Search

Our societies have now gone almost fully digital, which is bound to impact the ways in which the self apprehends itself, constructs itself and takes part in society. This collective essay draws upon philosophy, logics, literature and media studies to explore a variety of responses to what is currently at stake in the recomposition of the subject through our daily digital experience. Attempting to chart the digital traces human beings leave behind them, in writing or in archives, it situates them in the larger political, economic and financial frame of our times.

In the literary machine that is known as Finnegans Wake (1939), James Joyce has coined words and phrases that border wildly on nonsense, non-words, or linguistic idiosyncrasies characteristic of schizophrenia. The reader may get a taste of this from one of his sentences which goes: “the circumflexous wall of a singleminded men’s asylum, accentuated by bi tso fb rok engl a ssan dspl itch ina, —Yard inquiries pointed out — that they ad bîn ‘provoked’ ay ^ fork, of à grave Brofèsor; àth é’s Brèa...

Machine learning (ML) techniques are increasingly prevalent in education, from their use in predicting student dropout, to assisting in university admissions, and facilitating the rise of MOOCs. Given the rapid growth of these novel uses, there is a pressing need to investigate how ML techniques support long-standing education principles and goals. In this work, we shed light on this complex landscape drawing on qualitative insights from interviews with education experts. These interviews comprise in-depth evaluations of ML for education (ML4Ed) papers published in preeminent applied ML conferences over the past decade. Our central research goal is to critically examine how the stated or implied education and societal objectives of these papers are aligned with the ML problems they tackle. That is, to what extent does the technical problem formulation, objectives, approach, and interpretation of results align with the education problem at hand. We find that a cross-disciplinary gap exists and is particularly salient in two parts of the ML life cycle: the formulation of an ML problem from education goals and the translation of predictions to interventions. We use these insights to propose an extended ML life cycle, which may also apply to the use of ML in other domains. Our work joins a growing number of meta-analytical studies across education and ML research, as well as critical analyses of the societal impact of ML. Specifically, it fills a gap between the prevailing technical understanding of machine learning and the perspective of education researchers working with students and in policy., 29 pages, 1 figure, 2 tables. Supplementary material available upon request

À quelles recompositions du sujet la numérisation croissante de nos sociétés conduit-elle ? Comment le sujet s’éprouve-t-il lui-même ? Comment se construit-il ? Comment forge-t-il sa place au monde, dans ce nouveau paysage informé par l’ubiquité numérique ? En croisant les points de vue de la philosophie, de la logique, des sciences des médias et de la littérature, l’ouvrage apporte un faisceau de réponses à ce questionnement contemporain. Il explore les traces que l’humain laisse de lui-même - écriture ou archive -, pour les articuler à des problématiques plus larges, artistiques, financières, politiques et économiques.

Dans la machine littéraire du nom de Finnegans Wake, James Joyce a inventé des mots et des expressions à l’extrême limite du non-sens : des non-mots, des idiosyncrasies linguistiques caractéristiques de la schizophrénie. Il suffira au lecteur de lire la phrase suivante pour en avoir un avant-goût : « the circumflexous wall of a singleminded men’s asylum, accentuated by bi tso fb rok engl a ssan dspl itch ina, —Yard inquiries pointed out — that they ad bîn ‘provoked’ ay ^ fork, of à grave Brof...

Additional file 2. Fig. S1. Tumor/NAT differences. A Consensus clustering of samples (n = 81, K = 4) using the top 25% most variable proteins (n = 1,193, K = 4). B Differentially abundant protein-coding RNAs in tumors and NATs from The Cancer Genome Atlas (TCGA). Statistically significant genes (FDR < 0.05) are colored in black. C Associations of protein abundance changes between tumor versus NAT, and mpMRI-visible NAT versus mpMRI-invisible NAT. Only proteins that were significantly differentially expressed in tumor and NAT regions (FDR < 0.05) were used for this analysis. D Associations of protein-coding RNA abundance changes between tumor versus NAT, and mpMRI-visible tumor versus mpMRI-invisible tumor. Only protein-coding RNAs that were significantly differentially expressed in tumor and NAT regions (FDR < 0.05) were used for this analysis. Proteins or transcripts that were significant (FDR < 0.05) in the tumor-NAT comparison and had the same directionality are marked in black. NAT: histologically normal prostate adjacent to the tumor; mpMRI: multiparametric magnetic resonance imaging; FDR: Benjamini-Hochberg-adjusted p-value.

Additional file 1 Methods. Method details.

Prostate cancer is the second most frequently diagnosed non-skin cancer in men worldwide. Patient outcomes are remarkably heterogeneous and the best existing clinical prognostic tools such as International Society of Urological Pathology Grade Group, pretreatment serum PSA concentration and T-category, do not accurately predict disease outcome for individual patients. Thus, patients newly diagnosed with prostate cancer are often overtreated or undertreated, reducing quality of life and increasing disease-specific mortality. Biomarkers that can improve the risk stratification of these patients are, therefore, urgently needed. The ideal biomarker in this setting will be non-invasive and affordable, enabling longitudinal evaluation of disease status. Prostatic secretions, urine and blood can be sources of biomarker discovery, validation and clinical implementation, and mass spectrometry can be used to detect and quantify proteins in these fluids. Protein biomarkers currently in use for diagnosis, prognosis and relapse-monitoring of localized prostate cancer in fluids remain centred around PSA and its variants, and opportunities exist for clinically validating novel and complimentary candidate protein biomarkers and deploying them into the clinic.

Strategic classification studies the design of a classifier robust to the manipulation of input by strategic individuals. However, the existing literature does not consider the effect of competition among individuals as induced by the algorithm design. Motivated by constrained allocation settings such as college admissions, we introduce strategic ranking, in which the (designed) individual reward depends on an applicant's post-effort rank in a measurement of interest. Our results illustrate how competition among applicants affects the resulting equilibria and model insights. We analyze how various ranking reward designs, belonging to a family of step functions, trade off applicant, school, and societal utility, as well as how ranking design counters inequities arising from disparate access to resources. In particular, we find that randomization in the reward design can mitigate two measures of disparate impact, welfare gap and access., Comment: 30 pages. To appear in the conference proceedings of AISTATS 2022

While real-world decisions involve many competing objectives, algorithmic decisions are often evaluated with a single objective function. In this paper, we study algorithmic policies which explicitly trade off between a private objective (such as profit) and a public objective (such as social welfare). We analyze a natural class of policies which trace an empirical Pareto frontier based on learned scores, and focus on how such decisions can be made in noisy or data-limited regimes. Our theoretical results characterize the optimal strategies in this class, bound the Pareto errors due to inaccuracies in the scores, and show an equivalence between optimal strategies and a rich class of fairness-constrained profit-maximizing policies. We then present empirical results in two different contexts -- online content recommendation and sustainable abalone fisheries -- to underscore the applicability of our approach to a wide range of practical decisions. Taken together, these results shed light on inherent trade-offs in using machine learning for decisions that impact social welfare.

We study two-sided matching markets in which one side of the market (the players) does not have a priori knowledge about its preferences for the other side (the arms) and is required to learn its preferences from experience. Also, we assume the players have no direct means of communication. This model extends the standard stochastic multi-armed bandit framework to a decentralized multiple player setting with competition. We introduce a new algorithm for this setting that, over a time horizon $T$, attains $\mathcal{O}(\log(T))$ stable regret when preferences of the arms over players are shared, and $\mathcal{O}(\log(T)^2)$ regret when there are no assumptions on the preferences on either side. Moreover, in the setting where a single player may deviate, we show that the algorithm is incentive compatible whenever the arms' preferences are shared, but not necessarily so when preferences are fully general., Comment: 34 pages

Tumor DNA sequencing data can be interpreted by computational methods that analyze genomic heterogeneity to infer evolutionary dynamics. A growing number of studies have used these approaches to link cancer evolution with clinical progression and response to therapy. Although the inference of tumor phylogenies is rapidly becoming standard practice in cancer genome analyses, standards for evaluating them are lacking. To address this need, we systematically assess methods for reconstructing tumor subclonality. First, we elucidate the main algorithmic problems in subclonal reconstruction and develop quantitative metrics for evaluating them. Then we simulate realistic tumor genomes that harbor all known clonal and subclonal mutation types and processes. Finally, we benchmark 580 tumor reconstructions, varying tumor read depth, tumor type and somatic variant detection. Our analysis provides a baseline for the establishment of gold-standard methods to analyze tumor heterogeneity.

Stable matching, a classical model for two-sided markets, has long been studied with little consideration for how each side's preferences are learned. With the advent of massive online markets powered by data-driven matching platforms, it has become necessary to better understand the interplay between learning and market objectives. We propose a statistical learning model in which one side of the market does not have a priori knowledge about its preferences for the other side and is required to learn these from stochastic rewards. Our model extends the standard multi-armed bandits framework to multiple players, with the added feature that arms have preferences over players. We study both centralized and decentralized approaches to this problem and show surprising exploration-exploitation trade-offs compared to the single player multi-armed bandits setting., 15 pages, 3 figures. A version appears in the Proceedings of The 23nd International Conference on Artificial Intelligence and Statistics (AISTATS), 2020

In photon-limited imaging, the pixel intensities are affected by photon count noise. Many applications, such as 3-D reconstruction using correlation analysis in X-ray free electron laser (XFEL) single molecule imaging, require an accurate estimation of the covariance of the underlying 2-D clean images. Accurate estimation of the covariance from low-photon count images must take into account that pixel intensities are Poisson distributed, hence the classical sample covariance estimator is sub-optimal. Moreover, in single molecule imaging, including in-plane rotated copies of all images could further improve the accuracy of covariance estimation. In this paper we introduce an efficient and accurate algorithm for covariance matrix estimation of count noise 2-D images, including their uniform planar rotations and possibly reflections. Our procedure, steerable $e$PCA, combines in a novel way two recently introduced innovations. The first is a methodology for principal component analysis (PCA) for Poisson distributions, and more generally, exponential family distributions, called $e$PCA. The second is steerable PCA, a fast and accurate procedure for including all planar rotations for PCA. The resulting principal components are invariant to the rotation and reflection of the input images. We demonstrate the efficiency and accuracy of steerable $e$PCA in numerical experiments involving simulated XFEL datasets and rotated Yale B face data.

We clarify what fairness guarantees we can and cannot expect to follow from unconstrained machine learning. Specifically, we characterize when unconstrained learning on its own implies group calibration, that is, the outcome variable is conditionally independent of group membership given the score. We show that under reasonable conditions, the deviation from satisfying group calibration is upper bounded by the excess risk of the learned score relative to the Bayes optimal score function. A lower bound confirms the optimality of our upper bound. Moreover, we prove that as the excess risk of the learned score decreases, it strongly violates separation and independence, two other standard fairness criteria. Our results show that group calibration is the fairness criterion that unconstrained learning implicitly favors. On the one hand, this means that calibration is often satisfied on its own without the need for active intervention, albeit at the cost of violating other criteria that are at odds with calibration. On the other hand, it suggests that we should be satisfied with calibration as a fairness criterion only if we are at ease with the use of unconstrained machine learning in a given application., 37 pages, 9 figures

Population risk is always of primary interest in machine learning; however, learning algorithms only have access to the empirical risk. Even for applications with nonconvex nonsmooth losses (such as modern deep networks), the population risk is generally significantly more well-behaved from an optimization point of view than the empirical risk. In particular, sampling can create many spurious local minima. We consider a general framework which aims to optimize a smooth nonconvex function $F$ (population risk) given only access to an approximation $f$ (empirical risk) that is pointwise close to $F$ (i.e., $\|F-f\|_{\infty} \le \nu$). Our objective is to find the $\epsilon$-approximate local minima of the underlying function $F$ while avoiding the shallow local minima---arising because of the tolerance $\nu$---which exist only in $f$. We propose a simple algorithm based on stochastic gradient descent (SGD) on a smoothed version of $f$ that is guaranteed to achieve our goal as long as $\nu \le O(\epsilon^{1.5}/d)$. We also provide an almost matching lower bound showing that our algorithm achieves optimal error tolerance $\nu$ among all algorithms making a polynomial number of queries of $f$. As a concrete example, we show that our results can be directly used to give sample complexities for learning a ReLU unit., Comment: To appear in NIPS 2018

Although the number of engineering graduates has expanded rapidly in the last two decades, relatively little is known about the quality of engineering programs worldwide. In particular, few studies look at differences in the degree to which students are learning skills across different engineering programs within and between countries. There is particular interest in the investigation of the engineering education quality in the countries with the rapidly growing economy, such as BRICS countries. Until now, there was little research in this field and one of the main reasons for this is the difficulty in developing an assessment approach and the accompanying set of instruments, which would allow for measurement and international comparison. Our study describes a set of procedures for developing such an assessment framework of instruments, to measure and compare skill levels and gains across engineering programs. We first describe a systematic approach for constructing cross-nationally comparable instruments in maths and physics for students in the first two years of their undergraduate engineering programs. The approach includes both a priori procedures (including expert assessments to avoid construct, method, and item bias), and a posteriori procedures (including the psychometric analysis of test quality, differential item functioning, and identifying and reducing bias in the data). In addition to describing this set of procedures in theory, we also show how we systematically implemented these procedures. Drawing on data that we collected from over 24 engineering experts and 3,600 engineering students across Russia and China, we provide evidence that it is possible to create tests that are cross-culturally valid, equate-able, and free from bias.

The article seeks to develop a new angle for translation studies by rethinking its relationship to the political. It begins with the question “Can the eventfulness of translation itself be thought?” Since neither the familiar model of communication (translatable and untranslatable) nor the biblical model of the Tower of Babel (the promise or withdrawal of meaning) can help us work out a suitable answer to that question, the author proposes an alternative method that incorporates the notions of temporality, difference, and competing universals in the reframing of translation. This method requires close attention to the multiple temporalities of translation in concrete analyses of translingual practices, or what the author calls “differentially distributed discursive practices across languages.” The author’s textual analysis focuses on a few pivotal moments of translation in global history—chosen for their world transforming influences or actual and potential global impact—to demonstrate what is meant by the “eventfulness of translation.” These include, for example, the nineteenth-century Chinese translation of Henry Wheaton’s "Elements of International Law" or "Wanguo gongfa", the post-World War II multilingual fashioning of the Universal Declaration of Human Rights with a focus on P. C. Chang’s unique contribution, and the Afro-Asian writers’ translation project during the Cold War.

Introduction: Contrast-induced nephropathy (CIN), defined as an increase in serum creatinine (SCr) greater than 25% or ≥0.5 mg/dL within 3 days of intravenous (IV) contrast administration in the absence of an alternative cause, is the third most common cause of new acute renal failure in hospitalized patients. It is known to increase in-hospital mortality up to 27%. The purpose of this study was to investigate the rate of outpatient follow up and the occurrence of CIN in patients who presented to the emergency department (ED) and were discharged home after computed tomography (CT) of the abdomen and pelvis (AP) with IV contrast. Methods: We conducted a single center retrospective review of charts for patients who required CT of AP with IV contrast and who were discharged home. Patients’ clinical data included the presence of diabetes mellitus, hypertension, chronic kidney disease (CKD) and congestive heart failure (CHF). Results: Five hundred and thirty six patients underwent CT of AP with IV contrast in 2011 and were discharged home. Diabetes mellitus was documented in 96 patients (18%). Hypertension was present in 141 patients (26.3%), and 82 patients (15.3%) were on angiotensin-converting-enzyme inhibitors (ACEI). Five patients (0.9%) had documented CHF and all of them were taking furosemide. Seventy patients (13%) had a baseline SCr >1.2 mg/dL. One hundred fifty patients (28%) followed up in one of the clinics or the ED within one week after discharge, but only 40 patients (7.5%) had laboratory workup. Out of 40 patients who followed up within 1 week after discharge, 9 patients (22.5%) developed CIN. One hundred ninety patients (35.4%) followed up in one of the clinics or the ED after 7 days and within 1 month after discharge, but only 71 patients (13.2%) had laboratory workup completed. Out of 71 patients who followed up within 1 month, 11 patients (15%) developed CIN. The overall incidence of CIN was 15.3% (17 out of 111 patients). Conclusion: There was a poor outpatient follow up after CT of AP with IV contrast and biochemically CIN appears to be present in some patients. Unlike previous reports that CKD is the major risk factor for CIN, our results demonstrated that risk factors such as advanced age, DM and hypertension seem to predispose patients to CIN rather than abnormal baseline SCr. [West J Emerg Med. 2014;15(3):276–281.]

Chance put the text of Edgar Allan Poe's story “The Purloined Letter” at the disposal of Jacques Lacan and his psychoanalytic work, and this work has since made numerous surprising moves and detours through poststructuralist literary criticism. These moves and detours are guarding an open secret as to how Lacan discovered Poe's story for psychoanalysis. The secret—hiding in plain sight, as it were—has inadvertently barred us from knowing more. That is to say, something will remain unseen and unheard until we are prepared to reflect on what we know about Lacan through American literary criticism and, more importantly, what we do not know about American cybernetics in France or in the U.S. for that matter.

Chromosomal instability (CIN) results in the accumulation of large-scale losses, gains, and rearrangements of DNA(1). The broad genomic complexity caused by CIN is a hallmark of cancer(2), however, there is no systematic framework to measure different types of CIN and their impact on clinical phenotypes pan-cancer. Here, we evaluate the extent, diversity and origin of chromosomal instability across 7,880 tumours representing 33 cancer types. We present a compendium of 17 copy number signatures characterising specific types of CIN, with putative aetiologies supported by multiple independent data sources. The signatures predict drug response and identify new drug targets. Our framework refines the understanding of impaired homologous recombination, one of the most therapeutically targetable types of CIN. Our results illuminate a fundamental structure underlying genomic complexity in human cancers and provide a resource to guide future CIN research.

The afternoon session at the ICI Berlin addresses the lived entanglements of the global that are not adequately theorized by notions of the global and the local, which bifurcate experiences that are necessarily co-constituting. It will take up the question of how artworks, artists, cultural producers, art historians, historians, and theorists who simultaneously exist on multiple and intertwining scales, co-constitute the global in relation with one another through meaning making/world-making processes of (critical) interconnecting, social networks, transnational/transcultural historiography, the circulation of objects, and imagined communities. We aim to address methodological challenges in the study of art history that reimagines its possible narratives by simultaneously holding account of art’s global resonances and its local, national, and regional frameworks. The papers on this panel are situated at the intersection between diaspora, postcolonial, and global (art history) studies. They seek to decolonize and deimperialize new top-down narratives of so called ‘global art’, that often identify global capitalism as the universal condition and will instead shed light on the multiperspectival polyphony that co-constitutes the global.Programme 15:30-17:30 Panel Birgit Hopfener, Carleton University Devika Singh, Cambridge University Francesca Tarocco, NYU Shanghai/Cà Foscari University, Venice David Teh, National University of Singapore Ming Tiampo, Carleton University 17:30-18:00 Discussion/Q&A moderated by John Tain 18:00-18:15 Coffee Break 18:15-18:45 Book Launch for Sarah Dornhof, Nanne Buurman, Birgit Hopfener, Barbara Lutz (eds.) Situating Global Art: Topologies, Temporalities, Trajectories (Bielefeld 2018). 18:45-19:00 Break 19:00 Keynote Lydia H. Liu (Columbia University) Co-constituting the Global: Ethical Challenges and Implications, symposium, ICI Berlin, 22 June 2018