Context Of Machine Learning Research Articles

Composing recipes based on nutrients in food in a machine learning context

In the past few decades, numerous studies have demonstrated the effectiveness of diet modification in preventing and controlling different chronic diseases, bringing increased attention to the creation of nutritional recipes. Currently, the development of recipes is primarily based on personal experience rather than nutritional specifications. To this end, this paper first uses kernel canonical correlation analysis to demonstrate the relationship between nutrients in food and recipes. Based on this relationship, a new recipe expression method is put forward, which objectively reflects differing importance of an ingredient in differing recipes. Then, recipes are composed using an auto-encoder in deep learning, and a fusion model of two auto-encoders is proposed for better concocting recipes. This paper uses two machine learning methods, namely, non-negative matrix factorization and two-step regularized least squares, to form recipes. To tackle overfitting and instability in non-negative matrix factorization during the training of recipe model, we introduce the Frobenius norm to redefine the objective function and add non-smooth sparse matrices. Similar food has similar taste, but their nutrients might differ. This paper also considers nutrients as a kernel matrix of the two-step regularized least squares, which can effectively avoid the occurrence of different food combinations with similar taste. Experimental results show that developing recipes based on nutrients in food is feasible and effective in the context of machine learning.

Adaptive Context-Aware Energy Optimization for Services on Mobile Devices with Use of Machine Learning

In this paper we present an original adaptive task scheduling system, which optimizes the energy consumption of mobile devices using machine learning mechanisms and context information. The system learns how to allocate resources appropriately: how to schedule services/tasks optimally between the device and the cloud, which is especially important in mobile systems. Decisions are made taking the context into account (e.g. network connection type, location, potential time and cost of executing the application or service). In this study, a supervised learning agent architecture and service selection algorithm are proposed to solve this problem. Adaptation is performed online, on a mobile device. Information about the context, task description, the decision made and its results such as power consumption are stored and constitute training data for a supervised learning algorithm, which updates the knowledge used to determine the optimal location for the execution of a given type of task. To verify the solution proposed, appropriate software has been developed and a series of experiments have been conducted. Results show that as a result of the experience gathered and the learning process performed, the decision module has become more efficient in assigning the task to either the mobile device or cloud resources.

Finding True Clusters: On the Importance of Simplicity in Science

The main point of this paper is to underscore the link between simplicity and truth in an unsupervised machine learning context. More precisely, we argue that parametric and dimensional simplicity are not indicators of truth but the methodological principle that urges us to pay attention to such notions of simplicity is truth conducive. The truth that we are looking for are specific geometrical shapes and we know which algorithm can find which shapes provided that we pay attention to parametric and dimensional simplicity.

Constitutional Rights in the Machine Learning State

A new class of “machine learning” tools is able to make better predictions and inferences from data than was previously feasible. For the state, machine learning is a powerful and supple device to reveal citizens’ beliefs, actions, and expected behaviors. Its deployment to allocate investigative resources, material benefits, and coercive penalties to particular individuals, though, can implicate due process, privacy, and equality interests. Substantive doctrinal frameworks and enforcement regimes for those entitlements, however, arose in the context of human action. Neither is apt for a machine learning context. This Article offers a start to the larger project of developing a more general account of substantive rules and enforcement mechanisms to promote due process, privacy, and equality norms in the machine learning state. After cataloging notable state and municipal adoptions of machine learning tools, it considers how existing constitutional norms can be re-calibrated (in the case of due process and equality) or retooled (in the case of privacy). It further reexamines the enforcement regime for constitutional interests. Today, constitutional rights are (largely) enforced through discrete, individual legal actions. Machine learning’s normative implications arise from systemic design choices. The retail enforcement mechanisms that currently dominate the constitutional remedies context are therefore particularly inapt. Instead, a careful mix of ex ante regulation and ex post aggregate litigation, which are necessary complements, is more desirable.

Identifying Pareto-based solutions for regression subset selection via a feasible solution algorithm

The concept of Pareto optimality has been utilized in fields such as engineering and economics to understand fluid dynamics and consumer behavior. In machine learning contexts, Pareto-optimality has been used to identify tuning parameters that best optimize a set of m criteria (multi-objective optimization). During the process of regression model selection, data scientists are often concerned with choosing a model which has the best single criterion (e.g., Akaike information criterion (AIC) or R-squared ( $$R^2$$ )) before continuing to check a number of other regression model characteristics (e.g., model size, form, diagnostics, and interpretability). This strategy is multi-objective in nature but single objective in its numeric execution. This paper will first introduce a feasible solution algorithm (FSA) and explain how it can be applied to multi-objective problems for regression subset selection. Then we introduce the general framework of Pareto optimality within the regression setting. We then apply the algorithm in a simulation setting where we seek to estimate the first four Pareto boundaries for regression models using two model fit criteria. Finally, we present an application where we use a US communities and crime dataset.

A Natural Representation of Functions that Facilitates `Exact Learning'

We present a collection of mathematical tools and emphasise a fundamental representation of analytic functions. Connecting these concepts leads to a framework for `exact learning', where an unknown numeric distribution could in principle be assigned an exact mathematical description. This is a new perspective on machine learning with potential applications in all domains of the mathematical sciences and the generalised representations presented here have not yet been widely considered in the context ofmachine learning and data analysis. The moments of a multivariate function or distribution are extracted using a Mellin transform and the generalised form of the coefficients is trained assuming a highly generalised Mellin-Barnes integral representation. The fit functions use many fewer parameters contemporary machine learning methods and any implementation that connects these concepts successfully will likely carry across to non-exact problems and provide approximate solutions. We compare the equations for theexact learning method with those for a neural network which leads to a new perspective on understanding what a neural network may be learning and how to interpret the parameters of those networks.

On the utility of dreaming: A general model for how learning in artificial agents can benefit from data hallucination

We consider the benefits of dream mechanisms – that is, the ability to simulate new experiences based on past ones – in a machine learning context. Specifically, we are interested in learning for artificial agents that act in the world, and operationalize “dreaming” as a mechanism by which such an agent can use its own model of the learning environment to generate new hypotheses and training data. We first show that it is not necessarily a given that such a data-hallucination process is useful, since it can easily lead to a training set dominated by spurious imagined data until an ill-defined convergence point is reached. We then analyse a notably successful implementation of a machine learning-based dreaming mechanism by Ha and Schmidhuber (Ha, D., & Schmidhuber, J. (2018). World models. arXiv e-prints, arXiv:1803.10122). On that basis, we then develop a general framework by which an agent can generate simulated data to learn from in a manner that is beneficial to the agent. This, we argue, then forms a general method for an operationalized dream-like mechanism. We finish by demonstrating the general conditions under which such mechanisms can be useful in machine learning, wherein the implicit simulator inference and extrapolation involved in dreaming act without reinforcing inference error even when inference is incomplete.

Machine Learning Models Reveal Key Performance Metrics of Football Players to Win Matches in Qatar Stars League

As football (soccer) is one of the most popular sports worldwide, winning football matches is becoming an essential aspect of football clubs. In this study, we analyzed football players' performance in a total of 864 football matches of the Qatar Stars League (QSL) between the years 2012 and 2019. For each match, the collective performance of the players in key playing positions was analyzed to understand their effectiveness in winning games. We formulated this study as a classification framework in the machine learning (ML) context to distinguish the winning team from the losing team in a match. This allowed us to check the effectiveness of different performance metrics considered a feature vector for ML models. Different ML models were considered for this classification task, and the logistic regression-based model was considered the best performing model, with more than 80% accuracy. Multiple feature selection methods were leveraged to identify players' performance metrics that could be considered as contributing factors to determine the match result. The proposed ML model identified several features, including (a) shots on target by forwarders (b) distance covered by forwarders and midfielders at very high speed (c) successful passes, that can be considered as effective performance metrics for winning a football match in QSL. Interestingly, we revealed that the defenders' role could not be ignored for match results, and playing fair games improves the chance of winning matches in QSL. We also showed that players' performance metrics from the last five seasons would provide sufficient discriminative power to the proposed ML model to predict the match-winner in the upcoming season. The proposed ML model will support the players, coaching staff, and team management to focus on specific performance metrics that may lead to winning a match in QSL.

Algorithmic Decision-Making and the Control Problem

The danger of human operators devolving responsibility to machines and failing to detect cases where they fail has been recognised for many years by industrial psychologists and engineers studying the human operators of complex machines. We call it “the control problem”, understood as the tendency of the human within a human–machine control loop to become complacent, over-reliant or unduly diffident when faced with the outputs of a reliable autonomous system. While the control problem has been investigated for some time, up to this point its manifestation in machine learning contexts has not received serious attention. This paper aims to fill that gap. We argue that, except in certain special circumstances, algorithmic decision tools should not be used in high-stakes or safety-critical decisions unless the systems concerned are significantly “better than human” in the relevant domain or subdomain of decision-making. More concretely, we recommend three strategies to address the control problem, the most promising of which involves a complementary (and potentially dynamic) coupling between highly proficient algorithmic tools and human agents working alongside one another. We also identify six key principles which all such human–machine systems should reflect in their design. These can serve as a framework both for assessing the viability of any such human–machine system as well as guiding the design and implementation of such systems generally.

On the S-instability and degeneracy of discrete deep learning models

AbstractA probability model exhibits instability if small changes in a data outcome result in large and, often unanticipated, changes in probability. This instability is a property of the probability model, given by a distributional form and a given configuration of parameters. For correlated data structures found in several application areas, there is increasing interest in identifying such sensitivity in model probability structure. We consider the problem of quantifying instability for general probability models defined on sequences of observations, where each sequence of length $N$ has a finite number of possible values that can be taken at each point. A sequence of probability models, indexed by $N$, and an associated parameter sequence result to accommodate data of expanding dimension. Model instability is formally shown to occur when a certain log probability ratio under such models grows faster than $N$. In this case, a one component change in the data sequence can shift probability by orders of magnitude. Also, as instability becomes more extreme, the resulting probability models are shown to tend to degeneracy, placing all their probability on potentially small portions of the sample space. These results on instability apply to large classes of models commonly used in random graphs, network analysis and machine learning contexts.

Putting a bug in ML: The moth olfactory network learns to read MNIST

We seek to (i) characterize the learning architectures exploited in biological neural networks for training on very few samples, and (ii) port these algorithmic structures to a machine learning context. The moth olfactory network is among the simplest biological neural systems that can learn, and its architecture includes key structural elements and mechanisms widespread in biological neural nets, such as cascaded networks, competitive inhibition, high intrinsic noise, sparsity, reward mechanisms, and Hebbian plasticity. These structural biological elements, in combination, enable rapid learning.MothNet is a computational model of the moth olfactory network, closely aligned with the moth’s known biophysics and with in vivo electrode data collected from moths learning new odors. We assign this model the task of learning to read the MNIST digits. We show that MothNet successfully learns to read given very few training samples (1–10 samples per class). In this few-samples regime, it outperforms standard machine learning methods such as nearest-neighbors, support-vector machines, and neural networks (NNs), and matches specialized one-shot transfer-learning methods but without the need for pre-training.The MothNet architecture illustrates how algorithmic structures derived from biological brains can be used to build alternative NNs that may avoid the high training data demands of many current engineered NNs.

Fairness, accountability and transparency: notes on algorithmic decision-making in criminal justice

AbstractOver the last few years, legal scholars, policy-makers, activists and others have generated a vast and rapidly expanding literature concerning the ethical ramifications of using artificial intelligence, machine learning, big data and predictive software in criminal justice contexts. These concerns can be clustered under the headings of fairness, accountability and transparency. First, can we trust technology to be fair, especially given that the data on which the technology is based are biased in various ways? Second, whom can we blame if the technology goes wrong, as it inevitably will on occasion? Finally, does it matter if we do not know how an algorithm works or, relatedly, cannot understand how it reached its decision? I argue that, while these are serious concerns, they are not irresolvable. More importantly, the very same concerns of fairness, accountability and transparency apply, with even greater urgency, to existing modes of decision-making in criminal justice. The question, hence, is comparative: can algorithmic modes of decision-making improve upon the status quo in criminal justice? There is unlikely to be a categorical answer to this question, although there are some reasons for cautious optimism.

Machine Learning in Context, or Learning from LANDR: Artificial Intelligence and the Platformization of Music Mastering

This article proposes a contextualist approach to machine learning and aesthetics, using LANDR, an online platform that offers automated music mastering and that trumpets its use of supervised machine learning, branded as artificial intelligence (AI). Increasingly, machine learning will become an integral part of the processing of sounds and images, shaping the way our culture sounds, looks, and feels. Yet we cannot know exactly how much of a role or what role machine learning plays in LANDR. To parochialize the machine learning part of what LANDR does, this study spirals in from bigger contexts to smaller ones: LANDR’s place between the new media industry and the mastering industry; the music scene in their home city, Montreal, Quebec; LANDR use by DIY musicians and independent engineers; and, finally, the LANDR interface and the sound it produces in use. While LANDR claims to automate the work of mastering engineers, it appears to expand and morph the definition of mastering itself: it devalues people’s aesthetic labor as it establishes higher standards for recordings online. And unlike many other new media firms, LANDR’s connection to its local music scene has been essential to its development, growth, and authority, even as they have since moved on from that scene, and even as the relationship was never fully reciprocal.

Machine learning in empirical asset pricing

The tremendous speedup in computing in recent years, the low data storage costs of today, the availability of “big data” as well as the broad range of free open-source software, have created a renaissance in the application of machine learning techniques in science. However, this new wave of research is not limited to computer science or software engineering anymore. Among others, machine learning tools are now used in financial problem settings as well. Therefore, this paper mentions a specific definition of machine learning in an asset pricing context and elaborates on the usefulness of machine learning in this context. Most importantly, the literature review gives the reader a theoretical overview of the most recent academic studies in empirical asset pricing that employ machine learning techniques. Overall, the paper concludes that machine learning can offer benefits for future research. However, researchers should be critical about these methodologies as machine learning has its pitfalls and is relatively new to asset pricing.

Multiple classifiers fusion and CNN feature extraction for handwritten digits recognition

Handwritten digits recognition has been treated as a multi-class classification problem in the machine learning context, where each of the ten digits (0–9) is viewed as a class and the machine learning task is essentially to train a classifier that can effectively discriminate the ten classes. In practice, it is very usual that the performance of a single classifier trained using a standard learning algorithm is varied on different datasets, which indicates that the same learning algorithm may train strong classifiers on some datasets but weak classifiers may be trained on other datasets. It is also possible that the same classifier shows different performance on different test sets, especially when considering the case that image instances can be highly diverse due to the different handwriting styles of different people on the same digits. To address the above issue, development of ensemble learning approaches have been very necessary to improve the overall performance and make the performance more stable on different datasets. In this paper, we propose a framework that involves CNN-based feature extraction from the MINST dataset and algebraic fusion of multiple classifiers trained on different feature sets, which are prepared through feature selection applied to the original feature set extracted using CNN. The experimental results show that the classifiers fusion can achieve the classification accuracy of ≥ 98%.

Feature Selection for Binary Classification Within Functional Genomics Experiments via Interquartile Range and Clustering

Datasets produced in modern research, such as biomedical science, pose a number of challenges for machine learning techniques used in binary classification due to high dimensionality. Feature selection is one of the most important statistical techniques used for dimensionality reduction of the datasets. Therefore, techniques are needed to find an optimal number of features to obtain more desirable learning performance. In the machine learning context, gene selection is treated as a feature selection problem, the objective of which is to find a small subset of the most discriminative features for the target class. In this paper, a gene selection method is proposed that identifies the most discriminative genes in two stages. Genes that unambiguously assign the maximum number of samples to their respective classes using a greedy approach are selected in the first stage. The remaining genes are divided into a certain number of clusters. From each cluster, the most informative genes are selected via the lasso method and combined with genes selected in the first stage. The performance of the proposed method is assessed through comparison with other state-of-the-art feature selection methods using gene expression datasets. This is done by applying two classifiers i.e., random forest and support vector machine, on datasets with selected genes and training samples and calculating their classification accuracy, sensitivity, and Brier score on samples in the testing part. Boxplots based on the results and correlation matrices of the selected genes are thenceforth constructed. The results show that the proposed method outperforms the other methods.

Enhance PATE on Complex Tasks With Knowledge Transferred From Non-Private Data

Privacy protection is considered as an important problem in learning-based systems. Recently, various works based on differential privacy have been proposed to protect an individual's privacy in the machine learning and deep learning contexts. One of the state-of-the-art approaches is Private Aggregation of Teacher Ensembles (PATE), a generic framework which can be successfully applied to many different learning algorithms. In PATE, we need to split the private dataset into many disjoint subsets and train an ensemble of teachers on these subsets. Then, we transfer noisy predictions from the ensemble of teachers to a student model. In this paper, we show that for complex datasets and tasks, such as nature image classification, the training set allocated for one teacher may be too small with respect to the corresponding task to achieve an ideal performance. To alleviate this problem, we propose the TrPATE framework which extends PATE with transfer learning. Based on PATE, we transfer and share the knowledge extracted from a publicly available non-private dataset to the teachers. The extensive experiments are conducted on various datasets, and the empirical results demonstrate the effectiveness of our method.

Combining Ensemble Learning with a Fragment-Based Topological Approach To Generate New Molecular Diversity in Drug Discovery: In Silico Design of Hsp90 Inhibitors.

Machine learning methods have revolutionized modern science, providing fast and accurate solutions to multiple problems. However, they are commonly treated as “black boxes”. Therefore, in important scientific fields such as medicinal chemistry and drug discovery, machine learning methods are restricted almost exclusively to the task of performing predictions of large and heterogeneous data sets of chemicals. The lack of interpretability prevents the full exploitation of the machine learning models as generators of new chemical knowledge. This work focuses on the development of an ensemble learning model for the prediction and design of potent dual heat shock protein 90 (Hsp90) inhibitors. The model displays accuracy higher than 80% in both training and test sets. To use the ensemble model as a generator of new chemical knowledge, three steps were followed. First, a physicochemical and/or structural interpretation was provided for each molecular descriptor present in the ensemble learning model. Second, the term “pseudolinear equation” was introduced within the context of machine learning to calculate the relative quantitative contributions of different molecular fragments to the inhibitory activity against the two Hsp90 isoforms studied here. Finally, by assembling the fragments with positive contributions, new molecules were designed, being predicted as potent Hsp90 inhibitors. According to Lipinski’s rule of five, the designed molecules were found to exhibit potentially good oral bioavailability, a primordial property that chemicals must have to pass early stages in drug discovery. The present approach based on the combination of ensemble learning and fragment-based topological design holds great promise in drug discovery, and it can be adapted and applied to many different scientific disciplines.

Machine Learning: Deepest Learning as Statistical Data Assimilation Problems.

We formulate an equivalence between machine learning and the formulation of statistical data assimilation as used widely in physical and biological sciences. The correspondence is that layer number in a feedforward artificial network setting is the analog of time in the data assimilation setting. This connection has been noted in the machine learning literature. We add a perspective that expands on how methods from statistical physics and aspects of Lagrangian and Hamiltonian dynamics play a role in how networks can be trained and designed. Within the discussion of this equivalence, we show that adding more layers (making the network deeper) is analogous to adding temporal resolution in a data assimilation framework. Extending this equivalence to recurrent networks is also discussed. We explore how one can find a candidate for the global minimum of the cost functions in the machine learning context using a method from data assimilation. Calculations on simple models from both sides of the equivalence are reported. Also discussed is a framework in which the time or layer label is taken to be continuous, providing a differential equation, the Euler-Lagrange equation and its boundary conditions, as a necessary condition for a minimum of the cost function. This shows that the problem being solved is a two-point boundary value problem familiar in the discussion of variational methods. The use of continuous layers is denoted "deepest learning." These problems respect a symplectic symmetry in continuous layer phase space. Both Lagrangian versions and Hamiltonian versions of these problems are presented. Their well-studied implementation in a discrete time/layer, while respecting the symplectic structure, is addressed. The Hamiltonian version provides a direct rationale for backpropagation as a solution method for a certain two-point boundary value problem.

You are smarter than you think: (super) machine learning in context.

You are smarter than you think: (super) machine learning in context.