Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

/pdf/deep-residual-learning-for-image-recognition-b75jg61qrq.pdf

Deep Residual Learning for Image Recognition

We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. The method is straightforward to implement, is computationally efficient, has little memory requirements, is invariant to diagonal rescaling of the gradients, and is well suited for problems that are large in terms of data and/or parameters. The method is also appropriate for non-stationary objectives and problems with very noisy and/or sparse gradients. The hyper-parameters have intuitive interpretations and typically require little tuning. Some connections to related algorithms, on which Adam was inspired, are discussed. We also analyze the theoretical convergence properties of the algorithm and provide a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework. Empirical results demonstrate that Adam works well in practice and compares favorably to other stochastic optimization methods. Finally, we discuss AdaMax, a variant of Adam based on the infinity norm.

Adam: A Method for Stochastic Optimization

We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 different classes. On the test data, we achieved top-1 and top-5 error rates of 37.5% and 17.0% which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax. To make training faster, we used non-saturating neurons and a very efficient GPU implementation of the convolution operation. To reduce overriding in the fully-connected layers we employed a recently-developed regularization method called "dropout" that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3%, compared to 26.2% achieved by the second-best entry.

/pdf/imagenet-classification-with-deep-convolutional-neural-1190x9a6b9.pdf

ImageNet Classification with Deep Convolutional Neural Networks

Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. 
The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.

/pdf/going-deeper-with-convolutions-1yobw2o2ds.pdf

Going deeper with convolutions

High-level synthesis (HLS) enables designing at a higher level of abstraction to effectively cope with design complexity of emerging applications on modern programmable system-on-chip (SoC). While HLS continues to evolve with a growing set of algorithms, methodologies, and tools to efficiently map software designs onto optimized hardware architectures, there continues to lack realistic benchmark applications with sufficient complexity and enforceable constraints. In this paper we present a case study of accelerating face detection based on the Viola Jones algorithm on a programmable SoC using a C-based HLS flow. We also share our insights in porting a software-based design into a synthesizable implementation with HLS-specific data structures and optimizations. Our design is able to achieve a frame rate of 30 frames per second which is suitable for realtime applications. Our performance and quality of results are comparable to those of many traditional RTL implementations.

Accelerating Face Detection on Programmable SoC Using C-Based Synthesis

State-of-the-art learning-based monocular 3D reconstruction methods learn priors over object categories on the training set, and as a result struggle to achieve reasonable generalization to object categories unseen during training In this paper we study the inductive biases encoded in the model architecture that impact the generalization of learning-based 3D reconstruction methods We find that 3 inductive biases impact performance: the spatial extent of the encoder, the use of the underlying geometry of the scene to describe point features, and the mechanism to aggregate information from multiple views Additionally, we propose mechanisms to enforce those inductive biases: a point representation that is aware of camera position, and a variance cost to aggregate information across views Our model achieves state-of-the-art results on the standard ShapeNet 3D reconstruction benchmark in various settings

On the generalization of learning-based 3D reconstruction

A system includes a neural network organized into layers corresponding to stages of inferences. The neural network includes a common portion, a first portion, and a second portion. The first portion includes a first set of layers dedicated to performing a first inference task on an input data. The second portion includes a second set of layers dedicated to performing a second inference task on the same input data. The common portion includes a third set of layers, which may include an input layer to the neural network, that are used in the performance of both the first and second inference tasks. The system may receive an input data and perform both inference tasks on the input data in a single pass. During training, a training sample with annotations for both inference tasks may be used to train the neural network in a single pass.

Partially shared neural networks for multiple tasks

Modeling heterogeneity in field crops is a key issue for a better characterization of field production. This paper presents some experimental data on sugar beet illustrating this heterogeneity. Several sources of individual variability within plant populations are identified: namely, initial condition (seed biomass, emergence delay), genetic variability (including phyllochron) and environment (including spacing and competition). A mathematical framework is introduced to integrate the different sources of variability in plant growth models. It is based on the classical method of Taylor Series Expansion, which allows the propagation of uncertainty in the dynamic system of growth and the computation of the approximate means and standard deviations of the model outputs. The method is applied to the GreenLab model of plant growth and more specifically to sugar beet. It opens perspectives in order to assess the different sources of variability in plant populations and estimate their parameters from experimental data. However important issues like optimization of data collection and system identifiability have to be resolved first, since the uncertainty effects may be mixed in an inextricable way or may necessitate a too huge amount of experimental data for their estimation.

/pdf/modeling-inter-individual-variability-in-sugar-beet-owtse5sxn1.pdf

Modeling Inter-individual Variability in Sugar Beet Populations

A system employs an inspection neural network (INN) to inspect data generated during an inference process of a primary neural network (PNN) to generate an indication of reliability for an output generated by the PNN. The system includes a sensor configured to capture sensor data. Sensor data captured by the sensor is provided to a data analyzer to generate an output using the PNN. An analyzer inspector is configured to capture inspection data associated with the generation of the output by the data analyzer, and use the INN to generate an indication of reliability for the PNN's output based on the inspection data. The INN is trained using a set of training data that is distinct from the training data used to train the PNN.

Nitish Srivastava

Papers

Accelerating Face Detection on Programmable SoC Using C-Based Synthesis

On the generalization of learning-based 3D reconstruction

Partially shared neural networks for multiple tasks

Modeling Inter-individual Variability in Sugar Beet Populations

Inspection neural network for assessing neural network reliability