In proteomics, De novo peptide sequencing from tandem Mass Spectrometry (MS) data is the key technology for finding new peptide or protein sequences. It has successful applications in assemble monocolonal antibody sequences (mAbs)[1] and great potentials in identifying neoantigens for personalized cancer vaccines[2]. Given the importance of the de novo peptide sequencing technology, massive research have been done in this area and different tools have been proposed[3][4][5][6][7]. In 2017, Tran et al. first introduced deep learning to de novo peptide sequencing and proposed DeepNovo, a neural network based de novo peptide sequencing model[8] for Data Dependent Acquisition (DDA) MS data. Inspired by the success of image captioning model[9], DeepNovo integrated two fundamental types of neural networks, CNNs and LSTM, in order to extract features from both the spectrum and the "language model of peptides". In DeepNovo, each spectrum is represented as a long intensity vector and CNNs are applied on segments of this vector to extract features and make predictions of the next amino acid. CNNs have been proved as effective tools for pattern recognition in different applications like image classification, object detection and sentiment analysis[10][11][12]. By applying CNNs to the intensity vector, DeepNovo could learn from noisy spectrum.

The high rate of false alarms in intensive care units (ICUs) is one of the top challenges of using medical technology in hospitals. These false alarms are often caused by patients' movements, detachment of monitoring sensors, or different sources of noise and interference that impact the collected signals from different monitoring devices. In this paper, we propose a novel set of high-level features based on unsupervised feature learning technique in order to effectively capture the characteristics of different arrhythmia in electrocardiogram (ECG) signal and differentiate them from irregularity in signals due to different sources of signal disturbances. This unsupervised feature learning technique, first extracts a set of low-level features from all existing heart cycles of a patient, and then clusters these segments for each individual patient to provide a set of prominent high-level features. The objective of the clustering phase is to enable the classification method to differentiate between the high-level features extracted from normal and abnormal cycles (i.e., either due to arrhythmia or different sources of distortions in signal) in order to put more attention to the features extracted from abnormal portion of the signal that contribute to the alarm. The performance of this method is evaluated using the 2015 PhysioNet/Computing in Cardiology Challenge dataset for reducing false arrhythmia alarms in the ICUs. As confirmed by the experimental results, the proposed method offers a considerable performance in terms of accuracy, sensitivity and specificity of alarm detection only using a few high-level features that are extracted from one single lead ECG signal.

This technical report describes a practical field test on word-image classification in a very large collection of more than 300 diverse handwritten historical manuscripts, with 1.6 million unique labeled images and more than 11 million images used in testing. Results indicate that several deep-learning tests completely failed (mean accuracy 83%). In the tests with more than 1000 output units (lexical words) in one-hot encoding for classification, performance steeply drops to almost zero percent accuracy, even with a modest size of the pre-final (i.e., penultimate) layer (150 units). A traditional feature method (BOVW) displays a consistent performance over numbers of classes and numbers of training examples (mean accuracy 87%). Additional tests using nearest mean on the output of the pre-final layer of an Inception V3 network, for each book, only yielded mediocre results (mean accuracy 49\%), but was not sensitive to high numbers of classes. Notably, this experiment was only possible on the basis of labels that were harvested on the basis of a traditional method which already works starting from a single labeled image per class. It is expected that the performance of the failed deep learning tests can be repaired, but only on the basis of human handcrafting (sic) of network architecture and hyperparameters. When the failed problematic books are not considered, end-to-end CNN training yields about 95% accuracy. This average is dominated by a large subset of Chinese characters, performances for other script styles being lower.

For many automated driving functions, a highly accurate perception of the vehicle environment is a crucial prerequisite. Modern high-resolution radar sensors generate multiple radar targets per object, which makes these sensors particularly suitable for the 2D object detection task. This work presents an approach to detect object hypotheses solely depending on sparse radar data using PointNets. In literature, only methods are presented so far which perform either object classification or bounding box estimation for objects. In contrast, this method facilitates a classification together with a bounding box estimation of objects using a single radar sensor. To this end, PointNets are adjusted for radar data performing 2D object classification with segmentation, and 2D bounding box regression in order to estimate an amodal bounding box. The algorithm is evaluated using an automatically created dataset which consist of various realistic driving maneuvers. The results show the great potential of object detection in high-resolution radar data using PointNets.

This research note combines two methods that have recently improved the state of the art in language modeling: Transformers and dynamic evaluation. Transformers use stacked layers of self-attention that allow them to capture long range dependencies in sequential data. Dynamic evaluation fits models to the recent sequence history, allowing them to assign higher probabilities to reoccurring sequential patterns. By applying dynamic evaluation to Transformer-XL models, we improve the state of the art on enwik8 from 0.99 to 0.94 bits/char, text8 from 1.08 to 1.04 bits/char, and WikiText-103 from 18.3 to 16.4 perplexity points. Language modeling is a commonly used machine learning benchmark with applications to speech recognition, machine translation, text generation, and unsupervised learning in natural language processing tasks.

Frameworks for writing, compiling, and optimizing deep learning (DL) models have recently enabled progress in areas like computer vision and natural language processing. Extending these frameworks to accommodate the rapidly diversifying landscape of DL models and hardware platforms presents challenging tradeoffs between expressiveness, composability, and portability. We present Relay, a new intermediate representation (IR) and compiler framework for DL models. The functional, statically-typed Relay IR unifies and generalizes existing DL IRs and can express state-of-the-art models. Relay's expressive IR required careful design of the type system, automatic differentiation, and optimizations. Relay's extensible compiler can eliminate abstraction overhead and target new hardware platforms. The design insights from Relay can be applied to existing frameworks to develop IRs that support extension without compromising on expressivity, composibility, and portability. Our evaluation demonstrates that the Relay prototype can already provide competitive performance for a broad class of models running on CPUs, GPUs, and FPGAs.

This paper presents Rogue-Gym, that enables agents to learn and play a subset of the original Rogue game with the OpenAI Gym interface. In roguelike games, a player explores a dungeon where each floor is two dimensional grid maze with enemies, golds, and downstairs. Because the map of a dungeon is different each time an agent starts a new game, learning in Rogue-Gym inevitably involves generalization of experiences, in a highly abstract manner. We argue that this generalization in reinforcement learning is a big challenge for AI agents. Recently, deep reinforcement learning (DRL) has succeeded in many games. However, it has been pointed out that agents trained by DRL methods often overfit to the training environment. To investigate this problem, some research environments with procedural content generation have been proposed. Following these studies, we show that our Rogue-Gym imposes a new generalization problem of their policies. In our experiments, we evaluate a standard reinforcement learning method, PPO, with and without enhancements for generalization. The results show that some enhancements work effective, but that there is still a large room for improvement. Therefore, Rogue-Gym a is a new challenging domain for further studies.

Recently, researchers have started decomposing deep neural network models according to their semantics or functions. Recent work has shown the effectiveness of decomposed functional blocks for defending adversarial attacks, which add small input perturbation to the input image to fool the DNN models. This work proposes a profiling-based method to decompose the DNN models to different functional blocks, which lead to the effective path as a new approach to exploring DNNs' internal organization. Specifically, the per-image effective path can be aggregated to the class-level effective path, through which we observe that adversarial images activate effective path different from normal images. We propose an effective path similarity-based method to detect adversarial images with an interpretable model, which achieve better accuracy and broader applicability than the state-of-the-art technique.

Event cameras are bio-inspired sensors that work radically different from traditional cameras. Instead of capturing images at a fixed rate, they measure per-pixel brightness changes asynchronously. This results in a stream of events, which encode the time, location and sign of the brightness changes. Event cameras posses outstanding properties compared to traditional cameras: very high dynamic range (140 dB vs. 60 dB), high temporal resolution (in the order of microseconds), low power consumption, and do not suffer from motion blur. Hence, event cameras have a large potential for robotics and computer vision in challenging scenarios for traditional cameras, such as high speed and high dynamic range. However, novel methods are required to process the unconventional output of these sensors in order to unlock their potential. This paper provides a comprehensive overview of the emerging field of event-based vision, with a focus on the applications and the algorithms developed to unlock the outstanding properties of event cameras. We present event cameras from their working principle, the actual sensors that are available and the tasks that they have been used for, from low-level vision (feature detection and tracking, optic flow, etc.) to high-level vision (reconstruction, segmentation, recognition). We also discuss the techniques developed to process events, including learning-based techniques, as well as specialized processors for these novel sensors, such as spiking neural networks. Additionally, we highlight the challenges that remain to be tackled and the opportunities that lie ahead in the search for a more efficient, bio-inspired way for machines to perceive and interact with the world.

Conventional automatic speech recognition (ASR) systems trained from frame-level alignments can easily leverage posterior fusion to improve ASR accuracy and build a better single model with knowledge distillation. End-to-end ASR systems trained using the Connectionist Temporal Classification (CTC) loss do not require frame-level alignment and hence simplify model training. However, sparse and arbitrary posterior spike timings from CTC models pose a new set of challenges in posterior fusion from multiple models and knowledge distillation between CTC models. We propose a method to train a CTC model so that its spike timings are guided to align with those of a pre-trained guiding CTC model. As a result, all models that share the same guiding model have aligned spike timings. We show the advantage of our method in various scenarios including posterior fusion of CTC models and knowledge distillation between CTC models with different architectures. With the 300-hour Switchboard training data, the single word CTC model distilled from multiple models improved the word error rates to 13.7%/23.1% from 14.9%/24.1% on the Hub5 2000 Switchboard/CallHome test sets without using any data augmentation, language model, or complex decoder.