Brain-inspired perception-decision machine for fake speech detection.

Three disciplines address the codified forms and rules of human thought and reasoning: logic, available since antiquity; dialectics as a process of logical reasoning; and semiotics which focuses on the epistemological properties of the extant domain. However, both the paradigmatic-historical model of knowledge and the logical-semiotic model of thought tend to incorrectly emphasize the separation and differences between the respective domains vs. their overlap and interactions. We propose a sublation of linguistic logics of objects and static forms by a dynamic logic of real physical-mental processes designated as the Logic in Reality (LIR). In our generalized logical theory, dialectics and semiotics are recovered from reductionist interpretations and reunited in a new synthetic paradigm centered on meaning and its communication. Our theory constitutes a meta-thesis composed of elements from science, logic and philosophy. We apply the theory to gain new insights into the structure and role of semiosis, information and communication and propose the concept of ‘ontolon’ to define the element of reasoning as a real dynamic process. It is part of a project within natural philosophy, which will address broader aspects of the dynamics of the growth of civilizations and their potential implications for the information society.

Transformers have been shown to be able to perform deductive reasoning on a logical rulebase containing rules and statements written in natural language. Recent works show that such models can also produce the reasoning steps (i.e., the proof graph) that emulate the model's logical reasoning process. Currently, these black-box models generate both the proof graph and intermediate inferences within the same model and thus may be unfaithful. In this work, we frame the deductive logical reasoning task by defining three modular components: rule selection, fact selection, and knowledge composition. The rule and fact selection steps select the candidate rule and facts to be used and then the knowledge composition combines them to generate new inferences. This ensures model faithfulness by assured causal relation from the proof step to the inference reasoning. To test our framework, we propose FaiRR (Faithful and Robust Reasoner) where the above three components are independently modeled by transformers. We observe that FaiRR is robust to novel language perturbations, and is faster at inference than previous works on existing reasoning datasets. Additionally, in contrast to black-box generative models, the errors made by FaiRR are more interpretable due to the modular approach.

Recently M. Morris Mano presented a method for performing Boolean OR, AND and NOT operations by means of arithmetic and conditional transfer operations in a decimal computer lacking builtin logical instructions [1]. When A , B , C are variables whose defined value is 0 or 1 and a , b , c are the corresponding integer variables with values 0 or 1, his Boolean OR was defined by: “The result of an OR operation of Boolean variables is the same as the arithmetic addition of the a , b , c integer variables after which the following test is made: (a) If the sum is equal to zero, the result is correct; (b) If the sum is larger than zero, the answer is 1.”

Discontinuous transmission based on voice activity detection (VAD) represents a valid solution to improve the spectral efficiency of new-generation wireless communication systems. This paper initially emphasizes the need for more efficient methods to evaluate the performance of a VAD, proposing new performance evaluation criteria, and then presents a speech/pause detection algorithm which is robust to noisy environments. The VAD proposed is based on a pattern approach in which the matching phase is performed by a set of six fuzzy rules trained by means of a new hybrid learning tool. The decision module uses an adaptive threshold varying the estimated signal-to-noise ratio. A series of subjective tests showed that, as compared with the VAD G.729 Annex B, the new algorithm is statistically equivalent with the advantage of providing both a percentage of reduction in the activity increase of about 82% and a clipping reduction percentage of about 24%.

An all optical approach for developing a system involving Kerr material based switches for cryptographic encoding of binary data depending on the input parity

Inverse kinematics solves the problem of how to control robot arm joints to achieve desired end effector positions, which is critical to any robot arm design and implementations of control algorithms. It is a common misunderstanding that closed-form inverse kinematics analysis is solved. Popular software and algorithms, such as gradient descent or any multi-variant equations solving algorithm, claims solving inverse kinematics but only on the numerical level. While the numerical inverse kinematics solutions are relatively straightforward to obtain, these methods often fail, even when the inverse kinematics solutions exist. Therefore, closed-form inverse kinematics analysis is superior, but there is no generalized automated algorithm. Up till now, the high-level logical reasoning involved in solving closed-form inverse kinematics made it hard to automate, so it's handled by human experts. We developed IKBT, a knowledge-based intelligent system that can mimic human experts' behaviors in solving closed-from inverse kinematics using Behavior Tree. Knowledge and rules used by engineers when solving closed-from inverse kinematics are encoded as actions in Behavior Tree. The order of applying these rules is governed by higher level composite nodes, which resembles the logical reasoning process of engineers. It is also the first time that the dependency of joint variables, an important issue in inverse kinematics analysis, is automatically tracked in graph form. Besides generating closed-form solutions, IKBT also explains its solving strategies in human (engineers) interpretable form. This is a proof-of-concept of using Behavior Trees to solve high-cognitive problems.

The purpose of this study is to examine individuals’ computer based complex task performing processes and strategies in order to determine the reasons of failure by cognitive task analysis method and cued retrospective think aloud with eye movement data. Study group was five senior students from Computer Education and Instructional Technologies Department at a state university. A computer based complex task including a logical reasoning process was developed and utilized. Data was collected with cued retrospective think aloud containing a gaze video replay and eye tracking. All the retrospective protocols were transcribed for analysis then a coding schema was developed iteratively from segments. By drawing area of interest fields(AOI) of task panel, eye movement data in specific processes and fields were analyzed by using fixation duration and fixation count metrics. None of the participants completed the task successfully within the given amount of time (max-10 min.). The findings yielded seven cognitive strategies (“information gathering-reviewing”, “identification well-defined instructions”, “cue-seeking”, “using cues”, “assumption”, “trial and error” and “crosschecking”) and their actions obtained from the task performing process. Furthermore, this process is modeled by defining sequences and the relations between the actions and cognitive strategies. It has been revealed that trial and error is the most employed strategy and participants used trial and error without reasoning. Moreover, it has been revealed out that participants mostly failed in choosing and using the right strategy.

This paper is based on two questions about Baumgarten's aesthetics-how is it possible to go from perceptual cognition to beauty and how does perfection serve as a basis for judgement of appreciation. Firstly, this paper reanalyzes Baumgarten's definition of aesthetics, and concludes that the real research field of aesthetics is "free art" that can provide pleasure to people's emotions. "Free art" as the object of appreciation has its own kind of perfection. Secondly, this paper focuses on the consummation of perceptual cognitive ability of the appreciative subject, and restores the logical reasoning process from perceptual cognition to the possibility of beauty by analyzing three directions of consummation. Thirdly, combining with the historical evolution of the theory of perfection, this paper discusses the basis of perfection as the judgment of appreciation from two aspects of quality and quantity respectively. Finally, this paper combs the concept of clarity of extension proposed by Baumgarten, and makes some attempts to solve the universality problem in the judgment of appreciation.

Nuclear deterrence is an integral aspect of the current security architecture and the question has arisen whether adoption of AI will enhance the stability of this architecture or weaken it. The stakes are very high. Stable deterrence depends on a complex web of risk perceptions. All sorts of distortions and errors are possible, especially in moments of crisis. AI might contribute toward reinforcing the rationality of decision-making under these conditions (easily affected by the emotional disturbances and fallacious inferences to which human beings are prone), thereby preventing an accidental launch or unintended escalation. Conversely, judgments about what does or does not suit the “national interest” are not well suited to AI (at least in its current state of development). A purely logical reasoning process based on the wrong values could have disastrous consequences, which would clearly be the case if an AI-based machine were allowed to make the launch decision (this virtually all experts would emphatically exclude), but grave problems could similarly arise if a human actor relied too heavily on AI input.

The tasks of guard placement or sensor deployment in an art gallery, a museum or in the corridors of public and security buildings pose the same problem, which requires placing the guards or sensors so as to cover a specified set of nodes with a minimum number of sensors or guards, thereby reducing the overall cost of the system as well assist power consumption. Generally, minimization can be done using optimization techniques such as linear programming, but in case of sensor deployment or guard placement there is a need either to place or not to place the sensor or guard, and hence only Boolean or binary values are used. Therefore, in order to optimize such a problem, we use the special case of linear integer programming known as Boolean integer linear programming (0-1 ILP). Other algorithms like Pseudo-Boolean SAT Solvers can also be used for the minimization purpose. In this paper, we introduce these minimization algorithms for the sensor deployment problem. We also contribute a greed-based heuristic, which utilizes the fact that the pertinent propositional formulas have variables of purely un-complemented literals. This heuristic has much less computational cost compared to those of 0-1 ILP and the Pseudo-Boolean SAT Solvers.

Publication in the conference proceedings of EUSIPCO, Florence, Italy, 2006

There exists a wide range of constraint programming (CP) problems defined on Boolean functions depending on binary variables. One of the approaches to solving CP problems is using specific appropriate solvers, e.g., SAT solvers. An alternative is using the generic solvers for mixed-integer linear programming problems (MILP), but they require transforming expressions with Boolean functions to linear equations or inequalities. Here, we present two methods of such a transformation which applies to any Boolean function defined by explicit rules giving values of the Boolean function for all combinations of its Boolean variables. The first method represents the Boolean function as a linear equation in the original binary variables and, possibly, binary ancillaries, which become additional variables of the MILP problem being composed. The second method represents the Boolean function as a set of linear inequalities in the original binary variables and one additional continuous variable (representing the value of the function). The choice between the first or second method is a trade-off between the number of binary variables and number of linear constraints in the emerging MP problem. The advantage of the proposed approach is that both methods reduce important cryptanalysis problems, such as the preimaging of hash functions or breaking symmetric ciphers as the MILP problems, which are solved by the generic MILP solvers. Furthermore, the first method enables to reduce the binary linear equations to quadratic unconstrained binary optimization (QUBO), by the quantum annealer, e.g., D-Wave.

As a result of rapid progress in technology and the shrinking product lifecycles faster than ever before, has led creeping realization of additional profits by performing the effective and efficient product recovery operations at a world-class level quintessential. Realization of these motives is complex due to the multidimensional relationships associated with the quality, variety, timeliness, demand changes, and logical processing of product returns and inherent complexity of recovery process. Therefore an Enterprise System (ES) perspective will give us a scope to develop a profit oriented recovery process as a flexible system that can handle products with various options and greater return volume and structural variability. This paper proposes generic model to enterprises engaged in or to be engaged in product recovery processes. A semi or partially flexible decision process model that facilitates flexible decision and information sharing (DIS) functions in product returns. This DIS model leads us to conceptualize the evolution of information associated with a product returns and how it might be encapsulated by Reverse Enterprise System (RES) to improve its profit and system performance.

In this paper, we present our approach to detection of hate speech against women and immigrants in tweets for our participation in the SemEval-2019 Task 5. We trained an SVM and an RF classifier using character bi- and trigram features and a BiLSTM pre-initialized with external word embeddings. We combined the predictions of the SVM, RF and BiLSTM in two different ensemble models. The first was a majority vote of the binary values, and the second used the average of the confidence scores. For development, we got the highest accuracy (75%) by the final ensemble model with majority voting. For testing, all models scored substantially lower and the scores between the classifiers varied more. We believe that these large differences between the higher accuracies in the development phase and the lower accuracies we obtained in the testing phase have partly to do with differences between the training, development and testing data.

The concepts of nonlocal, or distributed, cortical and cognitive activation are examined for their usefulness in describing the relations between sleep and waking neurocognitive processes. Changes in the pattern of distributed activation and inhibition of selected portions of sensory, cognitive, and motor decision modules account for the differences in imagery and thought across sleep and waking states in comparable environments. The massive inhibition of sensory and proprioceptive input to perceptual modules in Stage 1 REM sleep leaves the perceptual and cognitive modules, by default, with their own output as their sole input. Given this constraint, the activation of portions of the cortical structures that execute waking perceptual, cognitive and motor responses is necessary and sufficient to produce the imagery and thought of dreaming sleep. Connectionist models are introduced so that neurophysiological and cognitive concepts of distributed and local activation and inhibition can be translated into a common language, and in so doing, are used to simulate several processes fundamental to the production of imaginal thought and dreaming.