Learning the structure of event sequences is a ubiquitous problem in cognition and particularly in language. One possible solution is to learn a probabilistic generative model of sequences that allows making predictions about upcoming events. Though appealing from a neurobiological standpoint, this approach is typically not pursued in connectionist modeling. Here, we investigated a sequential version of the restricted Boltzmann machine, a stochastic recurrent neural network that extracts high-order structure from sensory data through unsupervised generative learning and can (...) encode contextual information in the form of internal, distributed representations. We assessed whether this type of network can extract the orthographic structure of English monosyllables by learning a generative model of the letter sequences forming a word training corpus. We show that the network learned an accurate probabilistic model of English graphotactics, which can be used to make predictions about the letter following a given context as well as to autonomously generate high-quality pseudowords. The model was compared to an extended version of simple recurrent networks, augmented with a stochastic process that allows autonomous generation of sequences, and to non-connectionist probabilistic models. We conclude that sequential RBMs and stochastic simple recurrent networks are promising candidates for modeling cognition in the temporal domain. (shrink)

It is possible to learn multiple layers of non-linear features by backpropagating error derivatives through a feedforward neural network. This is a very effective learning procedure when there is a huge amount of labeled training data, but for many learning tasks very few labeled examples are available. In an effort to overcome the need for labeled data, several different generative models were developed that learned interesting features by modeling the higher order statistical structure of a set of input vectors. One (...) of these generative models, the restricted Boltzmann machine (RBM), has no connections between its hidden units and this makes perceptual inference and learning much simpler. More significantly, after a layer of hidden features has been learned, the activities of these features can be used as training data for another RBM. By applying this idea recursively, it is possible to learn a deep hierarchy of progressively more complicated features without requiring any labeled data. This deep hierarchy can then be treated as a feedforward neural network which can be discriminatively fine-tuned using backpropagation. Using a stack of RBMs to initialize the weights of a feedforward neural network allows backpropagation to work effectively in much deeper networks and it leads to much better generalization. A stack of RBMs can also be used to initialize a deep Boltzmann machine that has many hidden layers. Combining this initialization method with a new method for fine-tuning the weights finally leads to the first efficient way of training Boltzmann machines with many hidden layers and millions of weights. (shrink)

The research expects to explore the application of intelligent music recognition technology in music teaching. Based on the Long Short-Term Memory network knowledge, an algorithm model which can distinguish various music signals and generate various genres of music is designed and implemented. First, by analyzing the application of machine learning and deep learning in the field of music, the algorithm model is designed to realize the function of intelligent music generation, which provides a theoretical basis for relevant research. Then, by (...) selecting massive music data, the music style discrimination and generation model is tested. The experimental results show that when the number of hidden layers of the designed model is 4 and the number of neurons in each layer is 1,024, 512, 256, and 128, the training result difference of the model is the smallest. The classification accuracy of jazz, classical, rock, country, and disco music types can be more than 60% using the designed algorithm model. Among them, the classification effect of jazz schools is the best, which is 77.5%. Moreover, compared with the traditional algorithm, the frequency distribution of the music score generated by the designed algorithm is almost consistent with the spectrum of the original music. Therefore, the methods and models proposed can distinguish music signals and generate different music, and the discrimination accuracy of different music signals is higher, which is superior to the traditional restricted Boltzmann machine method. (shrink)

Because the traditional social media fuzzy static image interactive three-dimensional (3D) reconstruction method has the problem of poor reconstruction completeness and long reconstruction time, the social media fuzzy static image interactive 3D reconstruction method is proposed. For preprocessing the fuzzy static image of social media, the Harris corner detection method is used to extract the feature points of the preprocessed fuzzy static image of social media. According to the extraction results, the parameter estimation algorithm of contrast divergence is used to (...) learn the restricted Boltzmann machine (RBM) network model, and the RBM network model is divided into input, output, and hidden layers. By combining the RBM-based joint dictionary learning method and a sparse representation model, an interactive 3D reconstruction of fuzzy static images in social media is achieved. Experimental results based on the CAD software show that the proposed method has a reconstruction completeness of above 95% and the reconstruction time is less than 15 s, improving the completeness and efficiency of the reconstruction, effectively reconstructing the fuzzy static images in social media, and increasing the sense of reality of social media images. (shrink)

Happiness can be regarded as an evaluation of life satisfaction. A high level of wellbeing can promote self-fulfillment and build a rational, peaceful, self-esteem, self-confidence, and positive social mentality. Therefore, the analysis of the factors of happiness is of great significance for the continuous improvement of the individual’s sense of security and gain and the realization of the maximization of self-worth. Emotion is not only an important internal factor that affects happiness, but it can also accurately reflect the individual’s happiness. (...) However, most of current happiness evaluation methods based on the emotional analysis belong to shallow learning paradigm, making the deep learning method unexploited for automatically happiness decoding. In this article, we analyzed the emotions of graduates during their employment and studied its influence on personal happiness at work. We proposed deep restricted Boltzmann machine for graduates’ happiness evaluation during employment. Furthermore, to mitigate the information loss when passing through many network layers, we introduced the skip connections to DRBM and proposed a deep residual RBM for enhancing the valuable information. We further introduced an attention mechanism to DRRBM to focus on the important factors. To verify the effectiveness of the proposed method on the happiness evaluation tasks, we conducted extensive experiments on the statistical data of the China Comprehensive Social Survey. Compared with the state-of-the-art methods, our method shows better performance, which proves the practicability and feasibility of our method for happiness evaluation. (shrink)

We describe a deep generative model in which the lowest layer represents the word-count vector of a document and the top layer represents a learned binary code for that document. The top two layers of the generative model form an undirected associative memory and the remaining layers form a belief net with directed, top-down connections. We present efficient learning and inference procedures for this type of generative model and show that it allows more accurate and much faster retrieval than latent (...) semantic analysis. By using our method as a filter for a much slower method called TF-IDF we achieve higher accuracy than TF-IDF alone and save several orders of magnitude in retrieval time. By using short binary codes as addresses, we can perform retrieval on very large document sets in a time that is independent of the size of the document set using only one word of memory to describe each document. (shrink)

There are three main problems in track fastener defect detection based on image: The number of abnormal fastener pictures is scarce, and supervised learning detection model is difficult to establish. The potential data features obtained by different feature extraction methods are different. Some methods focus on edge features, and some methods focus on texture features. Different features have different detection capabilities, and these features are not effectively fused and utilized. The detection of the track fastener clip will be interfered by (...) the track fastener bolt subimage. Aiming at the above three problems, a method for track fastener defects detection based on Local Deep Feature Fusion Network is proposed. Firstly, the track fastener image segmentation method is used to obtain the track fastener clip subimage, which can effectively reduce the interference of bolt subimage features on the track fastener clip detection. Secondly, the edge features and texture features of track fastener clip subimages are extracted by Autoencoder and Restricted Boltzmann Machine, and the features are fused. Finally, the similarity measurement method Mahalanobis Distance is used to detect defects in track fasteners. The effectiveness of the proposed method is verified by real Pandrol track fastener images. (shrink)

Probabilistic argumentation and neuro-argumentative systems offer new computational perspectives for the theory and applications of argumentation, but their principled construction involves two entangled problems. On the one hand, probabilistic argumentation aims at combining the quantitative uncertainty addressed by probability theory with the qualitative uncertainty of argumentation, but probabilistic dependences amongst arguments as well as learning are usually neglected. On the other hand, neuro-argumentative systems offer the opportunity to couple the computational advantages of learning and massive parallel computation from neural networks (...) with argumentative reasoning and explanatory abilities, but the relation of probabilistic argumentation frameworks with these systems has been ignored so far. Towards the construction of neuro-argumentative systems based on probabilistic argumentation, we associate a model of abstract argumentation and the graphical model of Boltzmann machines in order to (i... (shrink)

Boltzmann’s Bildtheorie, which asserts that scientific theories are ‘mental pictures’ having at best a partial similarity to reality, was a core element of his philosophy of science. The aim of this article is to draw attention to a neglected aspect of it, namely its significance for the issue of scientific explanation and understanding, regarded by Boltzmann as central goals of science. I argue that, in addition to being an epistemological view of the interpretation of scientific theories Boltzmann’s (...) Bildtheorie has implications for the nature of scientific understanding. This aspect has as yet been ignored because discussion of the Bildtheorie has been restricted to the realism-instrumentalism debate. To elucidate my analysis of Boltzmann’s Bildtheorie concrete examples are presented, and the pragmatist and Darwinist roots of Boltzmann’s view are discussed.Moreover, I propose to use Boltzmann’s ideas as a starting-point for developing a novel analysis of the notion of scientific understanding, of which a brief impression is given. It shows that the study of Boltzmann’s philosophy is not only of historical interest but can be relevant also to modern philosophy of science and to the methodology of theoretical physics. (shrink)

Many commentators agree that Wittgenstein took the idea that propositions are Bilder, or at least the terminology of Bilder, from Heinrich Hertz, or from Hertz and Ludwig Boltzmann. Boltzmann, the great Viennese theoretical physicist, was the founder of statistical thermodynamics, the modern theory of heat. The context within which Hertz and Boltzmann worked was one in which many prominent theoretical physicists accepted the Kantian restriction that our thought cannot access 'things in themselves', but works only with representations. (...) Wittgenstein may have found what Hertz said attractive, but even if he did develop a theory of representation or a picture theory of meaning, neither was something that he just took over from Hertz, fully formed, and his main concerns were puzzles about intentionality, not science. Hertz and Boltzmann did influence various aspects of the Tractatus. Their greatest and most lasting influence on Wittgenstein, though, was on his conception of philosophy itself: its aim, and its method. (shrink)

Machine consciousness exists already in organic systems and it is only a matter of time -- and some agreement -- before it will be realised in reverse-engineered organic systems and forward- engineered inorganic systems. The agreement must be over the preconditions that must first be met if the enterprise is to be successful, and it is these preconditions, for instance, being a socially-embedded, structurally-coupled and dynamic, goal-directed entity that organises its perceptual input and enacts its world through the application of (...) both a cognitive and kinaesthetic imagination, that I shall concentrate on presenting in this paper. It will become clear that these preconditions will present engineers with a tall order, but not, I will argue, an impossible one. After all, we might agree with Freeman and Núñez's claim that the machine metaphor has restricted the expectations of the cognitive sciences (Freeman & Núñez, 1999); but it is a double-edged sword, since our limited expectations about machines also narrow the potential of our cognitive science. (shrink)

We introduce a restricted second-order logic $\textrm{SO}^{\textit{plog}}$ for finite structures where second-order quantification ranges over relations of size at most poly-logarithmic in the size of the structure. We demonstrate the relevance of this logic and complexity class by several problems in database theory. We then prove a Fagin’s style theorem showing that the Boolean queries which can be expressed in the existential fragment of $\textrm{SO}^{\textit{plog}}$ correspond exactly to the class of decision problems that can be computed by a non-deterministic (...) Turing machine with random access to the input in time $O^k)$ for some $k \ge 0$, i.e. to the class of problems computable in non-deterministic poly-logarithmic time. It should be noted that unlike Fagin’s theorem which proves that the existential fragment of second-order logic captures NP over arbitrary finite structures, our result only holds over ordered finite structures, since $\textrm{SO}^{\textit{plog}}$ is too weak as to define a total order of the domain. Nevertheless, $\textrm{SO}^{\textit{plog}}$ provides natural levels of expressibility within poly-logarithmic space in a way which is closely related to how second-order logic provides natural levels of expressibility within polynomial space. Indeed, we show an exact correspondence between the quantifier prefix classes of $\textrm{SO}^{\textit{plog}}$ and the levels of the non-deterministic poly-logarithmic time hierarchy, analogous to the correspondence between the quantifier prefix classes of second-order logic and the polynomial-time hierarchy. Our work closely relates to the constant depth quasipolynomial size AND/OR circuits and corresponding restricted second-order logic defined by David A. Mix Barrington in 1992. We explore this relationship in detail. (shrink)

There are plenty of intelligent machines in our world today: digital computers and autonomous robots. At the heart of each of these machines there are automatic formal systems (programs running on a digital computer). Now, if the interpretation of a formal system does not belong to the formal system itself, if the interpretation has to be added, it is worth asking: in the case of these intelligent machines that are massively interspersed in our social interactions, where does (...) the interpretation come from? In this paper, we analyse what we call the invisibility of interpretation. Dealing with various types of formal systems (computers, robots, formalist approaches to Economics), the human source of the interpretation of these systems is sometimes concealed by a formalist restriction. To show how the formalist restriction produces the invisibility of interpretation allows us to underline our responsibility, as human agents, for all this interpretative work—and its importance for us as human beings. (shrink)

The problem of valid induction could be stated as follows: are we justified in accepting a given hypothesis on the basis of observations that frequently confirm it? The present paper argues that this question is relevant for the understanding of Machine Learning, but insufficient. Recent research in inductive reasoning has prompted another, more fundamental question: there is not just one given rule to be tested, there are a large number of possible rules, and many of these are somehow confirmed by (...) the data — how are we to restrict the space of inductive hypotheses and choose effectively some rules that will probably perform well on future examples? We analyze if and how this problem is approached in standard accounts of induction and show the difficulties that are present. Finally, we suggest that the explanation-based learning approach and related methods of knowledge intensive induction could be, if not a solution, at least a tool for solving some of these problems. (shrink)

This paper argues that machine learning and epidemiology are on collision course over causation. The discipline of epidemiology lays great emphasis on causation, while ML research does not. Some epidemiologists have proposed imposing what amounts to a causal constraint on ML in epidemiology, requiring it either to engage in causal inference or restrict itself to mere projection. We whittle down the issues to the question of whether causal knowledge is necessary for underwriting predictions about the outcomes of public health interventions. (...) While there is great plausibility to the idea that it is, conviction that something is impossible does not by itself motivate a constraint to forbid trying. We disambiguate the possible motivations for such a constraint into definitional, metaphysical, epistemological, and pragmatic considerations and argue that “Proceed with caution” is the outcome of each. We then argue that there are positive reasons to proceed, albeit cautiously. Causal inference enforces existing classification schema prior to the testing of associational claims, but associations and classification schema are more plausibly discovered in a back-and-forth process of gaining reflective equilibrium. ML instantiates this kind of process, we argue, and thus offers the welcome prospect of uncovering meaningful new concepts in epidemiology and public health—provided it is not causally constrained. (shrink)

Certain research strands can yield “forbidden knowledge”. This term refers to knowledge that is considered too sensitive, dangerous or taboo to be produced or shared. Discourses about such publication restrictions are already entrenched in scientific fields like IT security, synthetic biology or nuclear physics research. This paper makes the case for transferring this discourse to machine learning research. Some machine learning applications can very easily be misused and unfold harmful consequences, for instance, with regard to generative video or text synthesis, (...) personality analysis, behavior manipulation, software vulnerability detection and the like. Up till now, the machine learning research community embraces the idea of open access. However, this is opposed to precautionary efforts to prevent the malicious use of machine learning applications. Information about or from such applications may, if improperly disclosed, cause harm to people, organizations or whole societies. Hence, the goal of this work is to outline deliberations on how to deal with questions concerning the dissemination of such information. It proposes a tentative ethical framework for the machine learning community on how to deal with forbidden knowledge and dual-use applications. (shrink)

This thesis shows morality to be broader and more diverse than its human instantiation. It uses the idea of machine morality to argue for this position. Specifically, it contrasts the possibilities open to humans with those open to machines to meaningfully engage with the moral domain. -/- This contrast identifies distinctive characteristics of human morality, which are not fundamental to morality itself, but constrain our thinking about morality and its possibilities. It also highlights the inherent potential of machine morality (...) to be radically different from its human counterpart and the implications this has for the moral significance of machines. -/- My argument is particularly focussed on moral theory, which is the study of the observable and hypothetical conceptual structures of morality. By identifying structures that are recognisably moral in nature but which sit outside the boundaries of human realisation, we have tangible proof that a meaningful distinction exists between human morality and the wider moral domain. -/- This is achieved by showing that certain essentially human limits restrict the conceptual possibilities open to human realisation. The tight coupling between these limits and the existing conceptual structures of human morality also explains why it is unjustifiable to assume that the same structures would be suitable for machines. They do not share these same limits with us, which leads me to conclude that many conceptual structures are quite distinctive to human morality and that the structures of machine morality would be significantly different. Four examples illustrate these conclusions concretely. -/- The first, supererogation, is an example of a moral concept that doesn't easily extend to machines. Human limits dictate what it is reasonable to expect from one another and restrict our ability to pursue aspirational moral goals. I show that machine supererogation, if it is at all possible, would require a very different justificatory basis to be coherent. -/- The second, agency, is an example of a concept whose structures extend beyond the bounds of human realisation. The greater flexibility of artificial identity allows machines to experiment with novel forms of inter- and intra-agency. In comparison, human agency structures are limited by their tight coupling with human conceptions of identity. -/- The third, moral aspiration, is a concept with a distinctive function in human morality. Certain aspirational ends are peculiar in that they are obviously unrealisable and even undesirable, yet their pursuit is instrumentally justifiable. This justification depends on cognitive limits that aren't shared by machines, which leads me to conclude that the role of moral aspiration in machine morality, if there is any, would necessarily differ from its human counterpart. -/- The fourth, moral responsibility, is an example of a concept whose existing practices don't translate over to machines. We don't understand machine agency well enough to be able to judge a machine's culpability or effectively blame it. Consequently, I suggest that a responsibility conception prioritising understanding over blame is a more promising avenue for a shared conception suitable for both humans and machines. -/- This thesis does not speculate about the existence of moral machines. This remains an open question, and one that is largely irrelevant for my conclusions as the idea alone is enough to advance our thinking. It does this by helping us identify the boundaries of human morality, and then, to think beyond them by recognising the possibilities of machine morality. (shrink)

The article discusses the strategies of translation of «machine texts» on the example of generative transformers (GPT). Currently, the study and development of machine text generation has become an important task for processing and analyzing texts in different languages. Modern technologies of artificial intelligence and neural networks allow us to create powerful tools for activities in this field, which are becoming more and more effective every year. Generative transformers are one of such tools. The study of generative transformers also allows (...) developers to create more accurate and efficient machine translation algorithms, which improves the quality of translations and improves the user experience. In this context, the features of machine texts created by generative transformers, their patterns, errors and imperfections, which require special translation strategies, deserve special interest. Today we can say that the generation of unique and relevant texts is a routine task that has been automated. Nevertheless, certain restrictions for the use of such texts still exist, in particular, their use requires the use of appropriate translation strategies. The paper proposes the author’s typology of translation strategies, where, taking into account the features of AST, it is proposed to add a substrategy of tertiary-moderation translation. (shrink)

Information security is perceived as an important and vital aspect for the survival of any business. Preserving user identity and limiting the access of web resources only to the humans and restricting ‘bots’ is an ever challenging area of study. With the increase in computing power and development of newer approaches towards circumvention and reverse-engineering, the recognition gap present between the machines and the humans is said to be decreasing. Turing test and its modified versions are in place to (...) deal with such problems and ways to resolve them by developing complex algorithms for bot prevention systems like CAPTCHA (Completely Automated Public Turing test to tell Computers and Humans Apart). This paper will deal with the use of “Machine Vision” for judging the ability of the machines to compete with humans in breaking sequences of security systems like CAPTCHA. Reverse Turing test will be put to practise here. Complex image recognition technologies and novel approaches towards using Human interactive proofs (HIP) are discussed. The progress of Turing test over the past 60 years has been paid due attention at the end. After all this experimentation, it can be said that the current machine vision is quite poor and is far worse than it is expected to be. (shrink)

As robots increasingly become part of our everyday lives, questions arise with regards to how to approach them and how to understand them in social contexts. The Western history of human–robot relations revolves around competition and control, which restricts our ability to relate to machines in other ways. In this study, we take a relational approach to explore different manners of socializing with robots, especially those that exceed an instrumental approach. The nonhuman subjects of this study are built to (...) explore non-purposeful behavior, in an attempt to break away from the assumptions of utility that underlie the hegemonic human–machine interactions. This breakaway is accompanied by ‘learning to be attuned’ on the side of the human subjects, which is facilitated by continuous relations at the level of everyday life. Our paper highlights this ground for the emergence of meanings and questions that could not be subsumed by frameworks of control and domination. The research-creation project Machine Ménagerie serves as a case study for these ideas, demonstrating a relational approach in which the designer and the machines co-constitute each other through sustained interactions, becoming attuned to one another through the performance of research. Machine Ménagerie attempts to produce affective and playful—if not unruly—nonhuman entities that invite interaction yet have no intention of serving human social or physical needs. We diverge from other social robotics research by creating machines that do not attempt to mimic human social behaviours. (shrink)

This paper examines the implications of brain–machine interfaces (BMIs) from a libertarian perspective, arguing that their widespread use necessitates careful scrutiny due to potential risks to individual autonomy, freedom, privacy, and dignity. BMIs, while offering significant technological advancements, pose severe threats by potentially undermining fundamental libertarian values. The paper discusses how BMIs could enable invasive surveillance, thought manipulation, and emotional control, drawing parallels to Robert Nozick’s Experience Machine thought experiment. Unlike the hypothetical machine, which offers simulated experiences, BMIs could facilitate (...) real-time control over individuals’ thoughts and emotions, leading to unprecedented forms of government overreach and coercion. The paper concludes that although libertarians advocate for minimal state intervention, the profound impact of BMIs on personal freedom and autonomy warrants a cautious approach, including potential restrictions on their use to safeguard against their misuse and to preserve individual self-ownership and dignity. (shrink)

Descartes developed an elaborate theory of animal physiology that he used to explain functionally organized, situationally adapted behavior in both human and nonhuman animals. Although he restricted true mentality to the human soul, I argue that he developed a purely mechanistic (or material) ‘psychology’ of sensory, motor, and low-level cognitive functions. In effect, he sought to mechanize the offices of the Aristotelian sensitive soul. He described the basic mechanisms in the Treatise on man, which he summarized in the Discourse. (...) However, the Passions of the soul contains his most ambitious claims for purely material brain processes. These claims arise in abstract discussions of the functions of the passions and in illustrations of those functions. Accordingly, after providing an intellectual context for Descartes’s theory of the passions, especially by comparison with that of Tho- mas Aquinas, I examine its ‘machine psychology’, including the role of habituation and association. I contend that Descartes put forth what may reasonably be called a ‘psychology’ of the unensouled animal body and, correspondingly, of the human body when the soul does not intervene. He thus conceptually distinguished a mechanistically explicable sensory and motor psychology, common to nonhuman and human animals, from true mentality involving higher cognition and volition and requiring (in his view) an immaterial mind. (shrink)

This paper outlines the development of a proof-of-concept workflow using machine learning and computer vision techniques to unlock the data within digitized handwritten US Census forms from the 1950s. The 1950s US Census includes over 6.5 million page images and was only recently made available to the public on April 1, 2022, following a 72-year access restriction period. Our project uses computational treatments to assist researchers in their efforts to recover and preserve the history of the erased Sacramento Japantown. Sacramento (...) once housed the fourth largest Japantown in the United States before experiencing WWII Japanese American Incarceration and the 1950s US Government program of urban renewal. The goal is to augment a researcher’s work in selecting a subset of Census pages for further transcription and analysis. We demonstrate a workflow for extracting demographic information using computer vision for image segmentation, and machine learning for handwritten character recognition. The workflow consists of a computational filtering process for Census records and a user interface for page review. These computational techniques are suitable for other cities, states, and communities, and demonstrate new strategies to unlock vital demographic information. The approach highlights the potential benefits of computational techniques for the analysis of form-based historical records of the twentieth century that can have an impact on social justice. (shrink)

We are engineers, and our view of consciousness is shaped by an engineering ambition: we would like to build a conscious machine. We begin by acknowledging that we may be a little disadvantaged, in that consciousness studies do not form part of the engineering curriculum, and so we may be starting from a position of considerable ignorance as regards the study of consciousness itself. In practice, however, this may not set us back very far; almost a decade ago, Crick wrote: (...) 'Everyone has a rough idea of what is meant by consciousness. It is better to avoid a precise definition of consciousness because of the dangers of premature definition. Until the problem is understood much better, any attempt at a formal definition is likely to be either misleading or overly restrictive, or both'. This seems to be as true now as it was then, although the identification of different aspects of consciousness by Block has certainly brought a degree of clarification. On the other hand, there is little doubt that consciousness does seem to be something to do with the operation of a sophisticated control system, and we can claim more familiarity with control systems than can most philosophers, so perhaps we can make up some ground there. (shrink)

Gödel's Theorem is often used in arguments against machine intelligence, suggesting humans are not bound by the rules of any formal system. However, Gödelian arguments can be used to support AI, provided we extend our notion of computation to include devices incorporating random number generators. A complete description scheme can be given for integer functions, by which nonalgorithmic functions are shown to be partly random. Not being restricted to algorithms can be accounted for by the availability of an arbitrary (...) random function. Humans, then, might not be rule-bound, but Gödelian arguments also suggest how the relevant sort of nonalgorithmicity may be trivially made available to machines. (shrink)

We are engineers, and our view of consciousness is shaped by an engineering ambition: we would like to build a conscious machine. We begin by acknowledging that we may be a little disadvantaged, in that consciousness studies do not form part of the engineering curriculum, and so we may be starting from a position of considerable ignorance as regards the study of consciousness itself. In practice, however, this may not set us back very far; almost a decade ago, Crick wrote: (...) 'Everyone has a rough idea of what is meant by consciousness. It is better to avoid a precise definition of consciousness because of the dangers of premature definition. Until the problem is understood much better, any attempt at a formal definition is likely to be either misleading or overly restrictive, or both'. This seems to be as true now as it was then, although the identification of different aspects of consciousness by Block has certainly brought a degree of clarification. On the other hand, there is little doubt that consciousness does seem to be something to do with the operation of a sophisticated control system, and we can claim more familiarity with control systems than can most philosophers, so perhaps we can make up some ground there. (shrink)

This paper extends three decades of work arguing that researchers who discuss consciousness should not restrict themselves only to (adult) human minds, but should study (and attempt to model) many kinds of minds, natural and artificial, thereby contributing to our understanding of the space containing all of them. We need to study what they do or can do, how they can do it, and how the natural ones can be emulated in synthetic minds. That requires: (a) understanding sets of requirements (...) that are met by different sorts of minds, i.e. the niches that they occupy, (b) understanding the space of possible designs, and (c) understanding complex and varied relationships between requirements and designs. Attempts to model or explain any particular phenomenon, such as vision, emotion, learning, language use, or consciousness lead to muddle and confusion unless they are placed in that broader context. A methodology for making progress is summarised and a novel requirement proposed for a theory of how human minds work: the theory should support a single generic design for a learning, developing system that, in addition to meeting familiar requirements, should be capable of developing different and opposed philosophical viewpoints about consciousness, and the so-called hard problem. In other words, we need a common explanation for the mental machinations of mysterians, materialists, functionalists, identity theorists, and those who regard all such theories as attempting to answer incoherent questions. No designs proposed so far come close. (shrink)

COVID-19, the most severe public health problem to occur in the past 10 years, has greatly impacted people's mental health. Colleges in China have reopened, and how to prevent college students from suffering secondary damage due to school reopening remains elusive. This cross-sectional study was aimed to evaluate the psychological impact of COVID-19 after school reopening and explore via machine learning the factors that influence anxiety and depression among students. Among the 478 valid online questionnaires collected between September 14th and (...) September 20th, 74 showed symptoms of anxiety, and 155 showed symptoms of depression. Descriptive analysis of basic personal characteristics indicated that students at a higher grade, having relatives or friends who have been infected, fearing being infected, and having a pessimistic attitude to COVID-19 easily experience anxiety or depression. The Synthetic Minority Oversampling Technique was utilized to counteract the imbalance of retrieved data. The Akaike Information Criterion and multivariate logistic regression were performed to explore significant influence factors. The results indicate that exercise frequency, alcohol use, school reopening, having relatives or friends who have been infected, self-quarantine, quarantine of classmates, taking temperature routinely, wearing masks routinely, sleep quality, retaining holiday, availability of package delivery, take-out availability, lockdown restriction, several areas in school closed due to COVID-19, living conditions in the school, taking the final examinations after school reopening, and the degree to which family economic status is influenced by COVID-19 are the primary influence factors for anxiety or depression. To evaluate the effect of our model, we used 5-fold cross-validation, and the average area under the curve values of the receiver operating characteristic curves of anxiety and depression on the test set reached 0.885 and 0.806, respectively. To conclude, we examined the presence of anxiety and depression symptoms among Chinese college students after school reopening and explored many factors influencing students' mental health, providing reasonable school management suggestions. (shrink)

Hypothesis-driven approaches identified important characteristics that differentiate violent from non-violent radicals. However, they produced a mosaic of explanations as they investigated a restricted number of preselected variables. Here we analyzed without a priory assumption all the variables of the “Profiles of Individual Radicalization in the United States” database by a machine learning approach. Out of the 79 variables considered, 19 proved critical, and predicted the emergence of violence with an accuracy of 86.3%. Typically, violent extremists came from criminal but (...) not radical backgrounds and were radicalized in late stages of their life. They were followers in terrorist groups, sought training, and were radicalized by social media. They belonged to low social strata and had problematic social relations. By contrast, non-violent but still criminal extremists were characterized by a family tradition of radicalism without having criminal backgrounds, belonged to higher social strata, were leaders in terrorist organizations, and backed terrorism by supporting activities. Violence was also promoted by anti-gay, Sunni Islam and Far Right, and hindered by Far Left, Anti-abortion, Animal Rights and Environment ideologies. Critical characteristics were used to elaborate a risk-matrix, which may be used to predict violence risk at individual level. (shrink)

Pozzi1 has thoroughly analysed testimonial injustices in the automated Prediction Drug Monitoring Programmes (PDMPs) case. Although Pozzi1 suggests that ‘the shift from an interpersonal to a structural dimension … bears a significant moral component’, her topical investigation does not further conceptualise the type of collective knowledge practices necessary to achieve epistemic justice. As Pozzi1 concludes: ‘this paper shows the limitations of systems such as automated PDMPs, it does not provide possible solutions’. In this commentary, we propose that an Ubuntu perspective—which, (...) so far, has received little attention in connection with epistemic injustice in medical machine learning (MML)—can fill this gap. This perspective can be used to mitigate some of the harmful effects of epistemic injustices by encouraging community involvement and forward-looking responsibility. It may be instructive to note first and foremost that the debate on epistemic justice and artificial intelligence in medicine has mostly been restricted to the Western World, that is, Europe and North America. There has been very little insight or accommodation of insights from the Global South. Ubuntu is a concept originating from Southern Africa that focuses on the close connections and mutual …. (shrink)

The uniform halting problem (UH) can be stated as follows:Give a decision procedure which for any given Turing machine (TM) will decide whether or not it has an immortal instantaneous description (ID).An ID is called immortal if it has no terminal successor. As it is generally the case in the literature (see e.g. Minsky [4, p. 118]) we assume that in an ID the tape must be blank except for some finite number of squares. If we remove this restriction the (...) UH becomes the immortality problem (IP). (shrink)

What I refer to is how our thought in inventing, designing, modifying, and using machines carries over into acts we do not consciously associate with them—like writing or reading poetry. An airplane in flight may be “pure poetry,” or a Ferrari “a poem in steel”; it intrigues me to consider that beneath such object comparisons an object-of-thought connection may be made. Or in other words, there may be really something to a hackneyed compliment like “poem in steel.” My preference (...) for thought form over object form makes me less interested in machines that we can see than in those that we can’t, and it makes me direct my inquiry along two lines, concerning two questions about invisibility. The first involves the history of technology: How is it that machines “disappear”—become less visible, impinge less and less upon popular consciousness? The effect cannot be imputed entirely to familiarity. The second involves the history of literature: If machines have disappeared, are there “disappeared”—or “invisible”—machines in literature? It is reasonably clear that we go to considerable lengths to hide the machines that surround us and that we choose, or our artists choose, insofar as such choice can be located, to restrict the appearance of machines in art.3 Commonly their restricted appearance in art is understood as a kind of resistance to some form of machine takeover, while their concealment has only uncommonly received analysis and is not generally discussed in context with the matter of artistic representation. But we do accept, at least in theory, the idea that style of living and style in literature are connected. If the machine penetrates our style of living, then—and this is the end of my inquiry—these invisibilities are of interest to literary criticism, for they have something to do with the way literature is written, with whether or not, that is, writers choose to describe machines, use them as characters, or give them any role at all to play in surface structure. Since the artist often works to reveal what his society works to conceal and since the postmodern period has so far been one of crisis in the relations between society and technology, we may expect to find in postmodern fiction, or in writing to come, some greater revelation or simple exposure of the workings, than we could have seen before. 3. I use “restrict” in a statistical sense, for since the beginning of the Industrial Revolution there have been artists to celebrate the machine; today’s hyperrealist movement offers contemporary examples. These artists—Tom Blackwell, Ron Kleemann, Ralph Goings, and others—work a transformation like that of Andy Warhol with the pop object. They present not so much a celebration as an effort to turn looking into seeing, here directed at automobiles in particular, so much a part of everyday life as to have become indistinguishable from one another and as a species from the other parts of the semiurban landscape—the mailbox, the front lawn, the visiting relatives staring into the camera. Strother B. Purdy is the author of The Hole in the Fabric: Science, Contemporary Literature, and Henry James. His previous contributions to Critical Inquiry are “Stalingrad and My Lai: A Literary-Political Speculation” and “Reply to Lawrence W. Hyman. (shrink)

With the development of the globalization of science and technology, innovation has become an important driving force for regional economic development. As a core element of regional innovation, financial R&D resources have also become a key element to enhance national innovation capabilities and national economic competitiveness. National and regional innovation capabilities have a direct impact. There are also many deep-seated problems behind the world-renowned achievements, such as irrational industrial structure, insufficient independent innovation capabilities, low resource utilization efficiency, and the service (...) quality and efficiency of financial institutions for the transformation of total factor productivity. These problems extremely restrict the efficiency upgrade and further development of our country’s total factor productivity. This study uses the DEA-Malmquist index model to measure the efficiency of fiscal R&D resource allocation in 28 provinces and regions in China in the past 10 years and uses Mapinfo12.0 software to analyze regional differences in the efficiency of fiscal R&D resource allocation in China from a spatial perspective. During the year, the overall R&D resource allocation efficiency of 28 provinces and autonomous regions in China has shown an upward trend. The efficiency of fiscal R&D resource allocation and the concentration of financial factors have had a positive impact on total factor productivity, transform and upgrade factors, increase total factor productivity, and provide empirical evidence for building a strong country. (shrink)

The development of computer science has raised ethical concerns regarding the potential negative impacts of machine learning tools on people and society. Some examples are pornographic deepfakes used as weapons of war against women; pattern recognition designed to uncover sexual orientation; and misuse of data and deep learning by private companies to influence democratic elections. We contend that these three examples are cases of automated evil. In this article, we defend that the concept of forbidden knowledge can help to inform (...) a coherent ethical framework in the context of machine learning research. We conclude that restricting generalised access to extensive data and limiting access to ready-to-use codes would mitigate potential harms caused by machine learning tools. In addition, the notions of intersectionality and interdisciplinarity should be systematically introduced in data and computer science research. (shrink)

This proposal introduces a novel decision-making framework to advance safe economic activities in cyberspace. We focus on identifying anomalies within crypto-asset trading, recognized as potential sources of criminal activity, severely undermining the credibility of such assets. Detecting and mitigating such anomalies holds significant societal implications, particularly in fostering trust within blockchain networks. We aim to bolster the “social trust” inherent to blockchain technology by facilitating informed economic activities in cyberspace. To achieve this, we propose integrating two artificial intelligence (AI) systems (...) into a blockchain-based decentralized autonomous organization (DAO). The first AI application involves amalgamating various anomaly indicators, spanning from cluster coefficient, entropy, triangular motif analysis, correlation tensor analysis, loop component by Hodge decomposition, loop causality detection, network classification using graph Laplacian, and persistent homology analysis, into a comprehensive indicator using a Boltzmann machine. The second AI application entails deploying conversational AI to guide and support traders, aiding them in making informed trading decisions. This system is designed to alert DAO members to anomalies based on the integrated indicators, especially during massive price fluctuations. We operate under the assumption of close collaboration between governments, experts, traders, system developers, and operators to effectively organize DAOs. The primary technical challenge in our proposal lies in developing a wallet assisted by an intelligent software agent capable of safe interactions with traders within a unified DAO. With this organization, we envision fostering a global economic ecosystem where physical and cyber worlds converge, allowing democratic economic participation. (shrink)

This work aims to explore the application of deep learning-based artificial intelligence technology in sentencing, to promote the reform and innovation of the judicial system. First, the concept and the principles of sentencing are introduced, and the deep learning model of intelligent robot in trials is proposed. According to related concepts, the issues that need to be solved in artificial intelligence sentencing based on deep learning are introduced. The deep learning model is integrated into the intelligent robot system, to assist (...) in the sentencing of cases. Finally, an example is adopted to illustrate the feasibility of the intelligent robot under deep learning in legal sentencing. The results show that the general final trial periods for cases of traffic accidents, copyright information, trademark infringement, copyright protection, and theft are 1,049, 796, 663, 847, and 201 days, respectively; while the final trial period under artificial intelligence evaluation based on the restricted Boltzmann deep learning model is 458, 387, 376, 438, and 247 days, respectively. The accuracy of trials is above 92%, showing a high application value. It can be observed that expect theft cases, the final trial period for others cases has been effectively reduced. The intelligent robot assistance under the restricted Boltzmann deep learning model can shorten the trial period of cases. The deep learning intelligent robot has a certain auxiliary role in legal sentencing, and this outcome provides a theoretical basis for the research of artificial intelligence technology in legal sentencing. (shrink)

The ‘Turing barrier’ is an evocative image for 0′, the degree of the unsolvability of the halting problem for Turing machines—equivalently, of the undecidability of Peano Arithmetic. The ‘barrier’ metaphor conveys the idea that effective computability is impaired by restrictions that could be removed by infinite methods. Assuming that the undecidability of PA is essentially depending on the finite nature of its computational means, decidability would be restored by the ω-rule. Hypercomputation, the hypothetical realization of infinitary machines through (...) relativistic and quantium models, would thus be capable of breaking the Turing barrier. The speculation is unfounded in principle, apart from issues of physical realizability: The point is that the ω-rule does not cope with all objects entailing the undecidability of PA. As long as the system is consistent, the computational boundary can be established by paradox-like constructions, such as Gödel-Rosser’s and Yablo’s, which are refractory to infinite induction, and do stand as the barrier’s buttresses. (shrink)

There is an existing debate regarding the view that technological instruments, devices, or machines can assert ‎or testify. A standard view in epistemology is that only humans can testify. However, the notion of quasi-‎testimony acknowledges that technological devices can assert or testify under some conditions, without ‎denying that humans and machines are not the same. Indeed, there are four relevant differences between ‎humans and instruments. First, unlike humans, machine assertion is not imaginative or playful. Second, ‎machine assertion is (...) prescripted and context restricted. As such, computers currently cannot easily switch ‎contexts or make meaningful relevant assertions in contexts for which they were not programmed. Third, ‎while both humans and computers make errors, they do so in different ways. Computers are very sensitive to ‎small errors in input, which may cause them to make big errors in output. Moreover, automatic error control ‎is based on finding irregularities in data without trying to establish whether they make sense. Fourth, ‎testimony is produced by a human with moral worth, while quasi-testimony is not. Ultimately, the notion of ‎quasi-testimony can serve as a bridge between different philosophical fields that deal with instruments and ‎testimony as sources of knowledge, allowing them to converse and agree on a shared description of reality, ‎while maintaining their distinct conceptions and ontological commitments about knowledge, humans, and ‎nonhumans.‎. (shrink)

We study the finitely repeated prisoner’s dilemma in which the players are restricted to choosing strategies which are implementable by a machine with a bound on its complexity. One player has to use a finite automaton while the other player has to use a finite perceptron. Some examples illustrate that the sets of strategies which are induced by these two types of machines are different and not ordered by set inclusion. Repeated game payoffs are evaluated according to the (...) limit of means. The main result establishes that a cooperation at almost all stages of the game is an equilibrium outcome if the complexity of the machines the players may use is limited enough and if the length T of the repeated game is sufficiently large. This result persists when more than T states are allowed in the player’s automaton. We further consider a variant of the model in which the two players are restricted to choosing strategies which are implementable by perceptrons and prove that the players can cooperate at most of the stages provided that the complexity of their perceptrons is sufficiently reduced. (shrink)

Machine ethics is the field that strives to develop ‘artificial moral agents’ (AMAs), artificial systems that can autonomously make moral decisions. Some authors have questioned the feasibility of machine ethics, by questioning whether artificial systems can possess moral competence, or the capacity to reach morally right decisions in various situations. This paper explores this question by drawing on the work of several moral philosophers (McDowell, Wiggins, Hampshire, and Nussbaum) who have characterised moral competence in a manner inspired by Aristotle. Although (...) disparate in many ways, these philosophers all emphasise what may be called ‘moral sensitivity’ as a precondition for moral competence. Moral sensitivity is the uncodified, practical skill to recognise, in a range of situations, which features of the situations are morally relevant, and how they are relevant. This paper argues that the main types of AMAs currently proposed are incapable of full moral sensitivity. First, top-down AMAs that proceed from fixed rule-sets are too rigid to respond appropriately to the wide range of qualitatively unique factors that moral sensitivity gives access to. Second, bottom-up AMAs that learn moral behaviour from examples are at risk of generalising from these examples in undesirable ways, as they lack embedding in what Wittgenstein calls a ‘form of life’, which allows humans to appropriately learn from moral examples. The paper concludes that AMAs are unlikely to possess full moral competence, but closes by suggesting that they may still be feasible in restricted domains of public morality, where moral sensitivity plays a smaller role. (shrink)

Oxford philosopher Nick Bostrom; in his recent and celebrated book Superintelligence; argues that advanced AI poses a potentially major existential risk to humanity; and that advanced AI development should be heavily regulated and perhaps even restricted to a small set of government-approved researchers. Bostrom’s ideas and arguments are reviewed and explored in detail; and compared with the thinking of three other current thinkers on the nature and implications of AI: Eliezer Yudkowsky of the Machine Intelligence Research Institute ; and (...) David Weinbaum and Viktoras Veitas of the Global Brain Institute. Relevant portions of Yudkowsky’s book Rationality: From AI to Zombies are briefly reviewed; and it is found that nearly all the core ideas of Bostrom’s work appeared previously or concurrently in Yudkowsky’s thinking. However; Yudkowsky often presents these shared ideas in a more plain-spoken and extreme form; making clearer the essence of what is being claimed. For instance; the elitist strain of thinking that one sees in the background in Bostrom is plainly and openly articulated in Yudkowsky; with many of the same practical conclusions. Bostrom and Yudkowsky view intelligent systems through the lens of reinforcement learning – they view them as “reward-maximizers” and worry about what happens when a very powerful and intelligent reward-maximizer is paired with a goal system that gives rewards for achieving foolish goals like tiling the universe with paperclips. Weinbaum and Veitas’s recent paper “Open-Ended Intelligence” presents a starkly alternative perspective on intelligence; viewing it as centered not on reward maximization; but rather on complex self-organization and self-transcending development that occurs in close coupling with a complex environment that is also ongoingly self-organizing; in only partially knowable ways. It is concluded that Bostrom and Yudkowsky’s arguments for existential risk have some logical foundation; but are often presented in an exaggerated way. For instance; formal arguments whose implication is that the “worst case scenarios” for advanced AI development are extremely dire; are often informally discussed as if they demonstrated the likelihood; rather than just the possibility; of highly negative outcomes. And potential dangers of reward-maximizing AI are taken as problems with AI in general; rather than just as problems of the reward-maximization paradigm as an approach to building superintelligence. If one views past; current; and future intelligence as “open-ended;” in the vernacular of Weaver and Veitas; the potential dangers no longer appear to loom so large; and one sees a future that is wide-open; complex and uncertain; just as it has always been. (shrink)

The ability of machines to make truly independent and autonomous decisions is a goal of many, not least of military leaders who wish to take the human out of the loop as much as possible, claiming that autonomous military weaponry—most notably drones—can make decisions more quickly and with greater accuracy. However, there is no clear understanding of how autonomous weapons should be conceptualized and of the implications that their “autonomous” nature has on them as ethical agents. It will be (...) argued that autonomous weapons are not full ethical agents due to the restrictions of their coding. However, the highly complex machine-learning nature gives the impression that they are making their own decisions and creates the illusion that their human operators are protected from the responsibility of the harm they cause. Therefore, it is important to distinguish between autonomous AI weapons and an AI with autonomy, a distinction that creates two different ethical problems for their use. For autonomous weapons, their limited agency combined with machine-learning means their human counterparts are still responsible for their actions while having no ability to control or intercede in the actual decisions made. If, on the other hand, an AI could reach the point of autonomy, the level of critical reflection would make its decisions unpredictable and dangerous in a weapon. (shrink)

BACKGROUND: Ontologies play a major role in life sciences, enabling a number of applications, from new data integration to knowledge verification. SNOMED CT is a large medical ontology that is formally defined so that it ensures global consistency and support of complex reasoning tasks. Most biomedical ontologies and taxonomies on the other hand define concepts only textually, without the use of logic. Here, we investigate how to automatically generate formal concept definitions from textual ones. We develop a method that uses (...) machine learning in combination with several types of lexical and semantic features and outputs formal definitions that follow the structure of SNOMED CT concept definitions. RESULTS: We evaluate our method on three benchmarks and test both the underlying relation extraction component as well as the overall quality of output concept definitions. In addition, we provide an analysis on the following aspects: How do definitions mined from the Web and literature differ from the ones mined from manually created definitions, e.g., MeSH? How do different feature representations, e.g., the restrictions of relations' domain and range, impact on the generated definition quality?, How do different machine learning algorithms compare to each other for the task of formal definition generation?, and, What is the influence of the learning data size to the task? We discuss all of these settings in detail and show that the suggested approach can achieve success rates of over 90%. In addition, the results show that the choice of corpora, lexical features, learning algorithm and data size do not impact the performance as strongly as semantic types do. Semantic types limit the domain and range of a predicted relation, and as long as relations' domain and range pairs do not overlap, this information is most valuable in formalizing textual definitions. CONCLUSIONS: The analysis presented in this manuscript implies that automated methods can provide a valuable contribution to the formalization of biomedical knowledge, thus paving the way for future applications that go beyond retrieval and into complex reasoning. The method is implemented and accessible to the public from: https://github.com/alifahsyamsiyah/learningDL. (shrink)

The Probably Approximately Correct (PAC) learning is a machine learning model introduced by Leslie Valiant in 1984. The PACi reducibility refers to the PAC reducibility independent of size and computation time. This reducibility in PAC learning resembles the reducibility in Turing computability. The ordering of concept classes under PAC reducibility is nonlinear, even when restricted to particular concrete examples.Due to the resemblance to Turing Reducibility, we suspected that there could be incomparable PACi and PAC degrees for the PACi and (...) PAC reducibilities as in Turing incomparable degrees. In 1957 Friedberg and in 1956 Muchnik independently solved the Post problem by constructing computably enumerable sets A and B of incomparable degrees using the priority construction method. We adapt this idea to PACi and PAC reducibilities and construct two effective concept classes C and D such that C is not reducible to D and vice versa. When considering PAC reducibility it was necessary to work on the size of an effective concept class, thus we use Kolmogorov complexity to obtain the size. The non-learnability of concept classes in the PAC learning model is explained by the existence of PAC incomparable degrees.Analogous to the Turing jump, we give a jump operation on effective concept classes for the zero jump. To define the zero jump operator for PACi degrees the join of all the effective concept classes is constructed and proved that it is a greatest element. There are many properties proven for existing degrees. Thus we can explore proving those properties to PACi and PAC degrees. But if we prove an embedding from those degrees to PACi and PAC degrees then those properties will be true for PACi and PAC degrees without explicitly proving them.Abstract prepared by Dodamgodage Gihnee M. Senadheera and taken directly from the thesisE-mail: [email protected]: https://www.proquest.com/docview/2717762461/abstract/ACD19F29A8774AF6PQ/1?accountid=13864. (shrink)

The notion of an automaton, as it is employed in the natural philosophy of Descartes and his closest followers, has three main components. None of them is new; what is new in early modern philosophy is the uses to which this old notion is put, and the idiosyncrasies into which its components are combined by subsequent philosophers. The thaumaturgic element is never entirely suppressed; but the more down-to-earth usage exemplified in antiquity by Aristotle’s references predominates. The automaton is quite often (...) the opposite of wonderful: phenomena that might excited wonder are proved to be unworthy of it, just by showing that they are the productions of an automaton. The automaton is, first of all, a machine, and therefore an artifact, human or (if the metaphor becomes literal) divine. It offers a model of intelligibility—to use Peter Dear’s term—for a certain class of natural phenomena, namely those we find in living things. But Descartes wants from it something more. In Descartes’ usage, a machine is that which makes itself available not just to “mechanical” explanation, but to a complete explanation, an explanation that makes all others superfluous. Not all machines, of course, are automata. Two further components figure in the notion. One is that automata are typically imitative. From the automata thaumata of Aristotle’s De Motu animalium to the nymphs of Salomon de Caus’s fountains, many automata are likenesses, partaking both of the iconic (to use Peirce’s term) and the symbolic. Moving sculptures proceed moving pictures by over two thousand years. They succeeded if they convinced their audiences that they could do what their prototypes did—where the doing is typically restricted to some few sorts of act. The wind-up mouse skitters across the floor like a real mouse; but it does not eat, nor does it seek the.. (shrink)