text
stringlengths 17
3.36M
| source
stringlengths 3
333
| __index_level_0__
int64 0
518k
|
|---|---|---|
Automatic Speech Recognition involves mainly two steps; feature extraction and classification . Mel Frequency Cepstral Coefficient is used as one of the prominent feature extraction techniques in ASR. Usually, the set of all 12 MFCC coefficients is used as the feature vector in the classification step. But the question is whether the same or improved classification accuracy can be achieved by using a subset of 12 MFCC as feature vector. In this paper, Fisher's ratio technique is used for selecting a subset of 12 MFCC coefficients that contribute more in discriminating a pattern. The selected coefficients are used in classification with Hidden Markov Model algorithm. The classification accuracies that we get by using 12 coefficients and by using the selected coefficients are compared.
|
Feature selection using Fisher's ratio technique for automatic speech
recognition
| 1,900
|
Statistical studies of languages have focused on the rank-frequency distribution of words. Instead, we introduce here a measure of how word ranks change in time and call this distribution \emph{rank diversity}. We calculate this diversity for books published in six European languages since 1800, and find that it follows a universal lognormal distribution. Based on the mean and standard deviation associated with the lognormal distribution, we define three different word regimes of languages: "heads" consist of words which almost do not change their rank in time, "bodies" are words of general use, while "tails" are comprised by context-specific words and vary their rank considerably in time. The heads and bodies reflect the size of language cores identified by linguists for basic communication. We propose a Gaussian random walk model which reproduces the rank variation of words in time and thus the diversity. Rank diversity of words can be understood as the result of random variations in rank, where the size of the variation depends on the rank itself. We find that the core size is similar for all languages studied.
|
Rank diversity of languages: Generic behavior in computational
linguistics
| 1,901
|
We present an annotation schema as part of an effort to create a manually annotated corpus for Arabic dialogue language understanding including spoken dialogue and written "chat" dialogue for inquiry-answer domain. The proposed schema handles mainly the request and response acts that occurs frequently in inquiry-answer debate conversations expressing request services, suggests, and offers. We applied the proposed schema on 83 Arabic inquiry-answer dialogues.
|
Arabic Inquiry-Answer Dialogue Acts Annotation Schema
| 1,902
|
Twitter, a popular social media outlet, has evolved into a vast source of linguistic data, rich with opinion, sentiment, and discussion. Due to the increasing popularity of Twitter, its perceived potential for exerting social influence has led to the rise of a diverse community of automatons, commonly referred to as bots. These inorganic and semi-organic Twitter entities can range from the benevolent (e.g., weather-update bots, help-wanted-alert bots) to the malevolent (e.g., spamming messages, advertisements, or radical opinions). Existing detection algorithms typically leverage meta-data (time between tweets, number of followers, etc.) to identify robotic accounts. Here, we present a powerful classification scheme that exclusively uses the natural language text from organic users to provide a criterion for identifying accounts posting automated messages. Since the classifier operates on text alone, it is flexible and may be applied to any textual data beyond the Twitter-sphere.
|
Sifting Robotic from Organic Text: A Natural Language Approach for
Detecting Automation on Twitter
| 1,903
|
This thesis presents a study about the integration of information about Multiword Expressions (MWEs) into parsing with Combinatory Categorial Grammar (CCG). We build on previous work which has shown the benefit of adding information about MWEs to syntactic parsing by implementing a similar pipeline with CCG parsing. More specifically, we collapse MWEs to one token in training and test data in CCGbank, a corpus which contains sentences annotated with CCG derivations. Our collapsing algorithm however can only deal with MWEs when they form a constituent in the data which is one of the limitations of our approach. We study the effect of collapsing training and test data. A parsing effect can be obtained if collapsed data help the parser in its decisions and a training effect can be obtained if training on the collapsed data improves results. We also collapse the gold standard and show that our model significantly outperforms the baseline model on our gold standard, which indicates that there is a training effect. We show that the baseline model performs significantly better on our gold standard when the data are collapsed before parsing than when the data are collapsed after parsing which indicates that there is a parsing effect. We show that these results can lead to improved performance on the non-collapsed standard benchmark although we fail to show that it does so significantly. We conclude that despite the limited settings, there are noticeable improvements from using MWEs in parsing. We discuss ways in which the incorporation of MWEs into parsing can be improved and hypothesize that this will lead to more substantial results. We finally show that turning the MWE recognition part of the pipeline into an experimental part is a useful thing to do as we obtain different results with different recognizers.
|
CCG Parsing and Multiword Expressions
| 1,904
|
Word embedding, which refers to low-dimensional dense vector representations of natural words, has demonstrated its power in many natural language processing tasks. However, it may suffer from the inaccurate and incomplete information contained in the free text corpus as training data. To tackle this challenge, there have been quite a few works that leverage knowledge graphs as an additional information source to improve the quality of word embedding. Although these works have achieved certain success, they have neglected some important facts about knowledge graphs: (i) many relationships in knowledge graphs are \emph{many-to-one}, \emph{one-to-many} or even \emph{many-to-many}, rather than simply \emph{one-to-one}; (ii) most head entities and tail entities in knowledge graphs come from very different semantic spaces. To address these issues, in this paper, we propose a new algorithm named ProjectNet. ProjecNet models the relationships between head and tail entities after transforming them with different low-rank projection matrices. The low-rank projection can allow non \emph{one-to-one} relationships between entities, while different projection matrices for head and tail entities allow them to originate in different semantic spaces. The experimental results demonstrate that ProjectNet yields more accurate word embedding than previous works, thus leads to clear improvements in various natural language processing tasks.
|
Learning Better Word Embedding by Asymmetric Low-Rank Projection of
Knowledge Graph
| 1,905
|
Most state-of-the-art named entity recognition (NER) systems rely on handcrafted features and on the output of other NLP tasks such as part-of-speech (POS) tagging and text chunking. In this work we propose a language-independent NER system that uses automatically learned features only. Our approach is based on the CharWNN deep neural network, which uses word-level and character-level representations (embeddings) to perform sequential classification. We perform an extensive number of experiments using two annotated corpora in two different languages: HAREM I corpus, which contains texts in Portuguese; and the SPA CoNLL-2002 corpus, which contains texts in Spanish. Our experimental results shade light on the contribution of neural character embeddings for NER. Moreover, we demonstrate that the same neural network which has been successfully applied to POS tagging can also achieve state-of-the-art results for language-independet NER, using the same hyperparameters, and without any handcrafted features. For the HAREM I corpus, CharWNN outperforms the state-of-the-art system by 7.9 points in the F1-score for the total scenario (ten NE classes), and by 7.2 points in the F1 for the selective scenario (five NE classes).
|
Boosting Named Entity Recognition with Neural Character Embeddings
| 1,906
|
Knowledge graph embedding refers to projecting entities and relations in knowledge graph into continuous vector spaces. State-of-the-art methods, such as TransE, TransH, and TransR build embeddings by treating relation as translation from head entity to tail entity. However, previous models can not deal with reflexive/one-to-many/many-to-one/many-to-many relations properly, or lack of scalability and efficiency. Thus, we propose a novel method, flexible translation, named TransF, to address the above issues. TransF regards relation as translation between head entity vector and tail entity vector with flexible magnitude. To evaluate the proposed model, we conduct link prediction and triple classification on benchmark datasets. Experimental results show that our method remarkably improve the performance compared with several state-of-the-art baselines.
|
Knowlege Graph Embedding by Flexible Translation
| 1,907
|
Translation Memory (TM) systems are one of the most widely used translation technologies. An important part of TM systems is the matching algorithm that determines what translations get retrieved from the bank of available translations to assist the human translator. Although detailed accounts of the matching algorithms used in commercial systems can't be found in the literature, it is widely believed that edit distance algorithms are used. This paper investigates and evaluates the use of several matching algorithms, including the edit distance algorithm that is believed to be at the heart of most modern commercial TM systems. This paper presents results showing how well various matching algorithms correlate with human judgments of helpfulness (collected via crowdsourcing with Amazon's Mechanical Turk). A new algorithm based on weighted n-gram precision that can be adjusted for translator length preferences consistently returns translations judged to be most helpful by translators for multiple domains and language pairs.
|
Translation Memory Retrieval Methods
| 1,908
|
We describe the latest improvements to the IBM English conversational telephone speech recognition system. Some of the techniques that were found beneficial are: maxout networks with annealed dropout rates; networks with a very large number of outputs trained on 2000 hours of data; joint modeling of partially unfolded recurrent neural networks and convolutional nets by combining the bottleneck and output layers and retraining the resulting model; and lastly, sophisticated language model rescoring with exponential and neural network LMs. These techniques result in an 8.0% word error rate on the Switchboard part of the Hub5-2000 evaluation test set which is 23% relative better than our previous best published result.
|
The IBM 2015 English Conversational Telephone Speech Recognition System
| 1,909
|
The development of methods to deal with the informative contents of the text units in the matching process is a major challenge in automatic summary evaluation systems that use fixed n-gram matching. The limitation causes inaccurate matching between units in a peer and reference summaries. The present study introduces a new Keyphrase based Summary Evaluator KpEval for evaluating automatic summaries. The KpEval relies on the keyphrases since they convey the most important concepts of a text. In the evaluation process, the keyphrases are used in their lemma form as the matching text unit. The system was applied to evaluate different summaries of Arabic multi-document data set presented at TAC2011. The results showed that the new evaluation technique correlates well with the known evaluation systems: Rouge1, Rouge2, RougeSU4, and AutoSummENG MeMoG. KpEval has the strongest correlation with AutoSummENG MeMoG, Pearson and spearman correlation coefficient measures are 0.8840, 0.9667 respectively.
|
Keyphrase Based Evaluation of Automatic Text Summarization
| 1,910
|
NLP tasks differ in the semantic information they require, and at this time no single se- mantic representation fulfills all requirements. Logic-based representations characterize sentence structure, but do not capture the graded aspect of meaning. Distributional models give graded similarity ratings for words and phrases, but do not capture sentence structure in the same detail as logic-based approaches. So it has been argued that the two are complementary. We adopt a hybrid approach that combines logic-based and distributional semantics through probabilistic logic inference in Markov Logic Networks (MLNs). In this paper, we focus on the three components of a practical system integrating logical and distributional models: 1) Parsing and task representation is the logic-based part where input problems are represented in probabilistic logic. This is quite different from representing them in standard first-order logic. 2) For knowledge base construction we form weighted inference rules. We integrate and compare distributional information with other sources, notably WordNet and an existing paraphrase collection. In particular, we use our system to evaluate distributional lexical entailment approaches. We use a variant of Robinson resolution to determine the necessary inference rules. More sources can easily be added by mapping them to logical rules; our system learns a resource-specific weight that corrects for scaling differences between resources. 3) In discussing probabilistic inference, we show how to solve the inference problems efficiently. To evaluate our approach, we use the task of textual entailment (RTE), which can utilize the strengths of both logic-based and distributional representations. In particular we focus on the SICK dataset, where we achieve state-of-the-art results.
|
Representing Meaning with a Combination of Logical and Distributional
Models
| 1,911
|
Meaning of a word varies from one domain to another. Despite this important domain dependence in word semantics, existing word representation learning methods are bound to a single domain. Given a pair of \emph{source}-\emph{target} domains, we propose an unsupervised method for learning domain-specific word representations that accurately capture the domain-specific aspects of word semantics. First, we select a subset of frequent words that occur in both domains as \emph{pivots}. Next, we optimize an objective function that enforces two constraints: (a) for both source and target domain documents, pivots that appear in a document must accurately predict the co-occurring non-pivots, and (b) word representations learnt for pivots must be similar in the two domains. Moreover, we propose a method to perform domain adaptation using the learnt word representations. Our proposed method significantly outperforms competitive baselines including the state-of-the-art domain-insensitive word representations, and reports best sentiment classification accuracies for all domain-pairs in a benchmark dataset.
|
Unsupervised Cross-Domain Word Representation Learning
| 1,912
|
In this paper, we give an overview for the shared task at the 4th CCF Conference on Natural Language Processing \& Chinese Computing (NLPCC 2015): Chinese word segmentation and part-of-speech (POS) tagging for micro-blog texts. Different with the popular used newswire datasets, the dataset of this shared task consists of the relatively informal micro-texts. The shared task has two sub-tasks: (1) individual Chinese word segmentation and (2) joint Chinese word segmentation and POS Tagging. Each subtask has three tracks to distinguish the systems with different resources. We first introduce the dataset and task, then we characterize the different approaches of the participating systems, report the test results, and provide a overview analysis of these results. An online system is available for open registration and evaluation at http://nlp.fudan.edu.cn/nlpcc2015.
|
Overview of the NLPCC 2015 Shared Task: Chinese Word Segmentation and
POS Tagging for Micro-blog Texts
| 1,913
|
Learning vector representation for words is an important research field which may benefit many natural language processing tasks. Two limitations exist in nearly all available models, which are the bias caused by the context definition and the lack of knowledge utilization. They are difficult to tackle because these algorithms are essentially unsupervised learning approaches. Inspired by deep learning, the authors propose a supervised framework for learning vector representation of words to provide additional supervised fine tuning after unsupervised learning. The framework is knowledge rich approacher and compatible with any numerical vectors word representation. The authors perform both intrinsic evaluation like attributional and relational similarity prediction and extrinsic evaluations like the sentence completion and sentiment analysis. Experiments results on 6 embeddings and 4 tasks with 10 datasets show that the proposed fine tuning framework may significantly improve the quality of the vector representation of words.
|
Supervised Fine Tuning for Word Embedding with Integrated Knowledge
| 1,914
|
In this introductory article we present the basics of an approach to implementing computational interpreting of natural language aiming to model the meanings of words and phrases. Unlike other approaches, we attempt to define the meanings of text fragments in a composable and computer interpretable way. We discuss models and ideas for detecting different types of semantic incomprehension and choosing the interpretation that makes most sense in a given context. Knowledge representation is designed for handling context-sensitive and uncertain / imprecise knowledge, and for easy accommodation of new information. It stores quantitative information capturing the essence of the concepts, because it is crucial for working with natural language understanding and reasoning. Still, the representation is general enough to allow for new knowledge to be learned, and even generated by the system. The article concludes by discussing some reasoning-related topics: possible approaches to generation of new abstract concepts, and describing situations and concepts in words (e.g. for specifying interpretation difficulties).
|
Modeling meaning: computational interpreting and understanding of
natural language fragments
| 1,915
|
This paper reports on the comparison of the subject and object of verbs in their usage between phishing emails and legitimate emails. The purpose of this research is to explore whether the syntactic structures and subjects and objects of verbs can be distinguishable features for phishing detection. To achieve the objective, we have conducted two series of experiments: the syntactic similarity for sentences, and the subject and object of verb comparison. The results of the experiments indicated that both features can be used for some verbs, but more work has to be done for others.
|
Using Syntactic Features for Phishing Detection
| 1,916
|
Sequence-to-sequence translation methods based on generation with a side-conditioned language model have recently shown promising results in several tasks. In machine translation, models conditioned on source side words have been used to produce target-language text, and in image captioning, models conditioned images have been used to generate caption text. Past work with this approach has focused on large vocabulary tasks, and measured quality in terms of BLEU. In this paper, we explore the applicability of such models to the qualitatively different grapheme-to-phoneme task. Here, the input and output side vocabularies are small, plain n-gram models do well, and credit is only given when the output is exactly correct. We find that the simple side-conditioned generation approach is able to rival the state-of-the-art, and we are able to significantly advance the stat-of-the-art with bi-directional long short-term memory (LSTM) neural networks that use the same alignment information that is used in conventional approaches.
|
Sequence-to-Sequence Neural Net Models for Grapheme-to-Phoneme
Conversion
| 1,917
|
We describe an approach to create a diverse set of predictions with spectral learning of latent-variable PCFGs (L-PCFGs). Our approach works by creating multiple spectral models where noise is added to the underlying features in the training set before the estimation of each model. We describe three ways to decode with multiple models. In addition, we describe a simple variant of the spectral algorithm for L-PCFGs that is fast and leads to compact models. Our experiments for natural language parsing, for English and German, show that we get a significant improvement over baselines comparable to state of the art. For English, we achieve the $F_1$ score of 90.18, and for German we achieve the $F_1$ score of 83.38.
|
Diversity in Spectral Learning for Natural Language Parsing
| 1,918
|
Representation learning of knowledge bases (KBs) aims to embed both entities and relations into a low-dimensional space. Most existing methods only consider direct relations in representation learning. We argue that multiple-step relation paths also contain rich inference patterns between entities, and propose a path-based representation learning model. This model considers relation paths as translations between entities for representation learning, and addresses two key challenges: (1) Since not all relation paths are reliable, we design a path-constraint resource allocation algorithm to measure the reliability of relation paths. (2) We represent relation paths via semantic composition of relation embeddings. Experimental results on real-world datasets show that, as compared with baselines, our model achieves significant and consistent improvements on knowledge base completion and relation extraction from text.
|
Modeling Relation Paths for Representation Learning of Knowledge Bases
| 1,919
|
In this paper, we propose two new features for estimating phrase-based machine translation parameters from mainly monolingual data. Our method is based on two recently introduced neural network vector representation models for words and sentences. It is the first time that these models have been used in an end to end phrase-based machine translation system. Scores obtained from our method can recover more than 80% of BLEU loss caused by removing phrase table probabilities. We also show that our features combined with the phrase table probabilities improve the BLEU score by absolute 0.74 points.
|
Monolingually Derived Phrase Scores for Phrase Based SMT Using Neural
Networks Vector Representations
| 1,920
|
In this paper, we present a novel approach for medical synonym extraction. We aim to integrate the term embedding with the medical domain knowledge for healthcare applications. One advantage of our method is that it is very scalable. Experiments on a dataset with more than 1M term pairs show that the proposed approach outperforms the baseline approaches by a large margin.
|
Medical Synonym Extraction with Concept Space Models
| 1,921
|
We present a three-pronged approach to improving Statistical Machine Translation (SMT), building on recent success in the application of neural networks to SMT. First, we propose new features based on neural networks to model various non-local translation phenomena. Second, we augment the architecture of the neural network with tensor layers that capture important higher-order interaction among the network units. Third, we apply multitask learning to estimate the neural network parameters jointly. Each of our proposed methods results in significant improvements that are complementary. The overall improvement is +2.7 and +1.8 BLEU points for Arabic-English and Chinese-English translation over a state-of-the-art system that already includes neural network features.
|
Statistical Machine Translation Features with Multitask Tensor Networks
| 1,922
|
This article presents an analysis of the influence of context information on dialog act recognition. We performed experiments on the widely explored Switchboard corpus, as well as on data annotated according to the recent ISO 24617-2 standard. The latter was obtained from the Tilburg DialogBank and through the mapping of the annotations of a subset of the Let's Go corpus. We used a classification approach based on SVMs, which had proved successful in previous work and allowed us to limit the amount of context information provided. This way, we were able to observe the influence patterns as the amount of context information increased. Our base features consisted of n-grams, punctuation, and wh-words. Context information was obtained from one to five preceding segments and provided either as n-grams or dialog act classifications, with the latter typically leading to better results and more stable influence patterns. In addition to the conclusions about the importance and influence of context information, our experiments on the Switchboard corpus also led to results that advanced the state-of-the-art on the dialog act recognition task on that corpus. Furthermore, the results obtained on data annotated according to the ISO 24617-2 standard define a baseline for future work and contribute for the standardization of experiments in the area.
|
The Influence of Context on Dialogue Act Recognition
| 1,923
|
Natural language generation of coherent long texts like paragraphs or longer documents is a challenging problem for recurrent networks models. In this paper, we explore an important step toward this generation task: training an LSTM (Long-short term memory) auto-encoder to preserve and reconstruct multi-sentence paragraphs. We introduce an LSTM model that hierarchically builds an embedding for a paragraph from embeddings for sentences and words, then decodes this embedding to reconstruct the original paragraph. We evaluate the reconstructed paragraph using standard metrics like ROUGE and Entity Grid, showing that neural models are able to encode texts in a way that preserve syntactic, semantic, and discourse coherence. While only a first step toward generating coherent text units from neural models, our work has the potential to significantly impact natural language generation and summarization\footnote{Code for the three models described in this paper can be found at www.stanford.edu/~jiweil/ .
|
A Hierarchical Neural Autoencoder for Paragraphs and Documents
| 1,924
|
While neural networks have been successfully applied to many NLP tasks the resulting vector-based models are very difficult to interpret. For example it's not clear how they achieve {\em compositionality}, building sentence meaning from the meanings of words and phrases. In this paper we describe four strategies for visualizing compositionality in neural models for NLP, inspired by similar work in computer vision. We first plot unit values to visualize compositionality of negation, intensification, and concessive clauses, allow us to see well-known markedness asymmetries in negation. We then introduce three simple and straightforward methods for visualizing a unit's {\em salience}, the amount it contributes to the final composed meaning: (1) gradient back-propagation, (2) the variance of a token from the average word node, (3) LSTM-style gates that measure information flow. We test our methods on sentiment using simple recurrent nets and LSTMs. Our general-purpose methods may have wide applications for understanding compositionality and other semantic properties of deep networks , and also shed light on why LSTMs outperform simple recurrent nets,
|
Visualizing and Understanding Neural Models in NLP
| 1,925
|
Learning a distinct representation for each sense of an ambiguous word could lead to more powerful and fine-grained models of vector-space representations. Yet while `multi-sense' methods have been proposed and tested on artificial word-similarity tasks, we don't know if they improve real natural language understanding tasks. In this paper we introduce a multi-sense embedding model based on Chinese Restaurant Processes that achieves state of the art performance on matching human word similarity judgments, and propose a pipelined architecture for incorporating multi-sense embeddings into language understanding. We then test the performance of our model on part-of-speech tagging, named entity recognition, sentiment analysis, semantic relation identification and semantic relatedness, controlling for embedding dimensionality. We find that multi-sense embeddings do improve performance on some tasks (part-of-speech tagging, semantic relation identification, semantic relatedness) but not on others (named entity recognition, various forms of sentiment analysis). We discuss how these differences may be caused by the different role of word sense information in each of the tasks. The results highlight the importance of testing embedding models in real applications.
|
Do Multi-Sense Embeddings Improve Natural Language Understanding?
| 1,926
|
The interest in statistical machine translation systems increases currently due to political and social events in the world. A proposed Statistical Machine Translation (SMT) based model that can be used to translate a sentence from the source Language (English) to the target language (Arabic) automatically through efficiently incorporating different statistical and Natural Language Processing (NLP) models such as language model, alignment model, phrase based model, reordering model, and translation model. These models are combined to enhance the performance of statistical machine translation (SMT). Many implementation tools have been used in this work such as Moses, Gizaa++, IRSTLM, KenLM, and BLEU. Based on the implementation, evaluation of this model, and comparing the generated translation with other implemented machine translation systems like Google Translate, it was proved that this proposed model has enhanced the results of the statistical machine translation, and forms a reliable and efficient model in this field of research.
|
A Hybrid Model for Enhancing Lexical Statistical Machine Translation
(SMT)
| 1,927
|
Although, the fair amount of works in sentiment analysis (SA) and opinion mining (OM) systems in the last decade and with respect to the performance of these systems, but it still not desired performance, especially for morphologically-Rich Language (MRL) such as Arabic, due to the complexities and challenges exist in the nature of the languages itself. One of these challenges is the detection of idioms or proverbs phrases within the writer text or comment. An idiom or proverb is a form of speech or an expression that is peculiar to itself. Grammatically, it cannot be understood from the individual meanings of its elements and can yield different sentiment when treats as separate words. Consequently, In order to facilitate the task of detection and classification of lexical phrases for automated SA systems, this paper presents AIPSeLEX a novel idioms/ proverbs sentiment lexicon for modern standard Arabic (MSA) and colloquial. AIPSeLEX is manually collected and annotated at sentence level with semantic orientation (positive or negative). The efforts of manually building and annotating the lexicon are reported. Moreover, we build a classifier that extracts idioms and proverbs, phrases from text using n-gram and similarity measure methods. Finally, several experiments were carried out on various data, including Arabic tweets and Arabic microblogs (hotel reservation, product reviews, and TV program comments) from publicly available Arabic online reviews websites (social media, blogs, forums, e-commerce web sites) to evaluate the coverage and accuracy of AIPSeLEX.
|
Idioms-Proverbs Lexicon for Modern Standard Arabic and Colloquial
Sentiment Analysis
| 1,928
|
The quality and quantity of articles in each Wikipedia language varies greatly. Translating from another Wikipedia is a natural way to add more content, but the translation process is not properly supported in the software used by Wikipedia. Past computer-assisted translation tools built for Wikipedia are not commonly used. We created a tool that adapts to the specific needs of an open community and to the kind of content in Wikipedia. Qualitative and quantitative data indicates that the new tool helps users translate articles easier and faster.
|
Content Translation: Computer-assisted translation tool for Wikipedia
articles
| 1,929
|
Current distributed representations of words show little resemblance to theories of lexical semantics. The former are dense and uninterpretable, the latter largely based on familiar, discrete classes (e.g., supersenses) and relations (e.g., synonymy and hypernymy). We propose methods that transform word vectors into sparse (and optionally binary) vectors. The resulting representations are more similar to the interpretable features typically used in NLP, though they are discovered automatically from raw corpora. Because the vectors are highly sparse, they are computationally easy to work with. Most importantly, we find that they outperform the original vectors on benchmark tasks.
|
Sparse Overcomplete Word Vector Representations
| 1,930
|
Twitter is often used in quantitative studies that identify geographically-preferred topics, writing styles, and entities. These studies rely on either GPS coordinates attached to individual messages, or on the user-supplied location field in each profile. In this paper, we compare these data acquisition techniques and quantify the biases that they introduce; we also measure their effects on linguistic analysis and text-based geolocation. GPS-tagging and self-reported locations yield measurably different corpora, and these linguistic differences are partially attributable to differences in dataset composition by age and gender. Using a latent variable model to induce age and gender, we show how these demographic variables interact with geography to affect language use. We also show that the accuracy of text-based geolocation varies with population demographics, giving the best results for men above the age of 40.
|
Confounds and Consequences in Geotagged Twitter Data
| 1,931
|
We introduce a corpus of 7,032 sentences rated by human annotators for formality, informativeness, and implicature on a 1-7 scale. The corpus was annotated using Amazon Mechanical Turk. Reliability in the obtained judgments was examined by comparing mean ratings across two MTurk experiments, and correlation with pilot annotations (on sentence formality) conducted in a more controlled setting. Despite the subjectivity and inherent difficulty of the annotation task, correlations between mean ratings were quite encouraging, especially on formality and informativeness. We further explored correlation between the three linguistic variables, genre-wise variation of ratings and correlations within genres, compatibility with automatic stylistic scoring, and sentential make-up of a document in terms of style. To date, our corpus is the largest sentence-level annotated corpus released for formality, informativeness, and implicature.
|
SQUINKY! A Corpus of Sentence-level Formality, Informativeness, and
Implicature
| 1,932
|
Natural Language Processing (NLP) systems commonly leverage bag-of-words co-occurrence techniques to capture semantic and syntactic word relationships. The resulting word-level distributed representations often ignore morphological information, though character-level embeddings have proven valuable to NLP tasks. We propose a new neural language model incorporating both word order and character order in its embedding. The model produces several vector spaces with meaningful substructure, as evidenced by its performance of 85.8% on a recent word-analogy task, exceeding best published syntactic word-analogy scores by a 58% error margin. Furthermore, the model includes several parallel training methods, most notably allowing a skip-gram network with 160 billion parameters to be trained overnight on 3 multi-core CPUs, 14x larger than the previous largest neural network.
|
Modeling Order in Neural Word Embeddings at Scale
| 1,933
|
Through a particular choice of a predicate (e.g., "x violated y"), a writer can subtly connote a range of implied sentiments and presupposed facts about the entities x and y: (1) writer's perspective: projecting x as an "antagonist"and y as a "victim", (2) entities' perspective: y probably dislikes x, (3) effect: something bad happened to y, (4) value: y is something valuable, and (5) mental state: y is distressed by the event. We introduce connotation frames as a representation formalism to organize these rich dimensions of connotation using typed relations. First, we investigate the feasibility of obtaining connotative labels through crowdsourcing experiments. We then present models for predicting the connotation frames of verb predicates based on their distributional word representations and the interplay between different types of connotative relations. Empirical results confirm that connotation frames can be induced from various data sources that reflect how people use language and give rise to the connotative meanings. We conclude with analytical results that show the potential use of connotation frames for analyzing subtle biases in online news media.
|
Connotation Frames: A Data-Driven Investigation
| 1,934
|
The Abstract Meaning Representation (AMR) is a representation for open-domain rich semantics, with potential use in fields like event extraction and machine translation. Node generation, typically done using a simple dictionary lookup, is currently an important limiting factor in AMR parsing. We propose a small set of actions that derive AMR subgraphs by transformations on spans of text, which allows for more robust learning of this stage. Our set of construction actions generalize better than the previous approach, and can be learned with a simple classifier. We improve on the previous state-of-the-art result for AMR parsing, boosting end-to-end performance by 3 F$_1$ on both the LDC2013E117 and LDC2014T12 datasets.
|
Robust Subgraph Generation Improves Abstract Meaning Representation
Parsing
| 1,935
|
Communicative interactions involve a kind of procedural knowledge that is used by the human brain for processing verbal and nonverbal inputs and for language production. Although considerable work has been done on modeling human language abilities, it has been difficult to bring them together to a comprehensive tabula rasa system compatible with current knowledge of how verbal information is processed in the brain. This work presents a cognitive system, entirely based on a large-scale neural architecture, which was developed to shed light on the procedural knowledge involved in language elaboration. The main component of this system is the central executive, which is a supervising system that coordinates the other components of the working memory. In our model, the central executive is a neural network that takes as input the neural activation states of the short-term memory and yields as output mental actions, which control the flow of information among the working memory components through neural gating mechanisms. The proposed system is capable of learning to communicate through natural language starting from tabula rasa, without any a priori knowledge of the structure of phrases, meaning of words, role of the different classes of words, only by interacting with a human through a text-based interface, using an open-ended incremental learning process. It is able to learn nouns, verbs, adjectives, pronouns and other word classes, and to use them in expressive language. The model was validated on a corpus of 1587 input sentences, based on literature on early language assessment, at the level of about 4-years old child, and produced 521 output sentences, expressing a broad range of language processing functionalities.
|
A cognitive neural architecture able to learn and communicate through
natural language
| 1,936
|
Building unified timelines from a collection of written news articles requires cross-document event coreference resolution and temporal relation extraction. In this paper we present an approach event coreference resolution according to: a) similar temporal information, and b) similar semantic arguments. Temporal information is detected using an automatic temporal information system (TIPSem), while semantic information is represented by means of LDA Topic Modeling. The evaluation of our approach shows that it obtains the highest Micro-average F-score results in the SemEval2015 Task 4: TimeLine: Cross-Document Event Ordering (25.36\% for TrackB, 23.15\% for SubtrackB), with an improvement of up to 6\% in comparison to the other systems. However, our experiment also showed some draw-backs in the Topic Modeling approach that degrades performance of the system.
|
Combining Temporal Information and Topic Modeling for Cross-Document
Event Ordering
| 1,937
|
The Paraphrase Database (PPDB; Ganitkevitch et al., 2013) is an extensive semantic resource, consisting of a list of phrase pairs with (heuristic) confidence estimates. However, it is still unclear how it can best be used, due to the heuristic nature of the confidences and its necessarily incomplete coverage. We propose models to leverage the phrase pairs from the PPDB to build parametric paraphrase models that score paraphrase pairs more accurately than the PPDB's internal scores while simultaneously improving its coverage. They allow for learning phrase embeddings as well as improved word embeddings. Moreover, we introduce two new, manually annotated datasets to evaluate short-phrase paraphrasing models. Using our paraphrase model trained using PPDB, we achieve state-of-the-art results on standard word and bigram similarity tasks and beat strong baselines on our new short phrase paraphrase tasks.
|
From Paraphrase Database to Compositional Paraphrase Model and Back
| 1,938
|
We propose Imaginet, a model of learning visually grounded representations of language from coupled textual and visual input. The model consists of two Gated Recurrent Unit networks with shared word embeddings, and uses a multi-task objective by receiving a textual description of a scene and trying to concurrently predict its visual representation and the next word in the sentence. Mimicking an important aspect of human language learning, it acquires meaning representations for individual words from descriptions of visual scenes. Moreover, it learns to effectively use sequential structure in semantic interpretation of multi-word phrases.
|
Learning language through pictures
| 1,939
|
Our dictionary-based lemmatizer for the Bulgarian language presented here is distributed as free software, publicly available to download and use under the GPL v3 license. The presented software is written entirely in Java and is distributed as a GATE plugin. To our best knowledge, at the time of writing this article, there are not any other free lemmatization tools specifically targeting the Bulgarian language. The presented lemmatizer is a work in progress and currently yields an accuracy of about 95% in comparison to the manually annotated corpus BulTreeBank-Morph, which contains 273933 tokens.
|
A Publicly Available Cross-Platform Lemmatizer for Bulgarian
| 1,940
|
This paper reveals the results of an analysis of the accuracy of developed software for automatic lemmatization for the Bulgarian language. This lemmatization software is written entirely in Java and is distributed as a GATE plugin. Certain statistical methods are used to define the accuracy of this software. The results of the analysis show 95% lemmatization accuracy.
|
Evaluation of the Accuracy of the BGLemmatizer
| 1,941
|
We propose a simple, scalable, fully generative model for transition-based dependency parsing with high accuracy. The model, parameterized by Hierarchical Pitman-Yor Processes, overcomes the limitations of previous generative models by allowing fast and accurate inference. We propose an efficient decoding algorithm based on particle filtering that can adapt the beam size to the uncertainty in the model while jointly predicting POS tags and parse trees. The UAS of the parser is on par with that of a greedy discriminative baseline. As a language model, it obtains better perplexity than a n-gram model by performing semi-supervised learning over a large unlabelled corpus. We show that the model is able to generate locally and syntactically coherent sentences, opening the door to further applications in language generation.
|
A Bayesian Model for Generative Transition-based Dependency Parsing
| 1,942
|
We present our work on semi-supervised parsing of natural language sentences, focusing on multi-source crosslingual transfer of delexicalized dependency parsers. We first evaluate the influence of treebank annotation styles on parsing performance, focusing on adposition attachment style. Then, we present KLcpos3, an empirical language similarity measure, designed and tuned for source parser weighting in multi-source delexicalized parser transfer. And finally, we introduce a novel resource combination method, based on interpolation of trained parser models.
|
Parsing Natural Language Sentences by Semi-supervised Methods
| 1,943
|
Low-frequency words place a major challenge for automatic speech recognition (ASR). The probabilities of these words, which are often important name entities, are generally under-estimated by the language model (LM) due to their limited occurrences in the training data. Recently, we proposed a word-pair approach to deal with the problem, which borrows information of frequent words to enhance the probabilities of low-frequency words. This paper presents an extension to the word-pair method by involving multiple `predicting words' to produce better estimation for low-frequency words. We also employ this approach to deal with out-of-language words in the task of multi-lingual speech recognition.
|
Recognize Foreign Low-Frequency Words with Similar Pairs
| 1,944
|
Data-driven representation learning for words is a technique of central importance in NLP. While indisputably useful as a source of features in downstream tasks, such vectors tend to consist of uninterpretable components whose relationship to the categories of traditional lexical semantic theories is tenuous at best. We present a method for constructing interpretable word vectors from hand-crafted linguistic resources like WordNet, FrameNet etc. These vectors are binary (i.e, contain only 0 and 1) and are 99.9% sparse. We analyze their performance on state-of-the-art evaluation methods for distributional models of word vectors and find they are competitive to standard distributional approaches.
|
Non-distributional Word Vector Representations
| 1,945
|
In the quest to give a formal compositional semantics to natural languages, semanticists have started turning their attention to phenomena that have been also considered as parts of pragmatics (e.g., discourse anaphora and presupposition projection). To account for these phenomena, the very kinds of meanings assigned to words and phrases are often revisited. To be more specific, in the prevalent paradigm of modeling natural language denotations using the simply-typed lambda calculus (higher-order logic) this means revisiting the types of denotations assigned to individual parts of speech. However, the lambda calculus also serves as a fundamental theory of computation, and in the study of computation, similar type shifts have been employed to give a meaning to side effects. Side effects in programming languages correspond to actions that go beyond the lexical scope of an expression (a thrown exception might propagate throughout a program, a variable modified at one point might later be read at an another) or even beyond the scope of the program itself (a program might interact with the outside world by e.g., printing documents, making sounds, operating robotic limbs...).
|
Pragmatic Side Effects
| 1,946
|
Recent years have witnessed the increase of competition in science. While promoting the quality of research in many cases, an intense competition among scientists can also trigger unethical scientific behaviors. To increase the total number of published papers, some authors even resort to software tools that are able to produce grammatical, but meaningless scientific manuscripts. Because automatically generated papers can be misunderstood as real papers, it becomes of paramount importance to develop means to identify these scientific frauds. In this paper, I devise a methodology to distinguish real manuscripts from those generated with SCIGen, an automatic paper generator. Upon modeling texts as complex networks (CN), it was possible to discriminate real from fake papers with at least 89\% of accuracy. A systematic analysis of features relevance revealed that the accessibility and betweenness were useful in particular cases, even though the relevance depended upon the dataset. The successful application of the methods described here show, as a proof of principle, that network features can be used to identify scientific gibberish papers. In addition, the CN-based approach can be combined in a straightforward fashion with traditional statistical language processing methods to improve the performance in identifying artificially generated papers.
|
Comparing the writing style of real and artificial papers
| 1,947
|
Recently, there has been a lot of effort to represent words in continuous vector spaces. Those representations have been shown to capture both semantic and syntactic information about words. However, distributed representations of phrases remain a challenge. We introduce a novel model that jointly learns word vector representations and their summation. Word representations are learnt using the word co-occurrence statistical information. To embed sequences of words (i.e. phrases) with different sizes into a common semantic space, we propose to average word vector representations. In contrast with previous methods which reported a posteriori some compositionality aspects by simple summation, we simultaneously train words to sum, while keeping the maximum information from the original vectors. We evaluate the quality of the word representations on several classical word evaluation tasks, and we introduce a novel task to evaluate the quality of the phrase representations. While our distributed representations compete with other methods of learning word representations on word evaluations, we show that they give better performance on the phrase evaluation. Such representations of phrases could be interesting for many tasks in natural language processing.
|
"The Sum of Its Parts": Joint Learning of Word and Phrase
Representations with Autoencoders
| 1,948
|
Conversational modeling is an important task in natural language understanding and machine intelligence. Although previous approaches exist, they are often restricted to specific domains (e.g., booking an airline ticket) and require hand-crafted rules. In this paper, we present a simple approach for this task which uses the recently proposed sequence to sequence framework. Our model converses by predicting the next sentence given the previous sentence or sentences in a conversation. The strength of our model is that it can be trained end-to-end and thus requires much fewer hand-crafted rules. We find that this straightforward model can generate simple conversations given a large conversational training dataset. Our preliminary results suggest that, despite optimizing the wrong objective function, the model is able to converse well. It is able extract knowledge from both a domain specific dataset, and from a large, noisy, and general domain dataset of movie subtitles. On a domain-specific IT helpdesk dataset, the model can find a solution to a technical problem via conversations. On a noisy open-domain movie transcript dataset, the model can perform simple forms of common sense reasoning. As expected, we also find that the lack of consistency is a common failure mode of our model.
|
A Neural Conversational Model
| 1,949
|
We present structured perceptron training for neural network transition-based dependency parsing. We learn the neural network representation using a gold corpus augmented by a large number of automatically parsed sentences. Given this fixed network representation, we learn a final layer using the structured perceptron with beam-search decoding. On the Penn Treebank, our parser reaches 94.26% unlabeled and 92.41% labeled attachment accuracy, which to our knowledge is the best accuracy on Stanford Dependencies to date. We also provide in-depth ablative analysis to determine which aspects of our model provide the largest gains in accuracy.
|
Structured Training for Neural Network Transition-Based Parsing
| 1,950
|
We introduce Discriminative BLEU (deltaBLEU), a novel metric for intrinsic evaluation of generated text in tasks that admit a diverse range of possible outputs. Reference strings are scored for quality by human raters on a scale of [-1, +1] to weight multi-reference BLEU. In tasks involving generation of conversational responses, deltaBLEU correlates reasonably with human judgments and outperforms sentence-level and IBM BLEU in terms of both Spearman's rho and Kendall's tau.
|
deltaBLEU: A Discriminative Metric for Generation Tasks with
Intrinsically Diverse Targets
| 1,951
|
It is the most effective way for quick translation of tremendous amount of explosively increasing science and technique information material to develop a practicable machine translation system and introduce it into translation practice. This essay treats problems arising from translation of isolated units on the basis of the practical materials and experiments obtained in the development and introduction of English-Korean machine translation system. In other words, this essay considers establishment of information for isolated units and their Korean equivalents and word order.
|
New Approach to translation of Isolated Units in English-Korean Machine
Translation
| 1,952
|
This paper presents the MAXQ approach to hierarchical reinforcement learning based on decomposing the target Markov decision process (MDP) into a hierarchy of smaller MDPs and decomposing the value function of the target MDP into an additive combination of the value functions of the smaller MDPs. The paper defines the MAXQ hierarchy, proves formal results on its representational power, and establishes five conditions for the safe use of state abstractions. The paper presents an online model-free learning algorithm, MAXQ-Q, and proves that it converges wih probability 1 to a kind of locally-optimal policy known as a recursively optimal policy, even in the presence of the five kinds of state abstraction. The paper evaluates the MAXQ representation and MAXQ-Q through a series of experiments in three domains and shows experimentally that MAXQ-Q (with state abstractions) converges to a recursively optimal policy much faster than flat Q learning. The fact that MAXQ learns a representation of the value function has an important benefit: it makes it possible to compute and execute an improved, non-hierarchical policy via a procedure similar to the policy improvement step of policy iteration. The paper demonstrates the effectiveness of this non-hierarchical execution experimentally. Finally, the paper concludes with a comparison to related work and a discussion of the design tradeoffs in hierarchical reinforcement learning.
|
Hierarchical Reinforcement Learning with the MAXQ Value Function
Decomposition
| 1,953
|
Many researchers have explored methods for hierarchical reinforcement learning (RL) with temporal abstractions, in which abstract actions are defined that can perform many primitive actions before terminating. However, little is known about learning with state abstractions, in which aspects of the state space are ignored. In previous work, we developed the MAXQ method for hierarchical RL. In this paper, we define five conditions under which state abstraction can be combined with the MAXQ value function decomposition. We prove that the MAXQ-Q learning algorithm converges under these conditions and show experimentally that state abstraction is important for the successful application of MAXQ-Q learning.
|
State Abstraction in MAXQ Hierarchical Reinforcement Learning
| 1,954
|
The multiplicative Newton-like method developed by the author et al. is extended to the situation where the dynamics is restricted to the orthogonal group. A general framework is constructed without specifying the cost function. Though the restriction to the orthogonal groups makes the problem somewhat complicated, an explicit expression for the amount of individual jumps is obtained. This algorithm is exactly second-order-convergent. The global instability inherent in the Newton method is remedied by a Levenberg-Marquardt-type variation. The method thus constructed can readily be applied to the independent component analysis. Its remarkable performance is illustrated by a numerical simulation.
|
Multiplicative Algorithm for Orthgonal Groups and Independent Component
Analysis
| 1,955
|
We construct new algorithms from scratch, which use the fourth order cumulant of stochastic variables for the cost function. The multiplicative updating rule here constructed is natural from the homogeneous nature of the Lie group and has numerous merits for the rigorous treatment of the dynamics. As one consequence, the second order convergence is shown. For the cost function, functions invariant under the componentwise scaling are choosen. By identifying points which can be transformed to each other by the scaling, we assume that the dynamics is in a coset space. In our method, a point can move toward any direction in this coset. Thus, no prewhitening is required.
|
Multiplicative Nonholonomic/Newton -like Algorithm
| 1,956
|
Given a reference computer, Kolmogorov complexity is a well defined function on all binary strings. In the standard approach, however, only the asymptotic properties of such functions are considered because they do not depend on the reference computer. We argue that this approach can be more useful if it is refined to include an important practical case of simple binary strings. Kolmogorov complexity calculus may be developed for this case if we restrict the class of available reference computers. The interesting problem is to define a class of computers which is restricted in a {\it natural} way modeling the real-life situation where only a limited class of computers is physically available to us. We give an example of what such a natural restriction might look like mathematically, and show that under such restrictions some error terms, even logarithmic in complexity, can disappear from the standard complexity calculus. Keywords: Kolmogorov complexity; Algorithmic information theory.
|
Complexity analysis for algorithmically simple strings
| 1,957
|
In real-world environments it usually is difficult to specify target operating conditions precisely, for example, target misclassification costs. This uncertainty makes building robust classification systems problematic. We show that it is possible to build a hybrid classifier that will perform at least as well as the best available classifier for any target conditions. In some cases, the performance of the hybrid actually can surpass that of the best known classifier. This robust performance extends across a wide variety of comparison frameworks, including the optimization of metrics such as accuracy, expected cost, lift, precision, recall, and workforce utilization. The hybrid also is efficient to build, to store, and to update. The hybrid is based on a method for the comparison of classifier performance that is robust to imprecise class distributions and misclassification costs. The ROC convex hull (ROCCH) method combines techniques from ROC analysis, decision analysis and computational geometry, and adapts them to the particulars of analyzing learned classifiers. The method is efficient and incremental, minimizes the management of classifier performance data, and allows for clear visual comparisons and sensitivity analyses. Finally, we point to empirical evidence that a robust hybrid classifier indeed is needed for many real-world problems.
|
Robust Classification for Imprecise Environments
| 1,958
|
An approach to clustering is presented that adapts the basic top-down induction of decision trees method towards clustering. To this aim, it employs the principles of instance based learning. The resulting methodology is implemented in the TIC (Top down Induction of Clustering trees) system for first order clustering. The TIC system employs the first order logical decision tree representation of the inductive logic programming system Tilde. Various experiments with TIC are presented, in both propositional and relational domains.
|
Top-down induction of clustering trees
| 1,959
|
When comparing inductive logic programming (ILP) and attribute-value learning techniques, there is a trade-off between expressive power and efficiency. Inductive logic programming techniques are typically more expressive but also less efficient. Therefore, the data sets handled by current inductive logic programming systems are small according to general standards within the data mining community. The main source of inefficiency lies in the assumption that several examples may be related to each other, so they cannot be handled independently. Within the learning from interpretations framework for inductive logic programming this assumption is unnecessary, which allows to scale up existing ILP algorithms. In this paper we explain this learning setting in the context of relational databases. We relate the setting to propositional data mining and to the classical ILP setting, and show that learning from interpretations corresponds to learning from multiple relations and thus extends the expressiveness of propositional learning, while maintaining its efficiency to a large extent (which is not the case in the classical ILP setting). As a case study, we present two alternative implementations of the ILP system Tilde (Top-down Induction of Logical DEcision trees): Tilde-classic, which loads all data in main memory, and Tilde-LDS, which loads the examples one by one. We experimentally compare the implementations, showing Tilde-LDS can handle large data sets (in the order of 100,000 examples or 100 MB) and indeed scales up linearly in the number of examples.
|
Scaling Up Inductive Logic Programming by Learning from Interpretations
| 1,960
|
In order for an agent to perform well in partially observable domains, it is usually necessary for actions to depend on the history of observations. In this paper, we explore a {\it stigmergic} approach, in which the agent's actions include the ability to set and clear bits in an external memory, and the external memory is included as part of the input to the agent. In this case, we need to learn a reactive policy in a highly non-Markovian domain. We explore two algorithms: SARSA(\lambda), which has had empirical success in partially observable domains, and VAPS, a new algorithm due to Baird and Moore, with convergence guarantees in partially observable domains. We compare the performance of these two algorithms on benchmark problems.
|
Learning Policies with External Memory
| 1,961
|
Cross-validation is a useful and generally applicable technique often employed in machine learning, including decision tree induction. An important disadvantage of straightforward implementation of the technique is its computational overhead. In this paper we show that, for decision trees, the computational overhead of cross-validation can be reduced significantly by integrating the cross-validation with the normal decision tree induction process. We discuss how existing decision tree algorithms can be adapted to this aim, and provide an analysis of the speedups these adaptations may yield. The analysis is supported by experimental results.
|
Efficient algorithms for decision tree cross-validation
| 1,962
|
The Markov Blanket Bayesian Classifier is a recently-proposed algorithm for construction of probabilistic classifiers. This paper presents an empirical comparison of the MBBC algorithm with three other Bayesian classifiers: Naive Bayes, Tree-Augmented Naive Bayes and a general Bayesian network. All of these are implemented using the K2 framework of Cooper and Herskovits. The classifiers are compared in terms of their performance (using simple accuracy measures and ROC curves) and speed, on a range of standard benchmark data sets. It is concluded that MBBC is competitive in terms of speed and accuracy with the other algorithms considered.
|
Evaluation of the Performance of the Markov Blanket Bayesian Classifier
Algorithm
| 1,963
|
In biological data, it is often the case that observed data are available only for a subset of samples. When a kernel matrix is derived from such data, we have to leave the entries for unavailable samples as missing. In this paper, we make use of a parametric model of kernel matrices, and estimate missing entries by fitting the model to existing entries. The parametric model is created as a set of spectral variants of a complete kernel matrix derived from another information source. For model fitting, we adopt the em algorithm based on the information geometry of positive definite matrices. We will report promising results on bacteria clustering experiments using two marker sequences: 16S and gyrB.
|
Approximating Incomplete Kernel Matrices by the em Algorithm
| 1,964
|
To learn (statistical) dependencies among random variables requires exponentially large sample size in the number of observed random variables if any arbitrary joint probability distribution can occur. We consider the case that sparse data strongly suggest that the probabilities can be described by a simple Bayesian network, i.e., by a graph with small in-degree \Delta. Then this simple law will also explain further data with high confidence. This is shown by calculating bounds on the VC dimension of the set of those probability measures that correspond to simple graphs. This allows to select networks by structural risk minimization and gives reliability bounds on the error of the estimated joint measure without (in contrast to a previous paper) any prior assumptions on the set of possible joint measures. The complexity for searching the optimal Bayesian networks of in-degree \Delta increases only polynomially in the number of random varibales for constant \Delta and the optimal joint measure associated with a given graph can be found by convex optimization.
|
Reliable and Efficient Inference of Bayesian Networks from Sparse Data
by Statistical Learning Theory
| 1,965
|
We make progress on two important problems regarding attribute efficient learnability. First, we give an algorithm for learning decision lists of length $k$ over $n$ variables using $2^{\tilde{O}(k^{1/3})} \log n$ examples and time $n^{\tilde{O}(k^{1/3})}$. This is the first algorithm for learning decision lists that has both subexponential sample complexity and subexponential running time in the relevant parameters. Our approach establishes a relationship between attribute efficient learning and polynomial threshold functions and is based on a new construction of low degree, low weight polynomial threshold functions for decision lists. For a wide range of parameters our construction matches a 1994 lower bound due to Beigel for the ODDMAXBIT predicate and gives an essentially optimal tradeoff between polynomial threshold function degree and weight. Second, we give an algorithm for learning an unknown parity function on $k$ out of $n$ variables using $O(n^{1-1/k})$ examples in time polynomial in $n$. For $k=o(\log n)$ this yields a polynomial time algorithm with sample complexity $o(n)$. This is the first polynomial time algorithm for learning parity on a superconstant number of variables with sublinear sample complexity.
|
Toward Attribute Efficient Learning Algorithms
| 1,966
|
Using naive Bayes for email classification has become very popular within the last few months. They are quite easy to implement and very efficient. In this paper we want to present empirical results of email classification using a combination of naive Bayes and k-nearest neighbor searches. Using this technique we show that the accuracy of a Bayes filter can be improved slightly for a high number of features and significantly for a small number of features.
|
Improving spam filtering by combining Naive Bayes with simple k-nearest
neighbor searches
| 1,967
|
It is offered to pool test points of different subjects and different aspects of the same subject together in order to get the unitary rating score, by the way of nonlinear transformation of indicator points in accordance with Zipf's distribution. It is proposed to use the well-studied distribution of Intellectuality Quotient IQ as the reference distribution for latent variable "progress in studies".
|
About Unitary Rating Score Constructing
| 1,968
|
For the last years, time-series mining has become a challenging issue for researchers. An important application lies in most monitoring purposes, which require analyzing large sets of time-series for learning usual patterns. Any deviation from this learned profile is then considered as an unexpected situation. Moreover, complex applications may involve the temporal study of several heterogeneous parameters. In that paper, we propose a method for mining heterogeneous multivariate time-series for learning meaningful patterns. The proposed approach allows for mixed time-series -- containing both pattern and non-pattern data -- such as for imprecise matches, outliers, stretching and global translating of patterns instances in time. We present the early results of our approach in the context of monitoring the health status of a person at home. The purpose is to build a behavioral profile of a person by analyzing the time variations of several quantitative or qualitative parameters recorded through a provision of sensors installed in the home.
|
Mining Heterogeneous Multivariate Time-Series for Learning Meaningful
Patterns: Application to Home Health Telecare
| 1,969
|
We discuss stability for a class of learning algorithms with respect to noisy labels. The algorithms we consider are for regression, and they involve the minimization of regularized risk functionals, such as L(f) := 1/N sum_i (f(x_i)-y_i)^2+ lambda ||f||_H^2. We shall call the algorithm `stable' if, when y_i is a noisy version of f*(x_i) for some function f* in H, the output of the algorithm converges to f* as the regularization term and noise simultaneously vanish. We consider two flavors of this problem, one where a data set of N points remains fixed, and the other where N -> infinity. For the case where N -> infinity, we give conditions for convergence to f_E (the function which is the expectation of y(x) for each x), as lambda -> 0. For the fixed N case, we describe the limiting 'non-noisy', 'non-regularized' function f*, and give conditions for convergence. In the process, we develop a set of tools for dealing with functionals such as L(f), which are applicable to many other problems in learning theory.
|
Stability Analysis for Regularized Least Squares Regression
| 1,970
|
We consider the probability hierarchy for Popperian FINite learning and study the general properties of this hierarchy. We prove that the probability hierarchy is decidable, i.e. there exists an algorithm that receives p_1 and p_2 and answers whether PFIN-type learning with the probability of success p_1 is equivalent to PFIN-type learning with the probability of success p_2. To prove our result, we analyze the topological structure of the probability hierarchy. We prove that it is well-ordered in descending ordering and order-equivalent to ordinal epsilon_0. This shows that the structure of the hierarchy is very complicated. Using similar methods, we also prove that, for PFIN-type learning, team learning and probabilistic learning are of the same power.
|
Probabilistic and Team PFIN-type Learning: General Properties
| 1,971
|
We analyze a new algorithm for probability forecasting of binary observations on the basis of the available data, without making any assumptions about the way the observations are generated. The algorithm is shown to be well calibrated and to have good resolution for long enough sequences of observations and for a suitable choice of its parameter, a kernel on the Cartesian product of the forecast space $[0,1]$ and the data space. Our main results are non-asymptotic: we establish explicit inequalities, shown to be tight, for the performance of the algorithm.
|
Non-asymptotic calibration and resolution
| 1,972
|
We consider a general class of forecasting protocols, called "linear protocols", and discuss several important special cases, including multi-class forecasting. Forecasting is formalized as a game between three players: Reality, whose role is to generate observations; Forecaster, whose goal is to predict the observations; and Skeptic, who tries to make money on any lack of agreement between Forecaster's predictions and the actual observations. Our main mathematical result is that for any continuous strategy for Skeptic in a linear protocol there exists a strategy for Forecaster that does not allow Skeptic's capital to grow. This result is a meta-theorem that allows one to transform any continuous law of probability in a linear protocol into a forecasting strategy whose predictions are guaranteed to satisfy this law. We apply this meta-theorem to a weak law of large numbers in Hilbert spaces to obtain a version of the K29 prediction algorithm for linear protocols and show that this version also satisfies the attractive properties of proper calibration and resolution under a suitable choice of its kernel parameter, with no assumptions about the way the data is generated.
|
Defensive forecasting for linear protocols
| 1,973
|
We propose a new framework for building and evaluating machine learning algorithms. We argue that many real-world problems require an agent which must quickly learn to respond to demands, yet can continue to perform and respond to new training throughout its useful life. We give a framework for how such agents can be built, describe several metrics for evaluating them, and show that subtle changes in system construction can significantly affect agent performance.
|
On the Job Training
| 1,974
|
We present in this work a new methodology to design kernels on data which is structured with smaller components, such as text, images or sequences. This methodology is a template procedure which can be applied on most kernels on measures and takes advantage of a more detailed "bag of components" representation of the objects. To obtain such a detailed description, we consider possible decompositions of the original bag into a collection of nested bags, following a prior knowledge on the objects' structure. We then consider these smaller bags to compare two objects both in a detailed perspective, stressing local matches between the smaller bags, and in a global or coarse perspective, by considering the entire bag. This multiresolution approach is likely to be best suited for tasks where the coarse approach is not precise enough, and where a more subtle mixture of both local and global similarities is necessary to compare objects. The approach presented here would not be computationally tractable without a factorization trick that we introduce before presenting promising results on an image retrieval task.
|
Multiresolution Kernels
| 1,975
|
This paper shows how universal learning can be achieved with expert advice. To this aim, we specify an experts algorithm with the following characteristics: (a) it uses only feedback from the actions actually chosen (bandit setup), (b) it can be applied with countably infinite expert classes, and (c) it copes with losses that may grow in time appropriately slowly. We prove loss bounds against an adaptive adversary. From this, we obtain a master algorithm for "reactive" experts problems, which means that the master's actions may influence the behavior of the adversary. Our algorithm can significantly outperform standard experts algorithms on such problems. Finally, we combine it with a universal expert class. The resulting universal learner performs -- in a certain sense -- almost as well as any computable strategy, for any online decision problem. We also specify the (worst-case) convergence speed, which is very slow.
|
Defensive Universal Learning with Experts
| 1,976
|
A main problem of "Follow the Perturbed Leader" strategies for online decision problems is that regret bounds are typically proven against oblivious adversary. In partial observation cases, it was not clear how to obtain performance guarantees against adaptive adversary, without worsening the bounds. We propose a conceptually simple argument to resolve this problem. Using this, a regret bound of O(t^(2/3)) for FPL in the adversarial multi-armed bandit problem is shown. This bound holds for the common FPL variant using only the observations from designated exploration rounds. Using all observations allows for the stronger bound of O(t^(1/2)), matching the best bound known so far (and essentially the known lower bound) for adversarial bandits. Surprisingly, this variant does not even need explicit exploration, it is self-stabilizing. However the sampling probabilities have to be either externally provided or approximated to sufficient accuracy, using O(t^2 log t) samples in each step.
|
FPL Analysis for Adaptive Bandits
| 1,977
|
Naive Bayes is a simple Bayesian classifier with strong independence assumptions among the attributes. This classifier, desipte its strong independence assumptions, often performs well in practice. It is believed that relaxing the independence assumptions of a naive Bayes classifier may improve the classification accuracy of the resulting structure. While finding an optimal unconstrained Bayesian Network (for most any reasonable scoring measure) is an NP-hard problem, it is possible to learn in polynomial time optimal networks obeying various structural restrictions. Several authors have examined the possibilities of adding augmenting arcs between attributes of a Naive Bayes classifier. Friedman, Geiger and Goldszmidt define the TAN structure in which the augmenting arcs form a tree on the attributes, and present a polynomial time algorithm that learns an optimal TAN with respect to MDL score. Keogh and Pazzani define Augmented Bayes Networks in which the augmenting arcs form a forest on the attributes (a collection of trees, hence a relaxation of the stuctural restriction of TAN), and present heuristic search methods for learning good, though not optimal, augmenting arc sets. The authors, however, evaluate the learned structure only in terms of observed misclassification error and not against a scoring metric, such as MDL. In this paper, we present a simple, polynomial time greedy algorithm for learning an optimal Augmented Bayes Network with respect to MDL score.
|
Learning Optimal Augmented Bayes Networks
| 1,978
|
We consider the problem of learning unions of rectangles over the domain $[b]^n$, in the uniform distribution membership query learning setting, where both b and n are "large". We obtain poly$(n, \log b)$-time algorithms for the following classes: - poly$(n \log b)$-way Majority of $O(\frac{\log(n \log b)} {\log \log(n \log b)})$-dimensional rectangles. - Union of poly$(\log(n \log b))$ many $O(\frac{\log^2 (n \log b)} {(\log \log(n \log b) \log \log \log (n \log b))^2})$-dimensional rectangles. - poly$(n \log b)$-way Majority of poly$(n \log b)$-Or of disjoint $O(\frac{\log(n \log b)} {\log \log(n \log b)})$-dimensional rectangles. Our main algorithmic tool is an extension of Jackson's boosting- and Fourier-based Harmonic Sieve algorithm [Jackson 1997] to the domain $[b]^n$, building on work of [Akavia, Goldwasser, Safra 2003]. Other ingredients used to obtain the results stated above are techniques from exact learning [Beimel, Kushilevitz 1998] and ideas from recent work on learning augmented $AC^{0}$ circuits [Jackson, Klivans, Servedio 2002] and on representing Boolean functions as thresholds of parities [Klivans, Servedio 2001].
|
Learning Unions of $ω(1)$-Dimensional Rectangles
| 1,979
|
We consider the problem of on-line prediction of real-valued labels, assumed bounded in absolute value by a known constant, of new objects from known labeled objects. The prediction algorithm's performance is measured by the squared deviation of the predictions from the actual labels. No stochastic assumptions are made about the way the labels and objects are generated. Instead, we are given a benchmark class of prediction rules some of which are hoped to produce good predictions. We show that for a wide range of infinite-dimensional benchmark classes one can construct a prediction algorithm whose cumulative loss over the first N examples does not exceed the cumulative loss of any prediction rule in the class plus O(sqrt(N)); the main differences from the known results are that we do not impose any upper bound on the norm of the considered prediction rules and that we achieve an optimal leading term in the excess loss of our algorithm. If the benchmark class is "universal" (dense in the class of continuous functions on each compact set), this provides an on-line non-stochastic analogue of universally consistent prediction in non-parametric statistics. We use two proof techniques: one is based on the Aggregating Algorithm and the other on the recently developed method of defensive forecasting.
|
On-line regression competitive with reproducing kernel Hilbert spaces
| 1,980
|
The problem of finding an optimum using noisy evaluations of a smooth cost function arises in many contexts, including economics, business, medicine, experiment design, and foraging theory. We derive an asymptotic bound E[ (x_t - x*)^2 ] >= O(1/sqrt(t)) on the rate of convergence of a sequence (x_0, x_1, >...) generated by an unbiased feedback process observing noisy evaluations of an unknown quadratic function maximised at x*. The bound is tight, as the proof leads to a simple algorithm which meets it. We further establish a bound on the total regret, E[ sum_{i=1..t} (x_i - x*)^2 ] >= O(sqrt(t)) These bounds may impose practical limitations on an agent's performance, as O(eps^-4) queries are made before the queries converge to x* with eps accuracy.
|
Bounds on Query Convergence
| 1,981
|
A key data preparation step in Text Mining, Term Extraction selects the terms, or collocation of words, attached to specific concepts. In this paper, the task of extracting relevant collocations is achieved through a supervised learning algorithm, exploiting a few collocations manually labelled as relevant/irrelevant. The candidate terms are described along 13 standard statistical criteria measures. From these examples, an evolutionary learning algorithm termed Roger, based on the optimization of the Area under the ROC curve criterion, extracts an order on the candidate terms. The robustness of the approach is demonstrated on two real-world domain applications, considering different domains (biology and human resources) and different languages (English and French).
|
Preference Learning in Terminology Extraction: A ROC-based approach
| 1,982
|
We consider the problem of on-line prediction competitive with a benchmark class of continuous but highly irregular prediction rules. It is known that if the benchmark class is a reproducing kernel Hilbert space, there exists a prediction algorithm whose average loss over the first N examples does not exceed the average loss of any prediction rule in the class plus a "regret term" of O(N^(-1/2)). The elements of some natural benchmark classes, however, are so irregular that these classes are not Hilbert spaces. In this paper we develop Banach-space methods to construct a prediction algorithm with a regret term of O(N^(-1/p)), where p is in [2,infty) and p-2 reflects the degree to which the benchmark class fails to be a Hilbert space.
|
Competing with wild prediction rules
| 1,983
|
Fitness functions based on test cases are very common in Genetic Programming (GP). This process can be assimilated to a learning task, with the inference of models from a limited number of samples. This paper is an investigation on two methods to improve generalization in GP-based learning: 1) the selection of the best-of-run individuals using a three data sets methodology, and 2) the application of parsimony pressure in order to reduce the complexity of the solutions. Results using GP in a binary classification setup show that while the accuracy on the test sets is preserved, with less variances compared to baseline results, the mean tree size obtained with the tested methods is significantly reduced.
|
Genetic Programming, Validation Sets, and Parsimony Pressure
| 1,984
|
It is suggested to insert into test matrix 1s for correct responses, 0s for response refusals, and negative corrective elements for incorrect responses. With the classical test theory approach test scores of examinees and items are calculated traditionally as sums of matrix elements, organized in rows and columns. Correlation coefficients are estimated using correction coefficients. In item response theory approach examinee and item logits are estimated using maximum likelihood method and probabilities of all matrix elements.
|
Processing of Test Matrices with Guessing Correction
| 1,985
|
Given a finite set of words w1,...,wn independently drawn according to a fixed unknown distribution law P called a stochastic language, an usual goal in Grammatical Inference is to infer an estimate of P in some class of probabilistic models, such as Probabilistic Automata (PA). Here, we study the class of rational stochastic languages, which consists in stochastic languages that can be generated by Multiplicity Automata (MA) and which strictly includes the class of stochastic languages generated by PA. Rational stochastic languages have minimal normal representation which may be very concise, and whose parameters can be efficiently estimated from stochastic samples. We design an efficient inference algorithm DEES which aims at building a minimal normal representation of the target. Despite the fact that no recursively enumerable class of MA computes exactly the set of rational stochastic languages over Q, we show that DEES strongly identifies tis set in the limit. We study the intermediary MA output by DEES and show that they compute rational series which converge absolutely to one and which can be used to provide stochastic languages which closely estimate the target.
|
Learning rational stochastic languages
| 1,986
|
Consider an agent interacting with an environment in cycles. In every interaction cycle the agent is rewarded for its performance. We compare the average reward U from cycle 1 to m (average value) with the future discounted reward V from cycle k to infinity (discounted value). We consider essentially arbitrary (non-geometric) discount sequences and arbitrary reward sequences (non-MDP environments). We show that asymptotically U for m->infinity and V for k->infinity are equal, provided both limits exist. Further, if the effective horizon grows linearly with k or faster, then existence of the limit of U implies that the limit of V exists. Conversely, if the effective horizon grows linearly with k or slower, then existence of the limit of V implies that the limit of U exists.
|
General Discounting versus Average Reward
| 1,987
|
Suppose we are given two probability measures on the set of one-way infinite finite-alphabet sequences and consider the question when one of the measures predicts the other, that is, when conditional probabilities converge (in a certain sense) when one of the measures is chosen to generate the sequence. This question may be considered a refinement of the problem of sequence prediction in its most general formulation: for a given class of probability measures, does there exist a measure which predicts all of the measures in the class? To address this problem, we find some conditions on local absolute continuity which are sufficient for prediction and which generalize several different notions which are known to be sufficient for prediction. We also formulate some open questions to outline a direction for finding the conditions on classes of measures for which prediction is possible.
|
On Sequence Prediction for Arbitrary Measures
| 1,988
|
Prediction is a complex notion, and different predictors (such as people, computer programs, and probabilistic theories) can pursue very different goals. In this paper I will review some popular kinds of prediction and argue that the theory of competitive on-line learning can benefit from the kinds of prediction that are now foreign to it.
|
Predictions as statements and decisions
| 1,989
|
A standard approach in pattern classification is to estimate the distributions of the label classes, and then to apply the Bayes classifier to the estimates of the distributions in order to classify unlabeled examples. As one might expect, the better our estimates of the label class distributions, the better the resulting classifier will be. In this paper we make this observation precise by identifying risk bounds of a classifier in terms of the quality of the estimates of the label class distributions. We show how PAC learnability relates to estimates of the distributions that have a PAC guarantee on their $L_1$ distance from the true distribution, and we bound the increase in negative log likelihood risk in terms of PAC bounds on the KL-divergence. We give an inefficient but general-purpose smoothing method for converting an estimated distribution that is good under the $L_1$ metric into a distribution that is good under the KL-divergence.
|
PAC Classification based on PAC Estimates of Label Class Distributions
| 1,990
|
In this paper we introduce the class of stationary prediction strategies and construct a prediction algorithm that asymptotically performs as well as the best continuous stationary strategy. We make mild compactness assumptions but no stochastic assumptions about the environment. In particular, no assumption of stationarity is made about the environment, and the stationarity of the considered strategies only means that they do not depend explicitly on time; we argue that it is natural to consider only stationary strategies even for highly non-stationary environments.
|
Competing with stationary prediction strategies
| 1,991
|
In probabilistic grammatical inference, a usual goal is to infer a good approximation of an unknown distribution P called a stochastic language. The estimate of P stands in some class of probabilistic models such as probabilistic automata (PA). In this paper, we focus on probabilistic models based on multiplicity automata (MA). The stochastic languages generated by MA are called rational stochastic languages; they strictly include stochastic languages generated by PA; they also admit a very concise canonical representation. Despite the fact that this class is not recursively enumerable, it is efficiently identifiable in the limit by using the algorithm DEES, introduced by the authors in a previous paper. However, the identification is not proper and before the convergence of the algorithm, DEES can produce MA that do not define stochastic languages. Nevertheless, it is possible to use these MA to define stochastic languages. We show that they belong to a broader class of rational series, that we call pseudo-stochastic rational languages. The aim of this paper is twofold. First we provide a theoretical study of pseudo-stochastic rational languages, the languages output by DEES, showing for example that this class is decidable within polynomial time. Second, we have carried out a lot of experiments in order to compare DEES to classical inference algorithms such as ALERGIA and MDI. They show that DEES outperforms them in most cases.
|
Using Pseudo-Stochastic Rational Languages in Probabilistic Grammatical
Inference
| 1,992
|
We investigate here concept learning from incomplete examples. Our first purpose is to discuss to what extent logical learning settings have to be modified in order to cope with data incompleteness. More precisely we are interested in extending the learning from interpretations setting introduced by L. De Raedt that extends to relational representations the classical propositional (or attribute-value) concept learning from examples framework. We are inspired here by ideas presented by H. Hirsh in a work extending the Version space inductive paradigm to incomplete data. H. Hirsh proposes to slightly modify the notion of solution when dealing with incomplete examples: a solution has to be a hypothesis compatible with all pieces of information concerning the examples. We identify two main classes of incompleteness. First, uncertainty deals with our state of knowledge concerning an example. Second, generalization (or abstraction) deals with what part of the description of the example is sufficient for the learning purpose. These two main sources of incompleteness can be mixed up when only part of the useful information is known. We discuss a general learning setting, referred to as "learning from possibilities" that formalizes these ideas, then we present a more specific learning setting, referred to as "assumption-based learning" that cope with examples which uncertainty can be reduced when considering contextual information outside of the proper description of the examples. Assumption-based learning is illustrated on a recent work concerning the prediction of a consensus secondary structure common to a set of RNA sequences.
|
Logical settings for concept learning from incomplete examples in First
Order Logic
| 1,993
|
We present a theory of boosting probabilistic classifiers. We place ourselves in the situation of a user who only provides a stopping parameter and a probabilistic weak learner/classifier and compare three types of boosting algorithms: probabilistic Adaboost, decision tree, and tree of trees of ... of trees, which we call matryoshka. "Nested tree," "embedded tree" and "recursive tree" are also appropriate names for this algorithm, which is one of our contributions. Our other contribution is the theoretical analysis of the algorithms, in which we give training error bounds. This analysis suggests that the matryoshka leverages probabilistic weak classifiers more efficiently than simple decision trees.
|
A Theory of Probabilistic Boosting, Decision Trees and Matryoshki
| 1,994
|
We start from a simple asymptotic result for the problem of on-line regression with the quadratic loss function: the class of continuous limited-memory prediction strategies admits a "leading prediction strategy", which not only asymptotically performs at least as well as any continuous limited-memory strategy but also satisfies the property that the excess loss of any continuous limited-memory strategy is determined by how closely it imitates the leading strategy. More specifically, for any class of prediction strategies constituting a reproducing kernel Hilbert space we construct a leading strategy, in the sense that the loss of any prediction strategy whose norm is not too large is determined by how closely it imitates the leading strategy. This result is extended to the loss functions given by Bregman divergences and by strictly proper scoring rules.
|
Leading strategies in competitive on-line prediction
| 1,995
|
We present basic notions of Gold's "learnability in the limit" paradigm, first presented in 1967, a formalization of the cognitive process by which a native speaker gets to grasp the underlying grammar of his/her own native language by being exposed to well formed sentences generated by that grammar. Then we present Lambek grammars, a formalism issued from categorial grammars which, although not as expressive as needed for a full formalization of natural languages, is particularly suited to easily implement a natural interface between syntax and semantics. In the last part of this work, we present a learnability result for Rigid Lambek grammars from structured examples.
|
A Study on Learnability for Rigid Lambek Grammars
| 1,996
|
An approach to the classification problem of machine learning, based on building local classification rules, is developed. The local rules are considered as projections of the global classification rules to the event we want to classify. A massive global optimization algorithm is used for optimization of quality criterion. The algorithm, which has polynomial complexity in typical case, is used to find all high--quality local rules. The other distinctive feature of the algorithm is the integration of attributes levels selection (for ordered attributes) with rules searching and original conflicting rules resolution strategy. The algorithm is practical; it was tested on a number of data sets from UCI repository, and a comparison with the other predicting techniques is presented.
|
A Massive Local Rules Search Approach to the Classification Problem
| 1,997
|
Competitive on-line prediction (also known as universal prediction of individual sequences) is a strand of learning theory avoiding making any stochastic assumptions about the way the observations are generated. The predictor's goal is to compete with a benchmark class of prediction rules, which is often a proper Banach function space. Metric entropy provides a unifying framework for competitive on-line prediction: the numerous known upper bounds on the metric entropy of various compact sets in function spaces readily imply bounds on the performance of on-line prediction strategies. This paper discusses strengths and limitations of the direct approach to competitive on-line prediction via metric entropy, including comparisons to other approaches.
|
Metric entropy in competitive on-line prediction
| 1,998
|
We propose and analyze a new vantage point for the learning of mixtures of Gaussians: namely, the PAC-style model of learning probability distributions introduced by Kearns et al. Here the task is to construct a hypothesis mixture of Gaussians that is statistically indistinguishable from the actual mixture generating the data; specifically, the KL-divergence should be at most epsilon. In this scenario, we give a poly(n/epsilon)-time algorithm that learns the class of mixtures of any constant number of axis-aligned Gaussians in n-dimensional Euclidean space. Our algorithm makes no assumptions about the separation between the means of the Gaussians, nor does it have any dependence on the minimum mixing weight. This is in contrast to learning results known in the ``clustering'' model, where such assumptions are unavoidable. Our algorithm relies on the method of moments, and a subalgorithm developed in previous work by the authors (FOCS 2005) for a discrete mixture-learning problem.
|
PAC Learning Mixtures of Axis-Aligned Gaussians with No Separation
Assumption
| 1,999
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.