Neural networks are vulnerable to adversarial examples and researchers have proposed many heuristic attack and defense mechanisms. We address this problem through the principled lens of distributionally robust optimization, which guarantees performance under adversarial input perturbations. By considering a Lagrangian penalty formulation of perturbing the underlying data distribution in a Wasserstein ball, we provide a training procedure that augments model parameter updates with worst-case perturbations of training data. For smooth losses, our procedure provably achieves moderate levels of robustness with little computational or statistical cost relative to empirical risk minimization. Furthermore, our statistical guarantees allow us to efficiently certify robustness for the population loss. For imperceptible perturbations, our method matches or outperforms heuristic approaches.

A multidisciplinary group of experts gathered in Parma Italy for a workshop hosted by the University of Parma, May 16–18, 2014 to address concerns about the potential relationship between environmental metabolic disrupting chemicals, obesity and related metabolic disorders. The objectives of the workshop were to: 1. Review findings related to the role of environmental chemicals, referred to as “metabolic disruptors”, in obesity and metabolic syndrome with special attention to recent discoveries from animal model and epidemiology studies; 2. Identify conclusions that could be drawn with confidence from existing animal and human data; 3. Develop predictions based on current data; and 4. Identify critical knowledge gaps and areas of uncertainty. The consensus statements are intended to aid in expanding understanding of the role of metabolic disruptors in the obesity and metabolic disease epidemics, to move the field forward by assessing the current state of the science and to identify research needs on the role of environmental chemical exposures in these diseases. We propose broadening the definition of obesogens to that of metabolic disruptors, to encompass chemicals that play a role in altered susceptibility to obesity, diabetes and related metabolic disorders including metabolic syndrome.

SUMMARYThe stimulatory effects of testosterone on erythropoiesis are very well known, but the mechanisms underlying the erythropoietic action of testosterone are still poorly understood, although erythropoietin has long been considered a potential mediator. A total of 108 healthy men >65 years old with serum testosterone concentration <475 ng/dL were recruited by direct mailings to alumni of the University of Pennsylvania and Temple University, and randomized to receive a 60-cm 2 testosterone or placebo patch for 36 months.Ninety-six subjects completed the trial. We used information and stored serum specimens from this trial to test the hypothesis that increasing testosterone increases haemoglobin by stimulating erythropoietin production. We used information of 67 men, 43 in the testosterone group and 24 in the placebo group who had banked specimens available for assays of testosterone, haemoglobin and erythropoietin at baseline and after 36 months. The original randomized clinical study was primarily designed to verify the effects of testosterone on bone mineral density. The primary outcome of this report was to investigate whether or not transdermal testosterone increases haemoglobin by increasing erythropoietin levels. The mean age ± SD of the 67 subjects at baseline was 71.8 ± 4.9 years. Testosterone replacement therapy for 36 months, as compared with placebo, induced a significant increase in haemoglobin (0.86 ± 0.31 g/dL, p = 0.01), but no change in erythropoietin levels (À0.24 ± 2.16 mIU/mL, p = 0.91). Included time-varying measure of erythropoietin did not significantly account for the effect of testosterone on haemoglobin (Treatment-by-time: b = 0.93, SE = 0.33, p = 0.01). No serious adverse effect was observed. Transdermal testosterone treatment of older men for 36 months significantly increased haemoglobin, but not erythropoietin levels. The haematopoietic effect of testosterone does not appear to be mediated by stimulation of erythropoietin production.

In older men there is a multiple hormonal dysregulation with a relative prevalence of catabolic hormones such as thyroid hormones and cortisol and a decline in anabolic hormones such as dehydroepiandrosterone sulphate, testosterone and insulin like growth factor 1 levels. Many studies suggest that this catabolic milieu is an important predictor of frailty and mortality in older persons. There is a close relationship between frailty and cognitive impairment with studies suggesting that development of frailty is consequence of cognitive impairment and others pointing out that physical frailty is a determinant of cognitive decline. Decline in cognitive function, typically memory, is a major symptom of dementia. The "preclinical phase" of cognitive impairment occurs many years before the onset of dementia. The identification of relevant modifiable factors, including the hormonal dysregulation, may lead to therapeutic strategies for preventing the cognitive dysfunction. There are several mechanisms by which anabolic hormones play a role in neuroprotection and neuromodulation. These hormones facilitate recovery after brain injury and attenuate the neuronal loss. In contrast, elevated thyroid hormones may increase oxidative stress and apoptosis, leading to neuronal damage or death. In this mini review we will address the relationship between low levels of anabolic hormones, changes in thyroid hormones and cognitive function in older men. Then, giving the contradictory data of the literature and the multi-factorial origin of dementia, we will introduce the hypothesis of multiple hormonal derangement as a better determinant of cognitive decline in older men.

The use of antineoplastic drugs has a central role in treatment of patients affected by cancer but is often associated with numerous electrolyte derangements which, in many cases, could represent life-threatening conditions. In fact, while several anti-cancer agents can interfere with kidney function leading to acute kidney injury, proteinuria, and hypertension, in many cases alterations of electrolyte tubular handling and water balance occur. This review summarizes the mechanisms underlying the disturbances of sodium, potassium, magnesium, calcium, and phosphate metabolism during anti-cancer treatment. Platinum compounds are associated with sodium, potassium, and magnesium derangements while alkylating agents and Vinca alkaloids with hyponatremia due to syndrome of inappropriate antidiuretic hormone secretion (SIADH). Novel anti-neoplastic agents, such as targeted therapies (monoclonal antibodies, tyrosine kinase inhibitors, immunomodulators, mammalian target of rapamycin), can induce SIADH-related hyponatremia and, less frequently, urinary sodium loss. The blockade of epidermal growth factor receptor (EGFR) by anti-EGFR antibodies can result in clinically significant magnesium and potassium losses. Finally, the tumor lysis syndrome is associated with hyperphosphatemia, hypocalcemia and hyperkalemia, all of which represent serious complications of chemotherapy. Thus, clinicians should be aware of these side effects of antineoplastic drugs, in order to set out preventive measures and start appropriate treatments.