May 7, 2019

3287 words 16 mins read

Paper Group ANR 135

The “Horse’’ Inside: Seeking Causes Behind the Behaviours of Music Content Analysis Systems. Development of a Fuzzy Expert System based Liveliness Detection Scheme for Biometric Authentication. Stacking for machine learning redshifts applied to SDSS galaxies. Speed-Constrained Tuning for Statistical Machine Translation Using Bayesian Optimization. …

The “Horse’’ Inside: Seeking Causes Behind the Behaviours of Music Content Analysis Systems


Title	The “Horse’’ Inside: Seeking Causes Behind the Behaviours of Music Content Analysis Systems
Authors	Bob L. Sturm
Abstract	Building systems that possess the sensitivity and intelligence to identify and describe high-level attributes in music audio signals continues to be an elusive goal, but one that surely has broad and deep implications for a wide variety of applications. Hundreds of papers have so far been published toward this goal, and great progress appears to have been made. Some systems produce remarkable accuracies at recognising high-level semantic concepts, such as music style, genre and mood. However, it might be that these numbers do not mean what they seem. In this paper, we take a state-of-the-art music content analysis system and investigate what causes it to achieve exceptionally high performance in a benchmark music audio dataset. We dissect the system to understand its operation, determine its sensitivities and limitations, and predict the kinds of knowledge it could and could not possess about music. We perform a series of experiments to illuminate what the system has actually learned to do, and to what extent it is performing the intended music listening task. Our results demonstrate how the initial manifestation of music intelligence in this state-of-the-art can be deceptive. Our work provides constructive directions toward developing music content analysis systems that can address the music information and creation needs of real-world users.
Tasks
Published	2016-06-09
URL	http://arxiv.org/abs/1606.03044v1
PDF	http://arxiv.org/pdf/1606.03044v1.pdf
PWC	https://paperswithcode.com/paper/the-horse-inside-seeking-causes-behind-the
Repo
Framework

Development of a Fuzzy Expert System based Liveliness Detection Scheme for Biometric Authentication


Title	Development of a Fuzzy Expert System based Liveliness Detection Scheme for Biometric Authentication
Authors	Avinash Kumar Singh, Piyush Joshi, G C Nandi
Abstract	Liveliness detection acts as a safe guard against spoofing attacks. Most of the researchers used vision based techniques to detect liveliness of the user, but they are highly sensitive to illumination effects. Therefore it is very hard to design a system, which will work robustly under all circumstances. Literature shows that most of the research utilize eye blink or mouth movement to detect the liveliness, while the other group used face texture to distinguish between real and imposter. The classification results of all these approaches decreases drastically in variable light conditions. Hence in this paper we are introducing fuzzy expert system which is sufficient enough to handle most of the cases comes in real time. We have used two testing parameters, (a) under bad illumination and (b) less movement in eyes and mouth in case of real user to evaluate the performance of the system. The system is behaving well in all, while in first case its False Rejection Rate (FRR) is 0.28, and in second case its FRR is 0.4.
Tasks
Published	2016-09-17
URL	http://arxiv.org/abs/1609.05296v1
PDF	http://arxiv.org/pdf/1609.05296v1.pdf
PWC	https://paperswithcode.com/paper/development-of-a-fuzzy-expert-system-based
Repo
Framework

Stacking for machine learning redshifts applied to SDSS galaxies


Title	Stacking for machine learning redshifts applied to SDSS galaxies
Authors	Roman Zitlau, Ben Hoyle, Kerstin Paech, Jochen Weller, Markus Michael Rau, Stella Seitz
Abstract	We present an analysis of a general machine learning technique called ‘stacking’ for the estimation of photometric redshifts. Stacking techniques can feed the photometric redshift estimate, as output by a base algorithm, back into the same algorithm as an additional input feature in a subsequent learning round. We shown how all tested base algorithms benefit from at least one additional stacking round (or layer). To demonstrate the benefit of stacking, we apply the method to both unsupervised machine learning techniques based on self-organising maps (SOMs), and supervised machine learning methods based on decision trees. We explore a range of stacking architectures, such as the number of layers and the number of base learners per layer. Finally we explore the effectiveness of stacking even when using a successful algorithm such as AdaBoost. We observe a significant improvement of between 1.9% and 21% on all computed metrics when stacking is applied to weak learners (such as SOMs and decision trees). When applied to strong learning algorithms (such as AdaBoost) the ratio of improvement shrinks, but still remains positive and is between 0.4% and 2.5% for the explored metrics and comes at almost no additional computational cost.
Tasks
Published	2016-02-19
URL	http://arxiv.org/abs/1602.06294v2
PDF	http://arxiv.org/pdf/1602.06294v2.pdf
PWC	https://paperswithcode.com/paper/stacking-for-machine-learning-redshifts
Repo
Framework

Speed-Constrained Tuning for Statistical Machine Translation Using Bayesian Optimization


Title	Speed-Constrained Tuning for Statistical Machine Translation Using Bayesian Optimization
Authors	Daniel Beck, Adrià de Gispert, Gonzalo Iglesias, Aurelien Waite, Bill Byrne
Abstract	We address the problem of automatically finding the parameters of a statistical machine translation system that maximize BLEU scores while ensuring that decoding speed exceeds a minimum value. We propose the use of Bayesian Optimization to efficiently tune the speed-related decoding parameters by easily incorporating speed as a noisy constraint function. The obtained parameter values are guaranteed to satisfy the speed constraint with an associated confidence margin. Across three language pairs and two speed constraint values, we report overall optimization time reduction compared to grid and random search. We also show that Bayesian Optimization can decouple speed and BLEU measurements, resulting in a further reduction of overall optimization time as speed is measured over a small subset of sentences.
Tasks	Machine Translation
Published	2016-04-18
URL	http://arxiv.org/abs/1604.05073v1
PDF	http://arxiv.org/pdf/1604.05073v1.pdf
PWC	https://paperswithcode.com/paper/speed-constrained-tuning-for-statistical
Repo
Framework

Interactive Debugging of Knowledge Bases


Title	Interactive Debugging of Knowledge Bases
Authors	Patrick Rodler
Abstract	Many AI applications rely on knowledge about a relevant real-world domain that is encoded by means of some logical knowledge base (KB). The most essential benefit of logical KBs is the opportunity to perform automatic reasoning to derive implicit knowledge or to answer complex queries about the modeled domain. The feasibility of meaningful reasoning requires KBs to meet some minimal quality criteria such as logical consistency. Without adequate tool assistance, the task of resolving violated quality criteria in KBs can be extremely tough even for domain experts, especially when the problematic KB includes a large number of logical formulas or comprises complicated logical formalisms. Published non-interactive debugging systems often cannot localize all possible faults (incompleteness), suggest the deletion or modification of unnecessarily large parts of the KB (non-minimality), return incorrect solutions which lead to a repaired KB not satisfying the imposed quality requirements (unsoundness) or suffer from poor scalability due to the inherent complexity of the KB debugging problem. Even if a system is complete and sound and considers only minimal solutions, there are generally exponentially many solution candidates to select one from. However, any two repaired KBs obtained from these candidates differ in their semantics in terms of entailments and non-entailments. Selection of just any of these repaired KBs might result in unexpected entailments, the loss of desired entailments or unwanted changes to the KB. This work proposes complete, sound and optimal methods for the interactive debugging of KBs that suggest the one (minimally invasive) error correction of the faulty KB that yields a repaired KB with exactly the intended semantics. Users, e.g. domain experts, are involved in the debugging process by answering automatically generated queries about the intended domain.
Tasks
Published	2016-05-19
URL	http://arxiv.org/abs/1605.05950v1
PDF	http://arxiv.org/pdf/1605.05950v1.pdf
PWC	https://paperswithcode.com/paper/interactive-debugging-of-knowledge-bases
Repo
Framework

Three Tiers Neighborhood Graph and Multi-graph Fusion Ranking for Multi-feature Image Retrieval: A Manifold Aspect


Title	Three Tiers Neighborhood Graph and Multi-graph Fusion Ranking for Multi-feature Image Retrieval: A Manifold Aspect
Authors	Shenglan Liu, Muxin Sun, Lin Feng, Yang Liu, Jun Wu
Abstract	Single feature is inefficient to describe content of an image, which is a shortcoming in traditional image retrieval task. We know that one image can be described by different features. Multi-feature fusion ranking can be utilized to improve the ranking list of query. In this paper, we first analyze graph structure and multi-feature fusion re-ranking from manifold aspect. Then, Three Tiers Neighborhood Graph (TTNG) is constructed to re-rank the original ranking list by single feature and to enhance precision of single feature. Furthermore, we propose Multi-graph Fusion Ranking (MFR) for multi-feature ranking, which considers the correlation of all images in multiple neighborhood graphs. Evaluations are conducted on UK-bench, Corel-1K, Corel-10K and Cifar-10 benchmark datasets. The experimental results show that our TTNG and MFR outperform than other state-of-the-art methods. For example, we achieve competitive results N-S score 3.91 and precision 65.00% on UK-bench and Corel-10K datasets respectively.
Tasks	Image Retrieval
Published	2016-09-24
URL	http://arxiv.org/abs/1609.07599v1
PDF	http://arxiv.org/pdf/1609.07599v1.pdf
PWC	https://paperswithcode.com/paper/three-tiers-neighborhood-graph-and-multi
Repo
Framework

Lattice-Based Recurrent Neural Network Encoders for Neural Machine Translation


Title	Lattice-Based Recurrent Neural Network Encoders for Neural Machine Translation
Authors	Jinsong Su, Zhixing Tan, Deyi Xiong, Rongrong Ji, Xiaodong Shi, Yang Liu
Abstract	Neural machine translation (NMT) heavily relies on word-level modelling to learn semantic representations of input sentences. However, for languages without natural word delimiters (e.g., Chinese) where input sentences have to be tokenized first, conventional NMT is confronted with two issues: 1) it is difficult to find an optimal tokenization granularity for source sentence modelling, and 2) errors in 1-best tokenizations may propagate to the encoder of NMT. To handle these issues, we propose word-lattice based Recurrent Neural Network (RNN) encoders for NMT, which generalize the standard RNN to word lattice topology. The proposed encoders take as input a word lattice that compactly encodes multiple tokenizations, and learn to generate new hidden states from arbitrarily many inputs and hidden states in preceding time steps. As such, the word-lattice based encoders not only alleviate the negative impact of tokenization errors but also are more expressive and flexible to embed input sentences. Experiment results on Chinese-English translation demonstrate the superiorities of the proposed encoders over the conventional encoder.
Tasks	Machine Translation, Tokenization
Published	2016-09-25
URL	http://arxiv.org/abs/1609.07730v2
PDF	http://arxiv.org/pdf/1609.07730v2.pdf
PWC	https://paperswithcode.com/paper/lattice-based-recurrent-neural-network
Repo
Framework

A Metaprogramming and Autotuning Framework for Deploying Deep Learning Applications


Title	A Metaprogramming and Autotuning Framework for Deploying Deep Learning Applications
Authors	Matthew W. Moskewicz, Ali Jannesari, Kurt Keutzer
Abstract	In recent years, deep neural networks (DNNs), have yielded strong results on a wide range of applications. Graphics Processing Units (GPUs) have been one key enabling factor leading to the current popularity of DNNs. However, despite increasing hardware flexibility and software programming toolchain maturity, high efficiency GPU programming remains difficult: it suffers from high complexity, low productivity, and low portability. GPU vendors such as NVIDIA have spent enormous effort to write special-purpose DNN libraries. However, on other hardware targets, especially mobile GPUs, such vendor libraries are not generally available. Thus, the development of portable, open, high-performance, energy-efficient GPU code for DNN operations would enable broader deployment of DNN-based algorithms. Toward this end, this work presents a framework to enable productive, high-efficiency GPU programming for DNN computations across hardware platforms and programming models. In particular, the framework provides specific support for metaprogramming, autotuning, and DNN-tailored data types. Using our framework, we explore implementing DNN operations on three different hardware targets: NVIDIA, AMD, and Qualcomm GPUs. On NVIDIA GPUs, we show both portability between OpenCL and CUDA as well competitive performance compared to the vendor library. On Qualcomm GPUs, we show that our framework enables productive development of target-specific optimizations, and achieves reasonable absolute performance. Finally, On AMD GPUs, we show initial results that indicate our framework can yield reasonable performance on a new platform with minimal effort.
Tasks
Published	2016-11-21
URL	http://arxiv.org/abs/1611.06945v1
PDF	http://arxiv.org/pdf/1611.06945v1.pdf
PWC	https://paperswithcode.com/paper/a-metaprogramming-and-autotuning-framework
Repo
Framework

Interactive algorithms: from pool to stream


Title	Interactive algorithms: from pool to stream
Authors	Sivan Sabato, Tom Hess
Abstract	We consider interactive algorithms in the pool-based setting, and in the stream-based setting. Interactive algorithms observe suggested elements (representing actions or queries), and interactively select some of them and receive responses. Pool-based algorithms can select elements at any order, while stream-based algorithms observe elements in sequence, and can only select elements immediately after observing them. We assume that the suggested elements are generated independently from some source distribution, and ask what is the stream size required for emulating a pool algorithm with a given pool size. We provide algorithms and matching lower bounds for general pool algorithms, and for utility-based pool algorithms. We further show that a maximal gap between the two settings exists also in the special case of active learning for binary classification.
Tasks	Active Learning
Published	2016-02-02
URL	http://arxiv.org/abs/1602.01132v3
PDF	http://arxiv.org/pdf/1602.01132v3.pdf
PWC	https://paperswithcode.com/paper/interactive-algorithms-from-pool-to-stream
Repo
Framework

The statistical trade-off between word order and word structure - large-scale evidence for the principle of least effort


Title	The statistical trade-off between word order and word structure - large-scale evidence for the principle of least effort
Authors	Alexander Koplenig, Peter Meyer, Sascha Wolfer, Carolin Mueller-Spitzer
Abstract	Languages employ different strategies to transmit structural and grammatical information. While, for example, grammatical dependency relationships in sentences are mainly conveyed by the ordering of the words for languages like Mandarin Chinese, or Vietnamese, the word ordering is much less restricted for languages such as Inupiatun or Quechua, as those languages (also) use the internal structure of words (e.g. inflectional morphology) to mark grammatical relationships in a sentence. Based on a quantitative analysis of more than 1,500 unique translations of different books of the Bible in more than 1,100 different languages that are spoken as a native language by approximately 6 billion people (more than 80% of the world population), we present large-scale evidence for a statistical trade-off between the amount of information conveyed by the ordering of words and the amount of information conveyed by internal word structure: languages that rely more strongly on word order information tend to rely less on word structure information and vice versa. In addition, we find that - despite differences in the way information is expressed - there is also evidence for a trade-off between different books of the biblical canon that recurs with little variation across languages: the more informative the word order of the book, the less informative its word structure and vice versa. We argue that this might suggest that, on the one hand, languages encode information in very different (but efficient) ways. On the other hand, content-related and stylistic features are statistically encoded in very similar ways.
Tasks
Published	2016-08-11
URL	http://arxiv.org/abs/1608.03587v2
PDF	http://arxiv.org/pdf/1608.03587v2.pdf
PWC	https://paperswithcode.com/paper/the-statistical-trade-off-between-word-order
Repo
Framework

A novel transfer learning method based on common space mapping and weighted domain matching


Title	A novel transfer learning method based on common space mapping and weighted domain matching
Authors	Ru-Ze Liang, Wei Xie, Weizhi Li, Hongqi Wang, Jim Jing-Yan Wang, Lisa Taylor
Abstract	In this paper, we propose a novel learning framework for the problem of domain transfer learning. We map the data of two domains to one single common space, and learn a classifier in this common space. Then we adapt the common classifier to the two domains by adding two adaptive functions to it respectively. In the common space, the target domain data points are weighted and matched to the target domain in term of distributions. The weighting terms of source domain data points and the target domain classification responses are also regularized by the local reconstruction coefficients. The novel transfer learning framework is evaluated over some benchmark cross-domain data sets, and it outperforms the existing state-of-the-art transfer learning methods.
Tasks	Transfer Learning
Published	2016-08-16
URL	http://arxiv.org/abs/1608.04581v1
PDF	http://arxiv.org/pdf/1608.04581v1.pdf
PWC	https://paperswithcode.com/paper/a-novel-transfer-learning-method-based-on
Repo
Framework

Authenticating users through their arm movement patterns


Title	Authenticating users through their arm movement patterns
Authors	Rajesh Kumar, Vir V Phoha, Rahul Raina
Abstract	In this paper, we propose four continuous authentication designs by using the characteristics of arm movements while individuals walk. The first design uses acceleration of arms captured by a smartwatch’s accelerometer sensor, the second design uses the rotation of arms captured by a smartwatch’s gyroscope sensor, third uses the fusion of both acceleration and rotation at the feature-level and fourth uses the fusion at score-level. Each of these designs is implemented by using four classifiers, namely, k nearest neighbors (k-NN) with Euclidean distance, Logistic Regression, Multilayer Perceptrons, and Random Forest resulting in a total of sixteen authentication mechanisms. These authentication mechanisms are tested under three different environments, namely an intra-session, inter-session on a dataset of 40 users and an inter-phase on a dataset of 12 users. The sessions of data collection were separated by at least ten minutes, whereas the phases of data collection were separated by at least three months. Under the intra-session environment, all of the twelve authentication mechanisms achieve a mean dynamic false accept rate (DFAR) of 0% and dynamic false reject rate (DFRR) of 0%. For the inter-session environment, feature level fusion-based design with classifier k-NN achieves the best error rates that are a mean DFAR of 2.2% and DFRR of 4.2%. The DFAR and DFRR increased from 5.68% and 4.23% to 15.03% and 14.62% respectively when feature level fusion-based design with classifier k-NN was tested under the inter-phase environment on a dataset of 12 users.
Tasks
Published	2016-03-07
URL	http://arxiv.org/abs/1603.02211v1
PDF	http://arxiv.org/pdf/1603.02211v1.pdf
PWC	https://paperswithcode.com/paper/authenticating-users-through-their-arm
Repo
Framework

Large-Scale Machine Translation between Arabic and Hebrew: Available Corpora and Initial Results


Title	Large-Scale Machine Translation between Arabic and Hebrew: Available Corpora and Initial Results
Authors	Yonatan Belinkov, James Glass
Abstract	Machine translation between Arabic and Hebrew has so far been limited by a lack of parallel corpora, despite the political and cultural importance of this language pair. Previous work relied on manually-crafted grammars or pivoting via English, both of which are unsatisfactory for building a scalable and accurate MT system. In this work, we compare standard phrase-based and neural systems on Arabic-Hebrew translation. We experiment with tokenization by external tools and sub-word modeling by character-level neural models, and show that both methods lead to improved translation performance, with a small advantage to the neural models.
Tasks	Machine Translation, Tokenization
Published	2016-09-25
URL	http://arxiv.org/abs/1609.07701v1
PDF	http://arxiv.org/pdf/1609.07701v1.pdf
PWC	https://paperswithcode.com/paper/large-scale-machine-translation-between
Repo
Framework

Improving Variational Auto-Encoders using Householder Flow


Title	Improving Variational Auto-Encoders using Householder Flow
Authors	Jakub M. Tomczak, Max Welling
Abstract	Variational auto-encoders (VAE) are scalable and powerful generative models. However, the choice of the variational posterior determines tractability and flexibility of the VAE. Commonly, latent variables are modeled using the normal distribution with a diagonal covariance matrix. This results in computational efficiency but typically it is not flexible enough to match the true posterior distribution. One fashion of enriching the variational posterior distribution is application of normalizing flows, i.e., a series of invertible transformations to latent variables with a simple posterior. In this paper, we follow this line of thinking and propose a volume-preserving flow that uses a series of Householder transformations. We show empirically on MNIST dataset and histopathology data that the proposed flow allows to obtain more flexible variational posterior and competitive results comparing to other normalizing flows.
Tasks
Published	2016-11-29
URL	http://arxiv.org/abs/1611.09630v4
PDF	http://arxiv.org/pdf/1611.09630v4.pdf
PWC	https://paperswithcode.com/paper/improving-variational-auto-encoders-using-1
Repo
Framework

Generating Object Cluster Hierarchies for Benchmarking


Title	Generating Object Cluster Hierarchies for Benchmarking
Authors	Łukasz P. Olech, Michał Spytkowski, Halina Kwasnicka
Abstract	Generation of Object Cluster Hierarchies is a new variant of Hierarchical Clustering that increasingly gains more interest in the field of Machine Learning. Being a novelty, the lack of tools for systematic analysis and comparison of Object Cluster Hierarchies inhibits its further development. In this paper, we propose a novel method for generating hierarchical structures of data based on Tree-Structured Stick Breaking Process that can be used for benchmarking. The article presents thorough empirical and theoretical analysis of the method revealing its characteristics. More importantly, the intuition how to operate with model parameters and a set of benchmarking datasets are provided. Conducted experiments show usefulness of the model as high flexibility in generating a wide range of differently-structured data is achieved. The developed generator together with proposed benchmarks are publicly available (http://kio.pwr.edu.pl/?page_id=396).
Tasks
Published	2016-06-17
URL	http://arxiv.org/abs/1606.05681v2
PDF	http://arxiv.org/pdf/1606.05681v2.pdf
PWC	https://paperswithcode.com/paper/generating-object-cluster-hierarchies-for
Repo
Framework