Correct Hypothesis Research Articles

Phone-based filter parameter optimization of filter and sum robust speech recognition using likelihood maximization

Because of noise and reverberation, accuracy of speech recognition systems decreases when the distance between talker and microphone increases. By the using of microphone arrays and appropriate filtering of received signals, the accuracy of recognizer can be increased. Many different methods for using microphone arrays have been proposed that can be classified into two main approaches: systems that perform in two independent stages of array processing and then recognition and systems that use array processing to generate a sequence of features which maximize the likelihood of generating the correct hypothesis in recognition phase. Following second approach, in this paper a new method for microphone array processing is proposed in which the parameters of array processing are adjusted in calibration phase based on phones used in language and maximum likelihood method. Optimized filter parameters are stored and used during recognition phase. A new modified Viterbi algorithm using optimal phone-based filter parameters is used for recognition phase. The proposed algorithm is analytically formulated and Persian language is used to find any improvement in speech recognition accuracy compared with results of delay and sum and utterance-based filter and sum algorithms. The results show 12.2% improvement in accuracy compared to utterance-based algorithm.

Expected Monotonicity – A Desirable Property for Evidence Measures?

We consider here the principle of ‘evidential consistency’ – that as one gathers more data, any well-behaved evidence measure should, in some sense, approach the true answer. Evidential consistency is essential for the genome-scan design (GWAS or linkage), where one selects the most promising locus(i) for follow-up, expecting that new data will increase evidence for the correct hypothesis. Earlier work [Vieland, Hum Hered 2006;61:144–156] showed that many popular statistics do not satisfy this principle; Vieland concluded that the problem stems from fundamental difficulties in how we measure evidence and argued for determining criteria to evaluate evidence measures. Here, we investigate in detail one proposed consistency criterion – expected monotonicity (ExpM) – for a simple statistical model (binomial) and four likelihood ratio (LR)-based evidence measures. We show that, with one limited exception, none of these measures displays ExpM; what they do display is sometimes counterintuitive. We conclude that ExpM is not a reasonable requirement for evidence measures; moreover, no requirement based on expected values seems feasible. We demonstrate certain desirable properties of the simple LR and demonstrate a connection between the simple and integrated LRs. We also consider an alternative version of consistency, which is satisfied by certain forms of the integrated LR and posterior probability of linkage.

The solution effects of initial fragments on perceptual interference in Chinese characters

A large word fragment (e.g., "r__ i__rop") is harder to solve when disclosed piecemeal from a smaller fragment (e.g., "r_____p") than when shown all at once. This perceptual interference has been found to decrease with the number of solutions to the initial fragments. In the present study, 2 experiments manipulated the number of solutions (1 vs. at least 3) to examine this solution effect using a fragment completion task with Chinese characters. The results revealed a solution effect and showed for the first time no interference in the multiple-solution condition. These findings provide evidence against a competition explanation, which attributes interference to activations of incorrect hypotheses in an early presentation, and in favor of a mismatch explanation, which attributes interference to an enduring inhibition of the correct hypothesis regarding stimulus identity after a mismatch is detected between the hypothesis and the stimulus early in the presentation sequence.

Henry Rose Carter.

Henry Rose Carter.

Cochlea Archimedea: Hitchcock's Fish Trail

A specimen figured by Hitchcock (1858) and named Cochlea archimedea presents interesting ichnotaxonomic and paleoecologic issues. The ichnogenus Cochlea is considered to be a preoccupied taxon and therefore invalid. The specimen in question is given its appropriate name, Undichna isp., based on its apparent morphology and is reinterpreted as a trail made by a fish. Two working hypotheses for the construction of the trail are: (A) as a single epichnial groove with side levees made by one fin in contact with the substrate and (B) as a hypichnial trail with two grooves made by two fins in contact with the substrate. A more specific approach cannot be taken without knowledge of the true toponomy of the trace fossil, as none is indicated on the original specimen. Regardless, the identification of the first trails made by fish in the Early Jurassic of the Newark Supergroup of Massachusetts indicates the need for more nondinosaurian trace fossil research in the Mesozoic rift basins of eastern North America. The most likely tracemaker, depending on the correct constructional hypothesis, is either a semionotid with a deeply set anal fin (hypothesis B) or Redfieldius (hypothesis A). Considering the abundance of fish body fossils preserved in rocks of this age in eastern North America, it is likely that once recognized, more Undichna will be discovered in future field explorations.

A Comprehensive Analysis of the Structure-Function Relationship in Proteins Based on Local Structure Similarity

BackgroundSequence similarity to characterized proteins provides testable functional hypotheses for less than 50% of the proteins identified by genome sequencing projects. With structural genomics it is believed that structural similarities may give functional hypotheses for many of the remaining proteins.Methodology/Principal FindingsWe provide a systematic analysis of the structure-function relationship in proteins using the novel concept of local descriptors of protein structure. A local descriptor is a small substructure of a protein which includes both short- and long-range interactions. We employ a library of commonly reoccurring local descriptors general enough to assemble most existing protein structures. We then model the relationship between these local shapes and Gene Ontology using rule-based learning. Our IF-THEN rule model offers legible, high resolution descriptions that combine local substructures and is able to discriminate functions even for functionally versatile folds such as the frequently occurring TIM barrel and Rossmann fold. By evaluating the predictive performance of the model, we provide a comprehensive quantification of the structure-function relationship based only on local structure similarity. Our findings are, among others, that conserved structure is a stronger prerequisite for enzymatic activity than for binding specificity, and that structure-based predictions complement sequence-based predictions. The model is capable of generating correct hypotheses, as confirmed by a literature study, even when no significant sequence similarity to characterized proteins exists.Conclusions/SignificanceOur approach offers a new and complete description and quantification of the structure-function relationship in proteins. By demonstrating how our predictions offer higher sensitivity than using global structure, and complement the use of sequence, we show that the presented ideas could advance the development of meta-servers in function prediction.

Different likelihood ratio approaches to evaluate the strength of evidence of MDMA tablet comparisons

Two likelihood ratio (LR) approaches are presented to evaluate the strength of evidence of MDMA tablet comparisons. The first one is based on a more ‘traditional’ comparison of MDMA tablets by using distance measures (e.g., Pearson correlation distance or a Euclidean distance). In this approach, LRs are calculated using the distribution of distances between tablets of the same-batch and that of different-batches. The second approach is based on methods used in some other fields of forensic comparison. Here LRs are calculated based on the distribution of values of MDMA tablet characteristics within a specific batch and from all batches. The data used in this paper must be seen as examples to illustrate both methods. In future research the methods can be applied to other and more complex data. In this paper, the methods and their results are discussed, considering their performance in evidence evaluation and several practical aspects. With respect to evidence in favor of the correct hypothesis, the second method proved to be better than the first one. It is shown that the LRs in same-batch comparisons are generally higher compared to the first method and the LRs in different-batch comparisons are generally lower. On the other hand, for operational purposes (where quick information is needed), the first method may be preferred, because it is less time consuming. With this method a model has to be estimated only once in a while, which means that only a few measurements have to be done, while with the second method more measurements are needed because each time a new model has to be estimated.

Discriminative Training of the Hidden Vector State Model for Semantic Parsing

In this paper, we discuss how discriminative training can be applied to the hidden vector state (HVS) model in different task domains. The HVS model is a discrete hidden Markov model (HMM) in which each HMM state represents the state of a push-down automaton with a finite stack size. In previous applications, maximum-likelihood estimation (MLE) is used to derive the parameters of the HVS model. However, MLE makes a number of assumptions and unfortunately some of these assumptions do not hold. Discriminative training, without making such assumptions, can improve the performance of the HVS model by discriminating the correct hypothesis from the competing hypotheses. Experiments have been conducted in two domains: the travel domain for the semantic parsing task using the DARPA Communicator data and the Air Travel Information Services (ATIS) data and the bioinformatics domain for the information extraction task using the GENIA corpus. The results demonstrate modest improvements of the performance of the HVS model using discriminative training. In the travel domain, discriminative training of the HVS model gives a relative error reduction rate of 31 percent in F-measure when compared with MLE on the DARPA Communicator data and 9 percent on the ATIS data. In the bioinformatics domain, a relative error reduction rate of 4 percent in F-measure is achieved on the GENIA corpus.

Evidence with uncertain likelihoods

An agent often has a number of hypotheses, and must choose among them based on observations, or outcomes of experiments. Each of these observations can be viewed as providing evidence for or against various hypotheses. All the attempts to formalize this intuition up to now have assumed that associated with each hypothesis h there is a likelihood function µh, which is a probability measure that intuitively describes how likely each observation is, conditional on h being the correct hypothesis. We consider an extension of this framework where there is uncertainty as to which of a number of likelihood functions is appropriate, and discuss how one formal approach to defining evidence, which views evidence as a function from priors to posteriors, can be generalized to accommodate this uncertainty.

Moral Methodology and the Third Theory of Rights

The paper engages the conceptual question of the nature of rights. First, moral methodology for developing criteria to judge the adequacy of theories for the concept of rights is discussed. Standard methodologies for conceptual theory, such as analysis of language practices, appealing to intuitions to test and correct hypotheses, and mixtures of these with appeals to substantive moral values, are shown to fail in important ways to give us reasons to adopt one or another view of the concept. An alternative methodology is proposed which appeals to the distinct role of concepts in our value scheme (the added value of being able to employ such concepts in our moral judgments). This is then used to develop criteria for a successful theory of rights. Two existing theories are shown, as a matter of illustration, to fail to satisfy the criteria, and finally an alternative theory of rights which does satisfy these criteria is proposed and defended.

Differentiating between two models of motor lateralization.

This study was designed to differentiate between two models of motor lateralization: "feedback corrections" and dynamic dominance. Whereas the feedback correction hypothesis suggests that handedness reflects a dominant hemisphere advantage for visual-mediated correction processes, dynamic dominance proposes that each hemisphere has become specialized for distinct aspects of control. This model suggests that the dominant hemisphere is specialized for controlling task dynamics, as required for coordinating efficient trajectories, and the nondominant hemisphere is specialized for controlling limb impedance, as required for maintaining stable postures. To differentiate between these two models, we examined whether visuomotor corrections are mediated differently for the nondominant and dominant arms. Participants performed targeted reaches in a virtual reality environment in which visuomotor rotations occurred in two directions that elicited corrections with different coordination requirements. The feedback correction model predicts a dominant arm advantage for the timing and accuracy of corrections in both directions. Dynamic dominance predicts that correction timing and accuracy will be similar for both arms, but that interlimb differences in the quality of corrections will depend on the coordination requirements, and thus, direction of corrections. Our results indicated that correction time and accuracy did not depend on arm. However, correction quality, as reflected by trajectory curvature, depended on both arm and rotation direction. Nondominant trajectories were systematically more curvilinear than dominant trajectories for corrections with the highest coordination requirement. These results support the dynamic dominance hypothesis.

Fast speech recognition to access a very large list of items on embedded devices

In this paper, we propose a fast decoding algorithm to recognize a very large number of item names on a resource-limited embedded device. The proposed algorithm is based on a multi-pass search scheme. The algorithm is composed of a two-stage HMM-based coarse match and a detailed match. The two-stage HMM-based coarse match is aimed at rapidly selecting a small set of candidates that are assumed to contain a correct hypothesis with high probability, and the detailed match re-ranks the candidates by performing acoustic rescoring. The proposed algorithm is implemented on an in-car navigation system with a 32-bit fixed-point processor operating at 620 MHz. The experimental result shows that the proposed method runs at maximum speed 1. 74 times real-time on the embedded device while minimizing the degradation of the recognition accuracy for a 220 K Korean Point-of-Interest (POI) recognition domain.

Sequential multi-hypothesis testing for compound Poisson processes

Suppose that there are finitely many simple hypotheses about the unknown arrival rate and mark distribution of a compound Poisson process, and that exactly one of them is correct. The objective is to determine the correct hypothesis with minimal error probability and as soon as possible after the observation of the process starts. This problem is formulated in a Bayesian framework, and its solution is presented. Provably convergent numerical methods and practical near-optimal strategies are described and illustrated on various examples.

The Effect of a Justification Memo and Hypothesis Set Quality on the Review Process

This study provides evidence on the effects of a preparer's supporting justification memo and hypothesis set quality on a reviewer's ability to identify a preparer's incorrect conclusion. We hypothesize that the presence of a supporting justification memo will lead to a significant performance decline for reviewers who evaluate working papers containing a correct hypothesis set (one that includes the correct cause), but not those with an incorrect hypothesis set. This expectation follows because reviewers of the correct hypothesis set can link each hypothesis in the working papers to an underlying piece of evidence while those with the incorrect hypothesis set cannot. Thus, the former are more likely to presume that evidence pointing to the actual cause has been integrated, and hence are more willing to rely on the preparer's supporting justification memo. To test our hypothesis, 97 audit seniors reviewed working papers documenting work done by a preparer while investigating five potential explanations (i.e., hypotheses) of an unexpected fluctuation in a client's gross margin. We manipulated justification (Supporting Justification Memo versus No Justification Memo) and hypothesis set quality (Correct Set versus Incorrect Set). All reviewers received work paper documentation showing the results of six tests and an incorrect preparer conclusion. The dependent variable was whether reviewers agreed or disagreed with the preparer's conclusion. The experimental rsesults are consistent with our hypothesis suggesting that in a correct hypothesis set scenario a supporting justification memo may undermine the effectiveness of the review process. The findings underscore the need for reviewers to first evaluate the underlying audit evidence before looking at the preparer's overall conclusion and justification memo.

When unlearning helps

Overregularization seen in child language learning, for example, verb tense constructs, involves abandoning correct behaviours for incorrect ones and later reverting to correct behaviours. Quite a number of other child development phenomena also follow this U-shaped form of learning, unlearning and relearning. A decisive learner does not do this and, more generally, never abandons an hypothesis H for an inequivalent one where it later conjectures an hypothesis equivalent to H, where equivalence means semantical or behavioural equivalence. The first main result of the present paper entails that decisiveness is a real restriction on Gold’s model of explanatory (or in the limit) learning of grammars for languages from positive data. This result also solves an open problem posed in 1986 by Osherson, Stob and Weinstein. Second-time decisive learners semantically conjecture each of their hypotheses for any language at most twice. By contrast, such learners are shown not to restrict Gold’s model of learning. Non U-shaped learning liberalizes the requirement of decisiveness from being a restriction on all hypotheses output to the same restriction but only on correct hypotheses. The situation regarding learning power for non U-shaped learning is a little more complex than that for decisiveness. This is explained shortly below. Gold’s original model for learning grammars from positive data, called EX- learning, requires, for success, syntactic convergence to a correct grammar. A slight variant, called BC- learning, requires only semantic convergence to a sequence of correct grammars that need not be syntactically identical to one another. The second main result says that non U-shaped learning does not restrict EX-learning. However, from an argument of Fulk, Jain and Osherson, non U-shaped learning does restrict BC-learning. In the final section is discussed the possible meaning, for cognitive science, of these results and, in this regard, indicated are some avenues worthy of future investigation.

On the data consumption benefits of accepting increased uncertainty

In the context of learning paradigms of identification in the limit, we address the question: why is uncertainty sometimes desirable? We use mind change bounds on the output hypotheses as a measure of uncertainty and interpret ‘desirable’ as reduction in data memorization, also defined in terms of mind change bounds. The resulting model is closely related to iterative learning with bounded mind change complexity, but the dual use of mind change bounds — for hypotheses and for data — is a key distinctive feature of our approach. We show that situations exist where the more mind changes the learner is willing to accept, the less the amount of data it needs to remember in order to converge to the correct hypothesis. We also investigate relationships between our model and learning from good examples, set-driven, monotonic and strong-monotonic learners, as well as class-comprising versus class-preserving learnability.

Sequential testing of simple hypotheses about compound Poisson processes

One of two simple hypotheses for the unknown arrival rate and jump distribution of a compound Poisson process is correct. We start observing the process, and the problem is to decide on the correct hypothesis as soon as possible and with the smallest probability of wrong decision. We find a Bayes-optimal sequential decision rule and describe completely how to calculate its parameters without any restrictions on the arrival rate and the jump distribution.

A new generalized residual multiple model adaptive estimator of parameters and states

This article develops a modification to the standard Multiple Model Adaptive Estimator (MMAE) which allows the use of a new “generalized residual” in the hypothesis conditional probability calculation. The generalized residual is a linear combination of the traditional Kalman filter residual and the “post-fit” Kalman filter residual which is calculated after measurement incorporation. This new modified MMAE is termed a Generalized Residual Multiple Model Adaptive Estimator (GRMMAE). A derivation is provided for the hypothesis conditional probability formula which the GRMMAE uses to calculate probabilities that each elemental filter contains the correct parameter value. Through appropriate choice of a single scalar GRMMAE design parameter, the GRMMAE can be designed to be equivalent to a traditional MMAE, a post-fit residual modified MMAE, or any linear combination of the two. The original GRMMAE design goal was to choose the GRMMAE design parameter which caused the fastest GRMMAE convergence to the correct hypothesis. However, this article demonstrates that the GRMMAE design parameter can lead to β -dominance, a negative performance effect in the GRMMAE. That fact is a key result of this research as other researchers have previously suggested that the use of post-fit residuals may be advantageous in certain MMAE applications. This article demonstrates the β -dominance effect and recommends that post-fit residuals not be used in an MMAE.

Human Development VIII: A Theory of “Deep” Quantum Chemistry and Cell Consciousness: Quantum Chemistry Controls Genes and Biochemistry to Give Cells and Higher Organisms Consciousness and Complex Behavior

Deep quantum chemistry is a theory of deeply structured quantum fields carrying the biological information of the cell, making it able to remember, intend, represent the inner and outer world for comparison, understand what it “sees”, and make choices on its structure, form, behavior and division. We suggest that deep quantum chemistry gives the cell consciousness and all the qualities and abilities related to consciousness. We use geometric symbolism, which is a pre-mathematical and philosophical approach to problems that cannot yet be handled mathematically. Using Occams razor we have started with the simplest model that works; we presume this to be a many-dimensional, spiral fractal. We suggest that all the electrons of the large biological molecules orbitals make one huge “cell-orbital”, which is structured according to the spiral fractal nature of quantum fields. Consciousness of single cells, multi cellular structures as e.g. organs, multi-cellular organisms and multi-individual colonies (like ants) and human societies can thus be explained by deep quantum chemistry. When biochemical activity is strictly controlled by the quantum-mechanical super-orbital of the cell, this orbital can deliver energetic quanta as biological information, distributed through many fractal levels of the cell to guide form and behavior of an individual single or a multi-cellular organism. The top level of information is the consciousness of the cell or organism, which controls all the biochemical processes. By this speculative work inspired by Penrose and Hameroff we hope to inspire other researchers to formulate more strict and mathematically correct hypothesis on the complex and coherence nature of matter, life and consciousness.

Mouse models of efficient and inefficient anti-tumor immunity, with emphasis on minimal residual disease and tumor escape.

Tumor escape from the host immune response remains the major problem holding the development of immunotherapies for cancer. In this review, congenic mouse lines are discussed that differ dramatically in their ability to respond to tumors tested and, thereby, to survive or to succumb to the tumor and/or its metastases. This ability is under the control of either MHC class I or nontrivial MHC class II beta genes expressed in a small subpopulation of antigen-presenting cells. Two hypotheses can explain the results obtained so far: (1) emergence of tumor cell variants that escape the host immune response in morbid mice but are eliminated in survivors, and (2) tumor-induced immunosuppression, which is either efficient or not, depending on the congenic line used. It is argued that further experimentation on these congenics will allow to choose the correct hypothesis, and to characterize the mechanism(s) of elimination of minimal residual disease and prevention of tumor escape by the immune system of survivors as well as the reason(s) for its failure in morbid mice. It is also argued that the use of these models will substantially increase the chance to resolve the controversy of poor correlation of immunotherapy testing in mice with clinical results.