforms the current state-of-the-art.

In this paper, we confront the problem of deep learning's big labeled data requirements, offer a rule based strategy for extreme augmentation of small data sets and apply that strategy with the image to image translation model by Isola et al. (2016) to automate cel style cartoon coloring with very limited training data. While our experimental results using geometric rules and transformations demonstrate the performance of our methods on an image translation task with industry applications in art, design and animation, we also propose the use of rules on partial data sets as a generalizable small data strategy, potentially applicable across data types and domains.