January 27, 2020

3315 words 16 mins read

Paper Group ANR 1124

Stealing Knowledge from Protected Deep Neural Networks Using Composite Unlabeled Data. A Human-in-the-loop Framework to Construct Context-dependent Mathematical Formulations of Fairness. Learning Probabilities: Towards a Logic of Statistical Learning. Adversarial Training Can Hurt Generalization. BenchCouncil’s View on Benchmarking AI and Other Eme …

Stealing Knowledge from Protected Deep Neural Networks Using Composite Unlabeled Data


Title	Stealing Knowledge from Protected Deep Neural Networks Using Composite Unlabeled Data
Authors	Itay Mosafi, Eli David, Nathan S. Netanyahu
Abstract	As state-of-the-art deep neural networks are deployed at the core of more advanced Al-based products and services, the incentive for copying them (i.e., their intellectual properties) by rival adversaries is expected to increase considerably over time. The best way to extract or steal knowledge from such networks is by querying them using a large dataset of random samples and recording their output, followed by training a student network to mimic these outputs, without making any assumption about the original networks. The most effective way to protect against such a mimicking attack is to provide only the classification result, without confidence values associated with the softmax layer.In this paper, we present a novel method for generating composite images for attacking a mentor neural network using a student model. Our method assumes no information regarding the mentor’s training dataset, architecture, or weights. Further assuming no information regarding the mentor’s softmax output values, our method successfully mimics the given neural network and steals all of its knowledge. We also demonstrate that our student network (which copies the mentor) is impervious to watermarking protection methods, and thus would not be detected as a stolen model.Our results imply, essentially, that all current neural networks are vulnerable to mimicking attacks, even if they do not divulge anything but the most basic required output, and that the student model which mimics them cannot be easily detected and singled out as a stolen copy using currently available techniques.
Tasks
Published	2019-12-09
URL	https://arxiv.org/abs/1912.03959v1
PDF	https://arxiv.org/pdf/1912.03959v1.pdf
PWC	https://paperswithcode.com/paper/stealing-knowledge-from-protected-deep-neural
Repo
Framework

A Human-in-the-loop Framework to Construct Context-dependent Mathematical Formulations of Fairness


Title	A Human-in-the-loop Framework to Construct Context-dependent Mathematical Formulations of Fairness
Authors	Mohammad Yaghini, Hoda Heidari, Andreas Krause
Abstract	Despite the recent surge of interest in designing and guaranteeing mathematical formulations of fairness, virtually all existing notions of algorithmic fairness fail to be adaptable to the intricacies and nuances of the decision-making context at hand. We argue that capturing such factors is an inherently human task, as it requires knowledge of the social background in which machine learning tools impact real people’s outcomes and a deep understanding of the ramifications of automated decisions for decision subjects and society. In this work, we present a framework to construct a context-dependent mathematical formulation of fairness utilizing people’s judgment of fairness. We utilize the theoretical model of Heidari et al. (2019)—which shows that most existing formulations of algorithmic fairness are special cases of economic models of Equality of Opportunity (EOP)—and present a practical human-in-the-loop approach to pinpoint the fairness notion in the EOP family that best captures people’s perception of fairness in the given context. To illustrate our framework, we run human-subject experiments designed to learn the parameters of Heidari et al.‘s EOP model (including circumstance, desert, and utility) in a hypothetical recidivism decision-making scenario. Our work takes an initial step toward democratizing the formulation of fairness and utilizing human-judgment to tackle a fundamental shortcoming of automated decision-making systems: that the machine on its own is incapable of understanding and processing the human aspects and social context of its decisions.
Tasks	Decision Making
Published	2019-11-08
URL	https://arxiv.org/abs/1911.03020v1
PDF	https://arxiv.org/pdf/1911.03020v1.pdf
PWC	https://paperswithcode.com/paper/a-human-in-the-loop-framework-to-construct
Repo
Framework

Learning Probabilities: Towards a Logic of Statistical Learning


Title	Learning Probabilities: Towards a Logic of Statistical Learning
Authors	Alexandru Baltag, Soroush Rafiee Rad, Sonja Smets
Abstract	We propose a new model for forming beliefs and learning about unknown probabilities (such as the probability of picking a red marble from a bag with an unknown distribution of coloured marbles). The most widespread model for such situations of ‘radical uncertainty’ is in terms of imprecise probabilities, i.e. representing the agent’s knowledge as a set of probability measures. We add to this model a plausibility map, associating to each measure a plausibility number, as a way to go beyond what is known with certainty and represent the agent’s beliefs about probability. There are a number of standard examples: Shannon Entropy, Centre of Mass etc. We then consider learning of two types of information: (1) learning by repeated sampling from the unknown distribution (e.g. picking marbles from the bag); and (2) learning higher-order information about the distribution (in the shape of linear inequalities, e.g. we are told there are more red marbles than green marbles). The first changes only the plausibility map (via a ‘plausibilistic’ version of Bayes’ Rule), but leaves the given set of measures unchanged; the second shrinks the set of measures, without changing their plausibility. Beliefs are defined as in Belief Revision Theory, in terms of truth in the most plausible worlds. But our belief change does not comply with standard AGM axioms, since the revision induced by (1) is of a non-AGM type. This is essential, as it allows our agents to learn the true probability: we prove that the beliefs obtained by repeated sampling converge almost surely to the correct belief (in the true probability). We end by sketching the contours of a dynamic doxastic logic for statistical learning.
Tasks
Published	2019-07-22
URL	https://arxiv.org/abs/1907.09472v1
PDF	https://arxiv.org/pdf/1907.09472v1.pdf
PWC	https://paperswithcode.com/paper/learning-probabilities-towards-a-logic-of
Repo
Framework

Adversarial Training Can Hurt Generalization


Title	Adversarial Training Can Hurt Generalization
Authors	Aditi Raghunathan, Sang Michael Xie, Fanny Yang, John C. Duchi, Percy Liang
Abstract	While adversarial training can improve robust accuracy (against an adversary), it sometimes hurts standard accuracy (when there is no adversary). Previous work has studied this tradeoff between standard and robust accuracy, but only in the setting where no predictor performs well on both objectives in the infinite data limit. In this paper, we show that even when the optimal predictor with infinite data performs well on both objectives, a tradeoff can still manifest itself with finite data. Furthermore, since our construction is based on a convex learning problem, we rule out optimization concerns, thus laying bare a fundamental tension between robustness and generalization. Finally, we show that robust self-training mostly eliminates this tradeoff by leveraging unlabeled data.
Tasks
Published	2019-06-14
URL	https://arxiv.org/abs/1906.06032v2
PDF	https://arxiv.org/pdf/1906.06032v2.pdf
PWC	https://paperswithcode.com/paper/adversarial-training-can-hurt-generalization
Repo
Framework

BenchCouncil’s View on Benchmarking AI and Other Emerging Workloads


Title	BenchCouncil’s View on Benchmarking AI and Other Emerging Workloads
Authors	Jianfeng Zhan, Lei Wang, Wanling Gao, Rui Ren
Abstract	This paper outlines BenchCouncil’s view on the challenges, rules, and vision of benchmarking modern workloads like Big Data, AI or machine learning, and Internet Services. We conclude the challenges of benchmarking modern workloads as FIDSS (Fragmented, Isolated, Dynamic, Service-based, and Stochastic), and propose the PRDAERS benchmarking rules that the benchmarks should be specified in a paper-and-pencil manner, relevant, diverse, containing different levels of abstractions, specifying the evaluation metrics and methodology, repeatable, and scaleable. We believe proposing simple but elegant abstractions that help achieve both efficiency and general-purpose is the final target of benchmarking in future, which may be not pressing. In the light of this vision, we shortly discuss BenchCouncil’s related projects.
Tasks
Published	2019-12-02
URL	https://arxiv.org/abs/1912.00572v2
PDF	https://arxiv.org/pdf/1912.00572v2.pdf
PWC	https://paperswithcode.com/paper/benchcouncils-view-on-benchmarking-ai-and
Repo
Framework

AugLabel: Exploiting Word Representations to Augment Labels for Face Attribute Classification


Title	AugLabel: Exploiting Word Representations to Augment Labels for Face Attribute Classification
Authors	Binod Bhattarai, Rumeysa Bodur, Tae-Kyun Kim
Abstract	Augmenting data in image space (eg. flipping, cropping etc) and activation space (eg. dropout) are being widely used to regularise deep neural networks and have been successfully applied on several computer vision tasks. Unlike previous works, which are mostly focused on doing augmentation in the aforementioned domains, we propose to do augmentation in label space. In this paper, we present a novel method to generate fixed dimensional labels with continuous values for images by exploiting the word2vec representations of the existing categorical labels. We then append these representations with existing categorical labels and train the model. We validated our idea on two challenging face attribute classification data sets viz. CelebA and LFWA. Our extensive experiments show that the augmented labels improve the performance of the competitive deep learning baseline and reduce the need of annotated real data up to 50%, while attaining a performance similar to the state-of-the-art methods.
Tasks
Published	2019-07-15
URL	https://arxiv.org/abs/1907.06757v1
PDF	https://arxiv.org/pdf/1907.06757v1.pdf
PWC	https://paperswithcode.com/paper/auglabel-exploiting-word-representations-to
Repo
Framework

Rethinking the Discount Factor in Reinforcement Learning: A Decision Theoretic Approach


Title	Rethinking the Discount Factor in Reinforcement Learning: A Decision Theoretic Approach
Authors	Silviu Pitis
Abstract	Reinforcement learning (RL) agents have traditionally been tasked with maximizing the value function of a Markov decision process (MDP), either in continuous settings, with fixed discount factor $\gamma < 1$, or in episodic settings, with $\gamma = 1$. While this has proven effective for specific tasks with well-defined objectives (e.g., games), it has never been established that fixed discounting is suitable for general purpose use (e.g., as a model of human preferences). This paper characterizes rationality in sequential decision making using a set of seven axioms and arrives at a form of discounting that generalizes traditional fixed discounting. In particular, our framework admits a state-action dependent “discount” factor that is not constrained to be less than 1, so long as there is eventual long run discounting. Although this broadens the range of possible preference structures in continuous settings, we show that there exists a unique “optimizing MDP” with fixed $\gamma < 1$ whose optimal value function matches the true utility of the optimal policy, and we quantify the difference between value and utility for suboptimal policies. Our work can be seen as providing a normative justification for (a slight generalization of) Martha White’s RL task formalism (2017) and other recent departures from the traditional RL, and is relevant to task specification in RL, inverse RL and preference-based RL.
Tasks	Decision Making
Published	2019-02-08
URL	http://arxiv.org/abs/1902.02893v1
PDF	http://arxiv.org/pdf/1902.02893v1.pdf
PWC	https://paperswithcode.com/paper/rethinking-the-discount-factor-in
Repo
Framework

Orthogonality Constrained Multi-Head Attention For Keyword Spotting


Title	Orthogonality Constrained Multi-Head Attention For Keyword Spotting
Authors	Mingu Lee, Jinkyu Lee, Hye Jin Jang, Byeonggeun Kim, Wonil Chang, Kyuwoong Hwang
Abstract	Multi-head attention mechanism is capable of learning various representations from sequential data while paying attention to different subsequences, e.g., word-pieces or syllables in a spoken word. From the subsequences, it retrieves richer information than a single-head attention which only summarizes the whole sequence into one context vector. However, a naive use of the multi-head attention does not guarantee such richness as the attention heads may have positional and representational redundancy. In this paper, we propose a regularization technique for multi-head attention mechanism in an end-to-end neural keyword spotting system. Augmenting regularization terms which penalize positional and contextual non-orthogonality between the attention heads encourages to output different representations from separate subsequences, which in turn enables leveraging structured information without explicit sequence models such as hidden Markov models. In addition, intra-head contextual non-orthogonality regularization encourages each attention head to have similar representations across keyword examples, which helps classification by reducing feature variability. The experimental results demonstrate that the proposed regularization technique significantly improves the keyword spotting performance for the keyword “Hey Snapdragon”.
Tasks	Keyword Spotting
Published	2019-10-10
URL	https://arxiv.org/abs/1910.04500v1
PDF	https://arxiv.org/pdf/1910.04500v1.pdf
PWC	https://paperswithcode.com/paper/orthogonality-constrained-multi-head
Repo
Framework

ExpertMatcher: Automating ML Model Selection for Users in Resource Constrained Countries


Title	ExpertMatcher: Automating ML Model Selection for Users in Resource Constrained Countries
Authors	Vivek Sharma, Praneeth Vepakomma, Tristan Swedish, Ken Chang, Jayashree Kalpathy-Cramer, Ramesh Raskar
Abstract	In this work we introduce ExpertMatcher, a method for automating deep learning model selection using autoencoders. Specifically, we are interested in performing inference on data sources that are distributed across many clients using pretrained expert ML networks on a centralized server. The ExpertMatcher assigns the most relevant model(s) in the central server given the client’s data representation. This allows resource-constrained clients in developing countries to utilize the most relevant ML models for their given task without having to evaluate the performance of each ML model. The method is generic and can be beneficial in any setup where there are local clients and numerous centralized expert ML models.
Tasks	Model Selection
Published	2019-10-05
URL	https://arxiv.org/abs/1910.02312v1
PDF	https://arxiv.org/pdf/1910.02312v1.pdf
PWC	https://paperswithcode.com/paper/expertmatcher-automating-ml-model-selection
Repo
Framework

Variational Inference for Computational Imaging Inverse Problems


Title	Variational Inference for Computational Imaging Inverse Problems
Authors	Francesco Tonolini, Jack Radford, Alex Turpin, Daniele Faccio, Roderick Murray-Smith
Abstract	Machine learning methods for computational imaging require uncertainty estimation to be reliable in real settings. While Bayesian models offer a computationally tractable way of recovering uncertainty, they need large data volumes to be trained, which in imaging applications implicates prohibitively expensive collections with specific imaging instruments. This paper introduces a novel framework to train variational inference for inverse problems exploiting in combination few experimentally collected data, domain expertise and existing image data sets. In such a way, Bayesian machine learning models can solve imaging inverse problems with minimal data collection efforts. Extensive simulated experiments show the advantages of the proposed framework. The approach is then applied to two real experimental optics settings: holographic image reconstruction and imaging through highly scattering media. In both settings, state of the art reconstructions are achieved with little collection of training data.
Tasks	Image Reconstruction
Published	2019-04-12
URL	https://arxiv.org/abs/1904.06264v2
PDF	https://arxiv.org/pdf/1904.06264v2.pdf
PWC	https://paperswithcode.com/paper/variational-inference-for-computational
Repo
Framework

Fast Panoptic Segmentation Network


Title	Fast Panoptic Segmentation Network
Authors	Daan de Geus, Panagiotis Meletis, Gijs Dubbelman
Abstract	In this work, we present an end-to-end network for fast panoptic segmentation. This network, called Fast Panoptic Segmentation Network (FPSNet), does not require computationally costly instance mask predictions or merging heuristics. This is achieved by casting the panoptic task into a custom dense pixel-wise classification task, which assigns a class label or an instance id to each pixel. We evaluate FPSNet on the Cityscapes and Pascal VOC datasets, and find that FPSNet is faster than existing panoptic segmentation methods, while achieving better or similar panoptic segmentation performance. On the Cityscapes validation set, we achieve a Panoptic Quality score of 55.1%, at prediction times of 114 milliseconds for images with a resolution of 1024x2048 pixels. For lower resolutions of the Cityscapes dataset and for the Pascal VOC dataset, FPSNet runs at 22 and 35 frames per second, respectively.
Tasks	Panoptic Segmentation
Published	2019-10-09
URL	https://arxiv.org/abs/1910.03892v1
PDF	https://arxiv.org/pdf/1910.03892v1.pdf
PWC	https://paperswithcode.com/paper/fast-panoptic-segmentation-network
Repo
Framework

Augmenting Transfer Learning with Semantic Reasoning


Title	Augmenting Transfer Learning with Semantic Reasoning
Authors	Freddy Lecue, Jiaoyan Chen, Jeff Z. Pan, Huajun Chen
Abstract	Transfer learning aims at building robust prediction models by transferring knowledge gained from one problem to another. In the semantic Web, learning tasks are enhanced with semantic representations. We exploit their semantics to augment transfer learning by dealing with when to transfer with semantic measurements and what to transfer with semantic embeddings. We further present a general framework that integrates the above measurements and embeddings with existing transfer learning algorithms for higher performance. It has demonstrated to be robust in two real-world applications: bus delay forecasting and air quality forecasting.
Tasks	Transfer Learning
Published	2019-05-31
URL	https://arxiv.org/abs/1905.13672v2
PDF	https://arxiv.org/pdf/1905.13672v2.pdf
PWC	https://paperswithcode.com/paper/augmenting-transfer-learning-with-semantic
Repo
Framework

RELINE: Point-of-Interest Recommendations using Multiple Network Embeddings


Title	RELINE: Point-of-Interest Recommendations using Multiple Network Embeddings
Authors	Giannis Christoforidis, Pavlos Kefalas, Apostolos N. Papadopoulos, Yannis Manolopoulos
Abstract	The rapid growth of users’ involvement in Location-Based Social Networks (LBSNs) has led to the expeditious growth of the data on a global scale. The need of accessing and retrieving relevant information close to users’ preferences is an open problem which continuously raises new challenges for recommendation systems. The exploitation of Points-of-Interest (POIs) recommendation by existing models is inadequate due to the sparsity and the cold start problems. To overcome these problems many models were proposed in the literature, but most of them ignore important factors such as: geographical proximity, social influence, or temporal and preference dynamics, which tackle their accuracy while personalize their recommendations. In this work, we investigate these problems and present a unified model that jointly learns users and POI dynamics. Our proposal is termed RELINE (REcommendations with muLtIple Network Embeddings). More specifically, RELINE captures: i) the social, ii) the geographical, iii) the temporal influence, and iv) the users’ preference dynamics, by embedding eight relational graphs into one shared latent space. We have evaluated our approach against state-of-the-art methods with three large real-world datasets in terms of accuracy. Additionally, we have examined the effectiveness of our approach against the cold-start problem. Performance evaluation results demonstrate that significant performance improvement is achieved in comparison to existing state-of-the-art methods.
Tasks	Recommendation Systems
Published	2019-02-02
URL	http://arxiv.org/abs/1902.00773v1
PDF	http://arxiv.org/pdf/1902.00773v1.pdf
PWC	https://paperswithcode.com/paper/reline-point-of-interest-recommendations
Repo
Framework

Learning Spatio-Temporal Features with Two-Stream Deep 3D CNNs for Lipreading


Title	Learning Spatio-Temporal Features with Two-Stream Deep 3D CNNs for Lipreading
Authors	Xinshuo Weng, Kris Kitani
Abstract	We focus on the word-level visual lipreading, which requires recognizing the word being spoken, given only the video but not the audio. State-of-the-art methods explore the use of end-to-end neural networks, including a shallow (up to three layers) 3D convolutional neural network (CNN) + a deep 2D CNN (e.g., ResNet) as the front-end to extract visual features, and a recurrent neural network (e.g., bidirectional LSTM) as the back-end for classification. In this work, we propose to replace the shallow 3D CNNs + deep 2D CNNs front-end with recent successful deep 3D CNNs — two-stream (i.e., grayscale video and optical flow streams) I3D. We evaluate different combinations of front-end and back-end modules with the grayscale video and optical flow inputs on the LRW dataset. The experiments show that, compared to the shallow 3D CNNs + deep 2D CNNs front-end, the deep 3D CNNs front-end with pre-training on the large-scale image and video datasets (e.g., ImageNet and Kinetics) can improve the classification accuracy. Also, we demonstrate that using the optical flow input alone can achieve comparable performance as using the grayscale video as input. Moreover, the two-stream network using both the grayscale video and optical flow inputs can further improve the performance. Overall, our two-stream I3D front-end with a Bi-LSTM back-end results in an absolute improvement of 5.3% over the previous art on the LRW dataset.
Tasks	Lipreading, Optical Flow Estimation
Published	2019-05-04
URL	https://arxiv.org/abs/1905.02540v2
PDF	https://arxiv.org/pdf/1905.02540v2.pdf
PWC	https://paperswithcode.com/paper/learning-spatio-temporal-features-with-two
Repo
Framework

A Novel Approach for Automatic Bengali Question Answering System using Semantic Similarity Analysis


Title	A Novel Approach for Automatic Bengali Question Answering System using Semantic Similarity Analysis
Authors	Arijit Das, Jaydeep Mandal, Zargham Danial, Alok Ranjan Pal, Diganta Saha
Abstract	Finding the semantically accurate answer is one of the key challenges in advanced searching. In contrast to keyword-based searching, the meaning of a question or query is important here and answers are ranked according to relevance. It is very natural that there is almost no common word between the question sentence and the answer sentence. In this paper, an approach is described to find out the semantically relevant answers in the Bengali dataset. In the first part of the algorithm, a set of statistical parameters like frequency, index, part-of-speech (POS), etc. is matched between a question and the probable answers. In the second phase, entropy and similarity are calculated in different modules. Finally, a sense score is generated to rank the answers. The algorithm is tested on a repository containing a total of 275000 sentences. This Bengali repository is a product of Technology Development for Indian Languages (TDIL) project sponsored by Govt. of India and provided by the Language Research Unit of Indian Statistical Institute, Kolkata. The shallow parser, developed by the LTRC group of IIIT Hyderabad is used for POS tagging. The actual answer is ranked as 1st in 82.3% cases. The actual answer is ranked within 1st to 5th in 90.0% cases. The accuracy of the system is coming as 97.32% and precision of the system is coming as 98.14% using confusion matrix. The challenges and pitfalls of the work are reported at last in this paper.
Tasks	Question Answering, Semantic Similarity, Semantic Textual Similarity
Published	2019-10-23
URL	https://arxiv.org/abs/1910.10758v1
PDF	https://arxiv.org/pdf/1910.10758v1.pdf
PWC	https://paperswithcode.com/paper/a-novel-approach-for-automatic-bengali
Repo
Framework