2,610 research outputs found
Like trainer, like bot? Inheritance of bias in algorithmic content moderation
Get PDFThe internet has become a central medium through which `networked publics'
express their opinions and engage in debate. Offensive comments and personal
attacks can inhibit participation in these spaces. Automated content moderation
aims to overcome this problem using machine learning classifiers trained on
large corpora of texts manually annotated for offence. While such systems could
help encourage more civil debate, they must navigate inherently normatively
contestable boundaries, and are subject to the idiosyncratic norms of the human
raters who provide the training data. An important objective for platforms
implementing such measures might be to ensure that they are not unduly biased
towards or against particular norms of offence. This paper provides some
exploratory methods by which the normative biases of algorithmic content
moderation systems can be measured, by way of a case study using an existing
dataset of comments labelled for offence. We train classifiers on comments
labelled by different demographic subsets (men and women) to understand how
differences in conceptions of offence between these groups might affect the
performance of the resulting models on various test sets. We conclude by
discussing some of the ethical choices facing the implementers of algorithmic
moderation systems, given various desired levels of diversity of viewpoints
amongst discussion participants.Comment: 12 pages, 3 figures, 9th International Conference on Social
Informatics (SocInfo 2017), Oxford, UK, 13--15 September 2017 (forthcoming in
Springer Lecture Notes in Computer Science
Two classes of nonlocal Evolution Equations related by a shared Traveling Wave Problem
Full text linkWe consider reaction-diffusion equations and Korteweg-de Vries-Burgers (KdVB)
equations, i.e. scalar conservation laws with diffusive-dispersive
regularization. We review the existence of traveling wave solutions for these
two classes of evolution equations. For classical equations the traveling wave
problem (TWP) for a local KdVB equation can be identified with the TWP for a
reaction-diffusion equation. In this article we study this relationship for
these two classes of evolution equations with nonlocal diffusion/dispersion.
This connection is especially useful, if the TW equation is not studied
directly, but the existence of a TWS is proven using one of the evolution
equations instead. Finally, we present three models from fluid dynamics and
discuss the TWP via its link to associated reaction-diffusion equations
Characterization of PRLR and PPARGC1A genes in buffalo (Bubalus bubalis)
Get PDFMore than 40 million households in India depend at least partially on livestock production. Buffaloes are one of the major milk producers in India. The prolactin receptor (PRLR) gene and peroxisome proliferators activated receptor-Îł coactivator 1-alpha (PPARGC1A) gene are reportedly associated with milk protein and milk fat yields in Bos taurus. In this study, we sequenced the PRLR and PPARGC1A genes in the water buffalo Bubalus bubalis. The PRLR and PPARGC1A genes coded for 581 and 819 amino acids, respectively. The B. bubalis PRLR gene differed from the corresponding Bos taurus at 21 positions and four differences with an additional arginine at position 620 in the PPARGC1A gene were found in the amino acid sequence. All of the changes were confirmed by cDNA sequencing. Twelve buffalo-specific single nucleotide polymorphisms (SNPs) were identified in both genes, with five of them being non-synonymous
Wake response to an ocean-feedback mechanism: Madeira Island case study
Full text linkThis discussion focused on the numerical study of a wake episode. The Weather
Research and Forecasting model was used in a downscale mode. The current
literature focuses the discussion on the adiabatic dynamics of atmospheric
wakes. Changes in mountain height and consequently on its relation to the
atmospheric inversion layer should explain the shift in wake regimes: from a
'strong-wake' to a 'weak-wake' scenario. Nevertheless, changes in SST
variability can also induce similar regime shifts. Increase in evaporation,
contributes to increase convection and thus to an uplift of the stratified
atmospheric layer, above the critical height, with subsequent internal gravity
wave activity.Comment: Under review proces
Learning Interpretable Rules for Multi-label Classification
Full text linkMulti-label classification (MLC) is a supervised learning problem in which,
contrary to standard multiclass classification, an instance can be associated
with several class labels simultaneously. In this chapter, we advocate a
rule-based approach to multi-label classification. Rule learning algorithms are
often employed when one is not only interested in accurate predictions, but
also requires an interpretable theory that can be understood, analyzed, and
qualitatively evaluated by domain experts. Ideally, by revealing patterns and
regularities contained in the data, a rule-based theory yields new insights in
the application domain. Recently, several authors have started to investigate
how rule-based models can be used for modeling multi-label data. Discussing
this task in detail, we highlight some of the problems that make rule learning
considerably more challenging for MLC than for conventional classification.
While mainly focusing on our own previous work, we also provide a short
overview of related work in this area.Comment: Preprint version. To appear in: Explainable and Interpretable Models
in Computer Vision and Machine Learning. The Springer Series on Challenges in
Machine Learning. Springer (2018). See
http://www.ke.tu-darmstadt.de/bibtex/publications/show/3077 for further
informatio
RIPK1-mediated immunogenic cell death promotes anti-tumour immunity against soft-tissue sarcoma.
Get PDFDrugs that mobilise the immune system against cancer are dramatically improving care for many people. Dying cancer cells play an active role in inducing anti-tumour immunity but not every form of death can elicit an immune response. Moreover, resistance to apoptosis is a major problem in cancer treatment and disease control. While the term "immunogenic cell death" is not fully defined, activation of receptor-interacting serine/threonine-protein kinase 1 (RIPK1) can induce a type of death that mobilises the immune system against cancer. However, no clinical treatment protocols have yet been established that would harness the immunogenic potential of RIPK1. Here, we report the first pre-clinical application of an in vivo treatment protocol for soft-tissue sarcoma that directly engages RIPK1-mediated immunogenic cell death. We find that RIPK1-mediated cell death significantly improves local disease control, increases activation of CD8+ T cells as well as NK cells, and enhances the survival benefit of immune checkpoint blockade. Our findings warrant a clinical trial to assess the survival benefit of RIPK1-induced cell death in patients with advanced disease at limb extremities
OX40 and 4-1BB delineate distinct immune profiles in sarcoma.
Get PDFSystemic relapse after radiotherapy and surgery is the major cause of disease-related mortality in sarcoma patients. Combining radiotherapy and immunotherapy is under investigation as a means to improve response rates. However, the immune contexture of sarcoma is understudied. Here, we use a retrospective cohort of sarcoma patients, treated with neoadjuvant radiotherapy, and TCGA data. We explore therapeutic targets of relevance to sarcoma, using genomics and multispectral immunohistochemistry to provide insights into the tumor immune microenvironment across sarcoma subtypes. Differential gene expression between radioresponsive myxoid liposarcoma (MLPS) and more radioresistant undifferentiated pleomorphic sarcoma (UPS) indicated UPS contained higher transcript levels of a number of immunotherapy targets (CD73/NT5E, CD39/ENTPD1, CD25/IL2RA, and 4-1BB/TNFRSF9). We focused on 4-1BB/TNFRSF9 and other costimulatory molecules. In TCGA data, 4-1BB correlated to an inflamed and exhausted phenotype. OX40/TNFRSF4 and 4-1BB/TNFRSF9 were highly expressed in sarcoma subtypes versus other cancers. Despite OX40 and 4-1BB being described as Treg markers, we identified that they delineate distinct tumor immune profiles. This was true for sarcoma and other cancers. While only a limited number of samples could be analyzed, spatial analysis of OX40 expression identified two diverse phenotypes of OX40+Â Tregs, one associated with and one independent of tertiary lymphoid structures (TLSs). Patient stratification is of intense interest for immunotherapies. We provide data supporting the viewpoint that a cohort of sarcoma patients, appropriately selected, are promising candidates for immunotherapies. Spatial profiling of OX40+Â Tregs, in relation to TLSs, could be an additional metric to improve future patient stratification
On stability regions of the modified midpoint method for a linear delay differential equation
Get PDF- …
