The options framework is a popular approach for building temporally extended actions in reinforcement learning. In particular, the option-critic architecture provides general purpose policy gradient theorems for learning actions from scratch that are extended in time. However, past work makes the key assumption that each of the components of option-critic has independent parameters. In this work we note that while this key assumption of the policy gradient theorems of option-critic holds in the tabular case, it is always violated in practice for the deep function approximation setting. We thus reconsider this assumption and consider more general extensions of option-critic and hierarchical option-critic training that optimize for the full architecture with each update. It turns out that not assuming parameter independence challenges a belief in prior work that training the policy over options can be disentangled from the dynamics of the underlying options. In fact, learning can be sped up by focusing the policy over options on states where options are actually likely to terminate. We put our new algorithms to the test in application to sample efficient learning of Atari games, and demonstrate significantly improved stability and faster convergence when learning long options.

We propose a stochastic optimization method for minimizing loss functions, which can be expressed as an expected value, that adap-tively controls the batch size used in the computation of gradient approximations and the step size used to move along such directions, eliminating the need for the user to tune the learning rate. The proposed method exploits local curvature information and ensures that search directions are descent directions with high probability using an acute-angle test. The method is proved to have, under reasonable assumptions, a global linear rate of convergence on self-concordant functions with high probability. Numerical experiments show that this method is able to choose the best learning rates and compares favorably to fine-tuned SGD for training logistic regression and Deep Neural Networks (DNNs). We also propose an adaptive version of ADAM that eliminates the need to tune the base learning rate and compares favorably to fine-tuned ADAM for training DNNs.

Accurate quantification of pulmonary nodules can greatly assist the early diagnosis of lung cancer, which can enhance patient survival possibilities. A number of nodule segmentation techniques have been proposed, however, all of the existing techniques rely on radiologist 3-D volume of interest (VOI) input or use the constant region of interest (ROI) and only investigate the presence of nodule voxels within the given VOI. Such approaches restrain the solutions to investigate the nodule presence outside the given VOI and also include the redundant structures into VOI, which may lead to inaccurate nodule segmentation. In this work, a novel semi-automated approach for 3-D segmentation of nodule in volumetric computerized tomography (CT) lung scans has been proposed. The proposed technique can be segregated into two stages, at the first stage, it takes a 2-D ROI containing the nodule as input and it performs patch-wise investigation along the axial axis with a novel adaptive ROI strategy. The adaptive ROI algorithm enables the solution to dynamically select the ROI for the surrounding slices to investigate the presence of nodule using deep residual U-Net architecture. The first stage provides the initial estimation of nodule which is further utilized to extract the VOI. At the second stage, the extracted VOI is further investigated along the coronal and sagittal axis with two different networks and finally, all the estimated masks are fed into the consensus module to produce the final volumetric segmentation of nodule. The proposed approach has been rigorously evaluated on the LIDC dataset, which is the largest publicly available dataset. The result suggests that the approach is significantly robust and accurate as compared to the previous state of the art techniques.

In this work, we conducted a study on building an automated testing system for deep learning systems based on differential behavior criteria. The automated testing goals were achieved by jointly optimizing two objective functions: maximizing differential behaviors from models under testing and maximizing neuron coverage. By observing differential behaviors from three pre-trained models during each testing iteration, the input image that triggered erroneous feedback was registered as a corner-case. The generated corner-cases can be used to examine the robustness of DNNs and consequently improve model accuracy. A project called DeepXplore was also used as a baseline model. After we fully implemented and optimized the baseline system, we explored its application as an augmenting training dataset with newly generated corner cases. With the GTRSB dataset, by retraining the model based on automated generated corner cases, the accuracy of three generic models increased by 259.2%, 53.6%, and 58.3%, respectively. Further, to extend the capability of automated testing, we explored other approaches based on differential behavior criteria to generate photo-realistic images for deep learning systems. One approach was to apply various transformations to the seed images for the deep learning framework. The other approach was to utilize the Generative Adversarial Networks (GAN) technique, which was implemented on MNIST and Driving datasets. The style transferring capability has been observed very effective in adding additional visual effects, replacing image elements, and style-shifting (virtual image to real images). The GAN-based testing sample generation system was shown to be the next frontier for automated testing for deep learning systems.

Neural architecture search (NAS) is an emerging topic in machine learning and computer vision. The fundamental ideology of NAS is using an automatic mechanism to replace manual designs for exploring powerful network architectures. One of the key factors of NAS is to scale-up the search space, e.g., increasing the number of operators, so that more possibilities are covered, but existing search algorithms often get lost in a large number of operators. This paper presents a scalable NAS algorithm by designing a factorizable set of architectural parameters, so that the size of the search space goes up quadratically while the burden of optimization increases linearly. As a practical example, we add a set of activation functions to the original set containing convolution, pooling and skip-connect, etc. With a marginal increase in search costs and no extra costs in retraining, we are able to find interesting architectures that were not explored before, and achieve state-of-the-art performance in CIF AR10 and ImageNet, two standard image classification benchmarks.

Many theories, based on neuroscientific and psychological empirical evidence and on computational concepts, have been elaborated to explain the emergence of consciousness in the central nervous system. These theories propose key fundamental mechanisms to explain consciousness, but they only partially connect such mechanisms to the possible functional and adaptive role of consciousness. Recently, some cognitive and neuroscientific models try to solve this gap by linking consciousness to various aspects of goal-directed behaviour, the pivotal cognitive process that allows mammals to flexibly act in challenging environments. Here we propose the Representation Internal-Manipulation (RIM) theory of consciousness, a theory that links the main elements of consciousness theories to components and functions of goal-directed behaviour, ascribing a central role for consciousness to the goal-directed manipulation of internal representations. This manipulation relies on four specific computational operations to perform the flexible internal adaptation of all key elements of goal-directed computation, from the representations of objects to those of goals, actions, and plans. Finally, we propose the concept of `manipulation agency' relating the sense of agency to the internal manipulation of representations. This allows us to propose that the subjective experience of consciousness is associated to the human capacity to generate and control a simulated internal reality that is vividly perceived and felt through the same perceptual and emotional mechanisms used to tackle the external world.

In this python machine learning course, learn both supervised and unsupervised learning in python from scratch. Enroll in this course and boost your career now In this age of big data, companies across the globe use Python to sift through the avalanche of information at their disposal. By becoming proficient in unsupervised & supervised learning in Python, you can give your company a competitive edge - and boost your career to the next level.

Engineers have created a new deep-learning software capable of assessing complex audience reactions to movies using the viewer's facial expressions. Developed by Disney Research in collaboration with Yisong Yue of Caltech and colleagues at Simon Fraser University, the software relies on a new algorithm known as factorized variational autoencoders (FVAEs). Variational autoencoders use deep learning to automatically translate images of complex objects, like faces, into sets of numerical data, also known as a latent representation or encoding. The contribution of Yue and his colleagues was to train the autoencoders to incorporate metadata (pertinent information about the data being analyzed). In the parlance of the field, they used the metadata to define an encoding space that can be factorized. In this case, the factorized variational autoencoder takes images of the faces of people watching movies and breaks them down into a series of numbers representing specific features: one number for how much a face is smiling, another for how wide open the eyes are, etc. Metadata then allow the algorithm to connect those numbers with other relevant bits of data--for example, with other images of the same face taken at different points in time, or of other faces at the same point in time.

Today we kick off our 2019 AI Rewind Series joined by Zack Lipton, a jointly appointed Professor in the Tepper School of Business and the Machine Learning Department at CMU. You might remember Zack from our conversation earlier this year, "Fairwashing" and the Folly of ML Solutionism, which you can find here. We want to hear from you! Send your thoughts on the year that was 2019 below in the comments, or via Twitter at @samcharrington or @twimlai. To follow along with the 2019 AI Rewind Series, head over to the series page! "More On That Later" by Lee Rosevere licensed under CC By 4.0

It's hard to keep up with the pace of change in the data science and machine learning fields. And when you're under pressure to deliver projects, learning new skills and technologies might be the last thing on your mind. But if you don't have at least one eye on what you need to learn next you run the risk of falling behind. In turn this means you miss out on new solutions and new opportunities to drive change: you might miss the chance to do things differently. That's why we want to make it easy for you with this quick list of what you need to watch out for and learn in 2020.