Papers.

Deployment of a Blockchain-Based Self-Sovereign Identity

End-to-End Dense Video Captioning with Masked Transformer

Dense video captioning aims to generate text descriptions for all events in an untrimmed video. This involves both detecting and describing events. Therefore, all previous methods on dense video captioning tackle this problem by building two models, i.e. an event proposal and a captioning model, for these two sub-problems. The models are either trained separately or in alternation. This prevents direct influence of the language description to the event proposal, which is important for generating accurate descriptions. To address this problem, we propose an end-to-end transformer model for dense video captioning. The encoder encodes the video into appropriate representations. The proposal decoder decodes from the encoding with different anchors to form video event proposals. The captioning decoder employs a masking network to restrict its attention to the proposal event over the encoding feature. This masking network converts the event proposal to a differentiable mask, which ensures the consistency between the proposal and captioning during training. In addition, our model employs a self-attention mechanism, which enables the use of efficient non-recurrent structure during encoding and leads to performance improvements. We demonstrate the effectiveness of this end-to-end model on ActivityNet Captions and YouCookII datasets, where we achieved 10.12 and 6.58 METEOR score, respectively.

Structured Set Matching Networks for One-Shot Part Labeling

Experimental Evaluation of Hash Function Performance on Embedded Devices.

Mango: Distributed Visualization for Genomic Analysis

The decreasing cost of DNA sequencing over the past decade has led to an explosion of available sequencing datasets, leaving us with terabytes to petabytes of data to explore and analyze. It is critical for analysts in research and clinical settings to be able to develop new data-driven hypotheses from these datasets through bias identification, analysis of data quality, and testing different algorithms and parameter settings. However, current interactive tools for sequence analysis are designed to run on single machines that do not scale to the size of modern genomic datasets, and rely on precomputed static views, rather than allowing direct interaction with the primary dataset. Mango is a genomic sequence visualization and analysis platform that removes these constraints regarding scalability and staticity by leveraging the power of multi-node compute clusters in the cloud to allow interactive analysis over terabytes of sequencing data. Mango provides both a genome browser graphical user interface and programmable notebook form factor to allow users of varying analytical experience to explore large sequencing datasets on both private clusters and in the cloud. These tools provide a flexible environment for interactive exploration of genomic datasets, while surpassing the computational limits of single-node genomic visualization tools.

Over-optimization of academic publishing metrics: observing Goodhart’s Law in action

Abstract Background The academic publishing world is changing significantly, with ever-growing numbers of publications each year and shifting publishing patterns. However, the metrics used to measure academic success, such as the number of publications, citation number, and impact factor, have not changed for decades. Moreover, recent studies indicate that these metrics have become targets and follow Goodhart’s Law, according to which, “when a measure becomes a target, it ceases to be a good measure.” Results In this study, we analyzed >120 million papers to examine how the academic publishing world has evolved over the last century, with a deeper look into the specific field of biology. Our study shows that the validity of citation-based measures is being compromised and their usefulness is lessening. In particular, the number of publications has ceased to be a good metric as a result of longer author lists, shorter papers, and surging publication numbers. Citation-based metrics, such citation number and h-index, are likewise affected by the flood of papers, self-citations, and lengthy reference lists. Measures such as a journal’s impact factor have also ceased to be good metrics due to the soaring numbers of papers that are published in top journals, particularly from the same pool of authors. Moreover, by analyzing properties of >2,600 research fields, we observed that citation-based metrics are not beneficial for comparing researchers in different fields, or even in the same department. Conclusions Academic publishing has changed considerably; now we need to reconsider how we measure success.

Facebook language predicts depression in medical records

Depression, the most prevalent mental illness, is underdiagnosed and undertreated, highlighting the need to extend the scope of current screening methods. Here, we use language from Facebook posts of consenting individuals to predict depression recorded in electronic medical records. We accessed the history of Facebook statuses posted by 683 patients visiting a large urban academic emergency department, 114 of whom had a diagnosis of depression in their medical records. Using only the language preceding their first documentation of a diagnosis of depression, we could identify depressed patients with fair accuracy [area under the curve (AUC) = 0.69], approximately matching the accuracy of screening surveys benchmarked against medical records. Restricting Facebook data to only the 6 months immediately preceding the first documented diagnosis of depression yielded a higher prediction accuracy (AUC = 0.72) for those users who had sufficient Facebook data. Significant prediction of future depression status was possible as far as 3 months before its first documentation. We found that language predictors of depression include emotional (sadness), interpersonal (loneliness, hostility), and cognitive (preoccupation with the self, rumination) processes. Unobtrusive depression assessment through social media of consenting individuals may become feasible as a scalable complement to existing screening and monitoring procedures.

The Dat Project, an open and decentralized research data tool

A Programming Model and Foundation for Lineage-Based Distributed Computation

Flare Prediction Using Photospheric and Coronal Image Data

A Hybrid Approach to Privacy-Preserving Federated Learning

Federated learning facilitates the collaborative training of models without the sharing of raw data. However, recent attacks demonstrate that simply maintaining data locality during training processes does not provide sufficient privacy guarantees. Rather, we need a federated learning system capable of preventing inference over both the messages exchanged during training and the final trained model while ensuring the resulting model also has acceptable predictive accuracy. Existing federated learning approaches either use secure multiparty computation (SMC) which is vulnerable to inference or differential privacy which can lead to low accuracy given a large number of parties with relatively small amounts of data each. In this paper, we present an alternative approach that utilizes both differential privacy and SMC to balance these trade-offs. Combining differential privacy with secure multiparty computation enables us to reduce the growth of noise injection as the number of parties increases without sacrificing privacy while maintaining a pre-defined rate of trust. Our system is therefore a scalable approach that protects against inference threats and produces models with high accuracy. Additionally, our system can be used to train a variety of machine learning models, which we validate with experimental results on 3 different machine learning algorithms. Our experiments demonstrate that our approach out-performs state of the art solutions.

Advances in Data Science

Preface: New Computing in Digital Marketplaces Unleashed

Complex Collaborative Physical Process Management: A Position on the Trinity of BPM, IoT and DA

In the modern economy, we see complex business processes with a physical character executed collaboratively by a set of autonomous business organizations. Examples are international container logistics, integrated supply and manufacturing networks, and collaborative healthcare chains - all of which handle physical objects. Over time, these processes have become more complex, more business-critical, more time-critical, and at the same time heavily mass-customized. This implies that the processes need to be managed more explicitly in an increasingly real-time fashion, with ample attention to individual process cases. To support this kind of processes, no single existing technology class suffices. Therefore, we propose to integrate technologies from the areas of business process management (BPM - to manage the processes), internet of things (IoT - to sense and actuate the physical objects) and distributed analytics (DA - to take the right decisions at the right place in real-time) into a trinity. We illustrate our position with an example from the domain of container logistics.

Learning Ranking Functions by Genetic Programming Revisited

End-to-End Benchmark

Marketing Applications Using Big Data

Guide to Big Data Applications

Requirements for an Enterprise AI Benchmark

Domain Knowledge Driven Key Term Extraction for IT Services

Revisiting the Inverted Indices for Billion-Scale Approximate Nearest Neighbors

This work addresses the problem of billion-scale nearest neighbor search. The state-of-the-art retrieval systems for billion-scale databases are currently based on the inverted multi-index, the recently proposed generalization of the inverted index structure. The multi-index provides a very fine-grained partition of the feature space that allows extracting concise and accurate short-lists of candidates for the search queries. In this paper, we argue that the potential of the simple inverted index was not fully exploited in previous works and advocate its usage both for the highly-entangled deep descriptors and relatively disentangled SIFT descriptors. We introduce a new retrieval system that is based on the inverted index and outperforms the multi-index by a large margin for the same memory consumption and construction complexity. For example, our system achieves the state-of-the-art recall rates up to six times faster on the dataset of one billion deep descriptors compared to the efficient implementation of the inverted multi-index from the FAISS library.

ScarGAN: Chained Generative Adversarial Networks to Simulate Pathological Tissue on Cardiovascular MR Scans

We consider the problem of segmenting the left ventricular (LV) myocardium on late gadolinium enhancement (LGE) cardiovascular magnetic resonance (CMR) scans of which only some of the scans have scar tissue. We propose ScarGAN to simulate scar tissue on healthy myocardium using chained generative adversarial networks (GAN). Our novel approach factorizes the simulation process into 3 steps: (1) a mask generator to simulate the shape of the scar tissue; (2) a domain-specific heuristic to produce the initial simulated scar tissue from the mask; (3) a refining generator to add details to the simulated scar tissue. Unlike other approaches that generate samples from scratch, we simulate scar tissue on normal scans resulting in highly realistic samples. We show that experienced radiologists are unable to distinguish between real and simulated scar tissue. Training a U-Net with additional scans with scar tissue simulated by ScarGAN increases the percentage of scar pixels in LV myocardium prediction from 75.9% to 80.5%.

Combining Online Social Networks with Text Analysis

Multi-datacenter Consistency Properties

Structured Text Retrieval Models

Structured Document Retrieval

Validated automatic speech biomarkers in primary progressive aphasia

Passive Detection of Atrial Fibrillation Using a Commercially Available Smartwatch

PyText: A Seamless Path from NLP research to production

We introduce PyText - a deep learning based NLP modeling framework built on PyTorch. PyText addresses the often-conflicting requirements of enabling rapid experimentation and of serving models at scale. It achieves this by providing simple and extensible interfaces for model components, and by using PyTorch's capabilities of exporting models for inference via the optimized Caffe2 execution engine. We report our own experience of migrating experimentation and production workflows to PyText, which enabled us to iterate faster on novel modeling ideas and then seamlessly ship them at industrial scale.

A Statistical Approach to Assessing Neural Network Robustness

We present a new approach to assessing the robustness of neural networks based on estimating the proportion of inputs for which a property is violated. Specifically, we estimate the probability of the event that the property is violated under an input model. Our approach critically varies from the formal verification framework in that when the property can be violated, it provides an informative notion of how robust the network is, rather than just the conventional assertion that the network is not verifiable. Furthermore, it provides an ability to scale to larger networks than formal verification approaches. Though the framework still provides a formal guarantee of satisfiability whenever it successfully finds one or more violations, these advantages do come at the cost of only providing a statistical estimate of unsatisfiability whenever no violation is found. Key to the practical success of our approach is an adaptation of multi-level splitting, a Monte Carlo approach for estimating the probability of rare events, to our statistical robustness framework. We demonstrate that our approach is able to emulate formal verification procedures on benchmark problems, while scaling to larger networks and providing reliable additional information in the form of accurate estimates of the violation probability.

Detecting Backdoor Attacks on Deep Neural Networks by Activation Clustering

While machine learning (ML) models are being increasingly trusted to make decisions in different and varying areas, the safety of systems using such models has become an increasing concern. In particular, ML models are often trained on data from potentially untrustworthy sources, providing adversaries with the opportunity to manipulate them by inserting carefully crafted samples into the training set. Recent work has shown that this type of attack, called a poisoning attack, allows adversaries to insert backdoors or trojans into the model, enabling malicious behavior with simple external backdoor triggers at inference time and only a blackbox perspective of the model itself. Detecting this type of attack is challenging because the unexpected behavior occurs only when a backdoor trigger, which is known only to the adversary, is present. Model users, either direct users of training data or users of pre-trained model from a catalog, may not guarantee the safe operation of their ML-based system. In this paper, we propose a novel approach to backdoor detection and removal for neural networks. Through extensive experimental results, we demonstrate its effectiveness for neural networks classifying text and images. To the best of our knowledge, this is the first methodology capable of detecting poisonous data crafted to insert backdoors and repairing the model that does not require a verified and trusted dataset.

Stress-Testing Neural Models of Natural Language Inference with Multiply-Quantified Sentences

Standard evaluations of deep learning models for semantics using naturalistic corpora are limited in what they can tell us about the fidelity of the learned representations, because the corpora rarely come with good measures of semantic complexity. To overcome this limitation, we present a method for generating data sets of multiply-quantified natural language inference (NLI) examples in which semantic complexity can be precisely characterized, and we use this method to show that a variety of common architectures for NLI inevitably fail to encode crucial information; only a model with forced lexical alignments avoids this damaging information loss.

Assessing Generalization in Deep Reinforcement Learning

Enabling Factorized Piano Music Modeling and Generation with the MAESTRO Dataset

Generating musical audio directly with neural networks is notoriously difficult because it requires coherently modeling structure at many different timescales. Fortunately, most music is also highly structured and can be represented as discrete note events played on musical instruments. Herein, we show that by using notes as an intermediate representation, we can train a suite of models capable of transcribing, composing, and synthesizing audio waveforms with coherent musical structure on timescales spanning six orders of magnitude (~0.1 ms to ~100 s), a process we call Wave2Midi2Wave. This large advance in the state of the art is enabled by our release of the new MAESTRO (MIDI and Audio Edited for Synchronous TRacks and Organization) dataset, composed of over 172 hours of virtuosic piano performances captured with fine alignment (~3 ms) between note labels and audio waveforms. The networks and the dataset together present a promising approach toward creating new expressive and interpretable neural models of music.

Learning to Learn without Forgetting By Maximizing Transfer and Minimizing Interference

Lack of performance when it comes to continual learning over non-stationary distributions of data remains a major challenge in scaling neural network learning to more human realistic settings. In this work we propose a new conceptualization of the continual learning problem in terms of a temporally symmetric trade-off between transfer and interference that can be optimized by enforcing gradient alignment across examples. We then propose a new algorithm, Meta-Experience Replay (MER), that directly exploits this view by combining experience replay with optimization based meta-learning. This method learns parameters that make interference based on future gradients less likely and transfer based on future gradients more likely. We conduct experiments across continual lifelong supervised learning benchmarks and non-stationary reinforcement learning environments demonstrating that our approach consistently outperforms recently proposed baselines for continual learning. Our experiments show that the gap between the performance of MER and baseline algorithms grows both as the environment gets more non-stationary and as the fraction of the total experiences stored gets smaller.

Universal Language Model Fine-Tuning with Subword Tokenization for Polish

Universal Language Model for Fine-tuning [arXiv:1801.06146] (ULMFiT) is one of the first NLP methods for efficient inductive transfer learning. Unsupervised pretraining results in improvements on many NLP tasks for English. In this paper, we describe a new method that uses subword tokenization to adapt ULMFiT to languages with high inflection. Our approach results in a new state-of-the-art for the Polish language, taking first place in Task 3 of PolEval'18. After further training, our final model outperformed the second best model by 35%. We have open-sourced our pretrained models and code.

Supervising strong learners by amplifying weak experts

Many real world learning tasks involve complex or hard-to-specify objectives, and using an easier-to-specify proxy can lead to poor performance or misaligned behavior. One solution is to have humans provide a training signal by demonstrating or judging performance, but this approach fails if the task is too complicated for a human to directly evaluate. We propose Iterated Amplification, an alternative training strategy which progressively builds up a training signal for difficult problems by combining solutions to easier subproblems. Iterated Amplification is closely related to Expert Iteration (Anthony et al., 2017; Silver et al., 2017), except that it uses no external reward function. We present results in algorithmic environments, showing that Iterated Amplification can efficiently learn complex behaviors.

LIT: Block-wise Intermediate Representation Training for Model Compression

Unrestricted Adversarial Examples

We introduce a two-player contest for evaluating the safety and robustness of machine learning systems, with a large prize pool. Unlike most prior work in ML robustness, which studies norm-constrained adversaries, we shift our focus to unconstrained adversaries. Defenders submit machine learning models, and try to achieve high accuracy and coverage on non-adversarial data while making no confident mistakes on adversarial inputs. Attackers try to subvert defenses by finding arbitrary unambiguous inputs where the model assigns an incorrect label with high confidence. We propose a simple unambiguous dataset ("bird-or- bicycle") to use as part of this contest. We hope this contest will help to more comprehensively evaluate the worst-case adversarial risk of machine learning models.

Skill Rating for Generative Models

We explore a new way to evaluate generative models using insights from evaluation of competitive games between human players. We show experimentally that tournaments between generators and discriminators provide an effective way to evaluate generative models. We introduce two methods for summarizing tournament outcomes: tournament win rate and skill rating. Evaluations are useful in different contexts, including monitoring the progress of a single model as it learns during the training process, and comparing the capabilities of two different fully trained models. We show that a tournament consisting of a single model playing against past and future versions of itself produces a useful measure of training progress. A tournament containing multiple separate models (using different seeds, hyperparameters, and architectures) provides a useful relative comparison between different trained GANs. Tournament-based rating methods are conceptually distinct from numerous previous categories of approaches to evaluation of generative models, and have complementary advantages and disadvantages.

Text Classification using Capsules

What kind of content are you prone to tweet? Multi-topic Preference\n Model for Tweeters

Beyond Data and Model Parallelism for Deep Neural Networks

The computational requirements for training deep neural networks (DNNs) have grown to the point that it is now standard practice to parallelize training. Existing deep learning systems commonly use data or model parallelism, but unfortunately, these strategies often result in suboptimal parallelization performance. In this paper, we define a more comprehensive search space of parallelization strategies for DNNs called SOAP, which includes strategies to parallelize a DNN in the Sample, Operation, Attribute, and Parameter dimensions. We also propose FlexFlow, a deep learning framework that uses guided randomized search of the SOAP space to find a fast parallelization strategy for a specific parallel machine. To accelerate this search, FlexFlow introduces a novel execution simulator that can accurately predict a parallelization strategy's performance and is three orders of magnitude faster than prior approaches that have to execute each strategy. We evaluate FlexFlow with six real-world DNN benchmarks on two GPU clusters and show that FlexFlow can increase training throughput by up to 3.8x over state-of-the-art approaches, even when including its search time, and also improves scalability.

An Intriguing Failing of Convolutional Neural Networks and the CoordConv\n Solution

Few ideas have enjoyed as large an impact on deep learning as convolution.\nFor any problem involving pixels or spatial representations, common intuition\nholds that convolutional neural networks may be appropriate. In this paper we\nshow a striking counterexample to this intuition via the seemingly trivial\ncoordinate transform problem, which simply requires learning a mapping between\ncoordinates in (x,y) Cartesian space and one-hot pixel space. Although\nconvolutional networks would seem appropriate for this task, we show that they\nfail spectacularly. We demonstrate and carefully analyze the failure first on a\ntoy problem, at which point a simple fix becomes obvious. We call this solution\nCoordConv, which works by giving convolution access to its own input\ncoordinates through the use of extra coordinate channels. Without sacrificing\nthe computational and parametric efficiency of ordinary convolution, CoordConv\nallows networks to learn either complete translation invariance or varying\ndegrees of translation dependence, as required by the end task. CoordConv\nsolves the coordinate transform problem with perfect generalization and 150\ntimes faster with 10--100 times fewer parameters than convolution. This stark\ncontrast raises the question: to what extent has this inability of convolution\npersisted insidiously inside other tasks, subtly hampering performance from\nwithin? A complete answer to this question will require further investigation,\nbut we show preliminary evidence that swapping convolution for CoordConv can\nimprove models on a diverse set of tasks. Using CoordConv in a GAN produced\nless mode collapse as the transform between high-level spatial latents and\npixels becomes easier to learn. A Faster R-CNN detection model trained on MNIST\nshowed 24%...

Adversarial Robustness Toolbox v1.0.0

Adversarial Robustness Toolbox (ART) is a Python library supporting developers and researchers in defending Machine Learning models (Deep Neural Networks, Gradient Boosted Decision Trees, Support Vector Machines, Random Forests, Logistic Regression, Gaussian Processes, Decision Trees, Scikit-learn Pipelines, etc.) against adversarial threats and helps making AI systems more secure and trustworthy. Machine Learning models are vulnerable to adversarial examples, which are inputs (images, texts, tabular data, etc.) deliberately modified to produce a desired response by the Machine Learning model. ART provides the tools to build and deploy defences and test them with adversarial attacks. Defending Machine Learning models involves certifying and verifying model robustness and model hardening with approaches such as pre-processing inputs, augmenting training data with adversarial samples, and leveraging runtime detection methods to flag any inputs that might have been modified by an adversary. The attacks implemented in ART allow creating adversarial attacks against Machine Learning models which is required to test defenses with state-of-the-art threat models. Supported Machine Learning Libraries include TensorFlow (v1 and v2), Keras, PyTorch, MXNet, Scikit-learn, XGBoost, LightGBM, CatBoost, and GPy. The source code of ART is released with MIT license at https://github.com/IBM/adversarial-robustness-toolbox. The release includes code examples, notebooks with tutorials and documentation (this http URL).

Self-Reproducing Coins as Universal Turing Machine

The Natural Language Decathlon: Multitask Learning as Question Answering

Presented on August 28, 2018 at 12:15 p.m. in the Pettit Microelectronics Research Center, Room 102 A/B.

Smallify: Learning Network Size while Training

Towards Binary-Valued Gates for Robust LSTM Training

Long Short-Term Memory (LSTM) is one of the most widely used recurrent structures in sequence modeling. It aims to use gates to control information flow (e.g., whether to skip some information or not) in the recurrent computations, although its practical implementation based on soft gates only partially achieves this goal. In this paper, we propose a new way for LSTM training, which pushes the output values of the gates towards 0 or 1. By doing so, we can better control the information flow: the gates are mostly open or closed, instead of in a middle state, which makes the results more interpretable. Empirical studies show that (1) Although it seems that we restrict the model capacity, there is no performance drop: we achieve better or comparable performances due to its better generalization ability; (2) The outputs of gates are not sensitive to their inputs: we can easily compress the LSTM unit in multiple ways, e.g., low-rank approximation and low-precision approximation. The compressed models are even better than the baseline models without compression.

Analysis of DAWNBench, a Time-to-Accuracy Machine Learning Performance\n Benchmark