Malware detection is a popular application of Machine Learning for Information Security (ML-Sec), in which an ML classifier is trained to predict whether a given file is malware or benignware. Parameters of this classifier are typically optimized such that outputs from the model over a set of input samples most closely match the samples' true malicious/benign (1/0) target labels. However, there are often a number of other sources of contextual metadata for each malware sample, beyond an aggregate malicious/benign label, including multiple labeling sources and malware type information (e.g., ransomware, trojan, etc.), which we can feed to the classifier as auxiliary prediction targets. In this work, we fit deep neural networks to multiple additional targets derived from metadata in a threat intelligence feed for Portable Executable (PE) malware and benignware, including a multi-source malicious/benign loss, a count loss on multi-source detections, and a semantic malware attribute tag loss. We find that incorporating multiple auxiliary loss terms yields a marked improvement in performance on the main detection task. We also demonstrate that these gains likely stem from a more informed neural network representation and are not due to a regularization artifact of multi-target learning. Our auxiliary loss architecture yields a significant reduction in detection error rate (false negatives) of 42.6% at a false positive rate (FPR) of $10^{-3}$ when compared to a similar model with only one target, and a decrease of 53.8% at $10^{-5}$ FPR.

MIT has responded to Donald Trump's bizarre suggestion that its scientists are better at flying planes than trained pilots. In a strange tweet posted on Tuesday morning, the president had suggested that computer scientists from the Massachusetts Institute of Technology were required to to fly new planes, because they are so "complicated". "Pilots are no longer needed," he wrote. The post appeared to be a response to the Ethiopian Airlines disaster, in which 157 people were killed in a crash that had strange parallels to another accident involving the same model of plane just months before. Authorities across Europe have temporarily grounded all of the same planes in an attempt to stop further disasters.

Many facial recognition systems are being trained using millions of online photos uploaded by everyday people and, more often than not, the photos are being taken without users' consent, an NBC News investigation has found. In one worrying case, IBM scraped almost a million photos from unsuspecting users on Flickr to build its facial recognition database. The practice not only raises privacy concerns, but also fuels fears that the systems could one day be used to disproportionately target minorities. Many facial recognition systems are being trained using millions of online photos uploaded by everyday people and, more often than not, the photos are being taken without users' consent IBM's database, called'Diversity in Faces,' was released in January as part of the company's efforts to'advance the study of fairness and accuracy in facial recognition technology.' The database was released following a study from MIT Media Lab researcher Joy Buolamwini, which found that popular facial recognition services from Microsoft, IBM and Face vary in accuracy based on gender and race.

The recently prevalent tensor train (TT) and tensor ring (TR) decompositions can be graphically interpreted as (locally) linear interconnected latent factors and possess exponential decay of correlation. The projected entangled pair state (PEPS, also called two-dimensional TT) extends the spatial dimension of TT and its polycyclic structure can be considered as a square grid. Compared with TT, its algebraic decay of correlation means the enhancement of interaction between tensor modes. In this paper we adopt the PEPS and develop a tensor grid (TG) decomposition with its efficient realization termed splitting singular value decomposition (SSVD). By utilizing the alternating least squares (ALS) a method called TG-ALS is used to interpolate the missing entries of a tensor from its partial observations. Different kinds of data are used in the experiments, including synthetic data, color images and real-world videos. Experimental results demonstrate that the TG has much power of representation than TT and TR.

How do we know if communication is emerging in a multi-agent system? The vast majority of recent papers on emergent communication show that adding a communication channel leads to an increase in reward or task success. This is a useful indicator, but provides only a coarse measure of the agent's learned communication abilities. As we move towards more complex environments, it becomes imperative to have a set of finer tools that allow qualitative and quantitative insights into the emergence of communication. This may be especially useful to allow humans to monitor agents' behaviour, whether for fault detection, assessing performance, or even building trust. In this paper, we examine a few intuitive existing metrics for measuring communication, and show that they can be misleading. Specifically, by training deep reinforcement learning agents to play simple matrix games augmented with a communication channel, we find a scenario where agents appear to communicate (their messages provide information about their subsequent action), and yet the messages do not impact the environment or other agent in any way. We explain this phenomenon using ablation studies and by visualizing the representations of the learned policies. We also survey some commonly used metrics for measuring emergent communication, and provide recommendations as to when these metrics should be used.

Today's customers demand more options, more creative solutions, greater flexibility and faster responses from banks and other financial institutions. Survival and success for financial institutions in this new world requires that they operate with intelligence, agility and speed to keep up with evolving customer preferences and technologies. Consequently, more and more customer interactions and financial transactions are going digital as online and mobile payments, customer on-boarding and account opening are on the rise. Yet, while digital interfaces present an opening for innovative business services, they also yield new challenges, such as pressure on back office operations or increased regulatory scrutiny. Largely automated interactions generate more data to analyse, demand higher volumes of sample testing, and expand the compliance burden. To create a flawless customer experience, the back office has to keep up as well.

We introduce a dynamic generative model, Bayesian allocation model (BAM), which establishes explicit connections between nonnegative tensor factorization (NTF), graphical models of discrete probability distributions and their Bayesian extensions, and the topic models such as the latent Dirichlet allocation. BAM is based on a Poisson process, whose events are marked by using a Bayesian network, where the conditional probability tables of this network are then integrated out analytically. We show that the resulting marginal process turns out to be a Polya urn, an integer valued self-reinforcing process. This urn processes, which we name a Polya-Bayes process, obey certain conditional independence properties that provide further insight about the nature of NTF. These insights also let us develop space efficient simulation algorithms that respect the potential sparsity of data: we propose a class of sequential importance sampling algorithms for computing NTF and approximating their marginal likelihood, which would be useful for model selection. The resulting methods can also be viewed as a model scoring method for topic models and discrete Bayesian networks with hidden variables. The new algorithms have favourable properties in the sparse data regime when contrasted with variational algorithms that become more accurate when the total sum of the elements of the observed tensor goes to infinity. We illustrate the performance on several examples and numerically study the behaviour of the algorithms for various data regimes.

Neuroscience and artificial intelligence (AI) are two very different scientific disciplines. Neuroscience traces back to ancient civilizations, and AI is a decidedly modern phenomenon. At a cursory glance, it would seem that a branch of science of living systems would have little in common with one that springs from inanimate machines wholly created by humans. Yet discoveries in one field may result in breakthroughs in the other-- the two fields share a significant problem, and future opportunities. The origins of modern neuroscience is rooted in ancient human civilizations. One of the first descriptions of the brain's structure and neurosurgery can be traced back to 3000 - 2500 B.C. largely due to the efforts of the American Egyptologist Edwin Smith.

Consider a data set collected by (individuals-features) pairs in different times. It can be represented as a tensor of three dimensions (Individuals, features and times). The tensor biclustering problem computes a subset of individuals and a subset of features whose signal trajectories over time lie in a low-dimensional subspace, modeling similarity among the signal trajectories while allowing different scalings across different individuals or different features. This approach are based on spectral decomposition in order to build the desired biclusters. We evaluate the quality of the results from each algorithms with both synthetic and real data set.

Genomics is set to become the biggest source of data on the planet, overtaking the current leading heavyweights – astronomy, YouTube and Twitter. Genome sequencing currently produces a staggering 25 petabytes of digital information per year. A petabyte is 1015 bytes, or about 1,000 times the average storage on a personal computer. And there is no sign of a slowdown. The amount of DNA sequencing data produced around the world is doubling approximately every seven months.