Oceania
Artificial intelligence: Professor Toby Walsh on 10 ways society will change by 2050
Go player Lee Sedol (R) during the third game of the Google DeepMind Challenge Match against Google-developed supercomputer AlphaGo. Leading Australian artificial intelligence scientist Professor Toby Walsh is warning that we are "sleepwalking" into an AI future in which billions of machines and computers will be able to think. Professor Walsh, from the University of New South Wales, is calling for a national discussion about whether society needs to adopt clear boundaries and guidelines around how AI is developed and how it's used in our lives. In his book It's Alive: Artificial Intelligence From The Logic Piano to Killer Robots, he has highlighted key questions in a series of predictions that describe how our future could be far better or far worse because of AI. Here's how he thinks society might change by 2050 thanks to artificial intelligence.
Continual Learning for Sentence Representations Using Conceptors
Liu, Tianlin, Ungar, Lyle, Sedoc, João
Distributed representations of sentences have become ubiquitous in natural language processing tasks. In this paper, we consider a continual learning scenario for sentence representations: Given a sequence of corpora, we aim to optimize the sentence encoder with respect to the new corpus while maintaining its accuracy on the old corpora. To address this problem, we propose to initialize sentence encoders with the help of corpus-independent features, and then sequentially update sentence encoders using Boolean operations of conceptor matrices to learn corpus-dependent features. We evaluate our approach on semantic textual similarity tasks and show that our proposed sentence encoder can continually learn features from new corpora while retaining its competence on previously encountered corpora.
TTS Skins: Speaker Conversion via ASR
Polyak, Adam, Wolf, Lior, Taigman, Yaniv
We present a fully convolutional wav-to-wav network for converting between speakers' voices, without relying on text. Our network is based on an encoder-decoder architecture, where the encoder is pre-trained for the task of Automatic Speech Recognition (ASR), and a multi-speaker waveform decoder is trained to reconstruct the original signal in an autoregressive manner. We train the network on narrated audiobooks, and demonstrate the ability to perform multi-voice TTS in those voices, by converting the voice of a TTS robot. We observe no degradation in the quality of the generated voices, in comparison to the reference TTS voice. The modularity of our approach, which separates the target voice generation from the TTS module, enables client-side personalized TTS in a privacy-aware manner.
Disentangled Representation Learning with Information Maximizing Autoencoder
Haque, Kazi Nazmul, Latif, Siddique, Rana, Rajib
Learning disentangled representation from any unlabelled data is a nontrivial problem. In this paper we propose Information Maximising Autoencoder (InfoAE) where the encoder learns powerful disentangled representation through maximizing the mutual information between the representation and given information in an unsupervised fashion. We have evaluated our model on MNIST dataset and achieved 98.9 ( .1) Learning disentangled representation from any unlabelled data is an active area of research [1]. Self supervised learning [2, 3, 4] is a way to learn representation from the unlabelled data but the supervised signal is needed to be developed manually, which usually varies depending on the problem and the dataset. Generative Adversarial Neural Networks (GANs) [5] is a potential candidate for learning disentangled representation from unlabelled data ([6, 7, 8]).
COCO-GAN: Generation by Parts via Conditional Coordinating
Lin, Chieh Hubert, Chang, Chia-Che, Chen, Yu-Sheng, Juan, Da-Cheng, Wei, Wei, Chen, Hwann-Tzong
Humans can only interact with part of the surrounding environment due to biological restrictions. Therefore, we learn to reason the spatial relationships across a series of observations to piece together the surrounding environment. Inspired by such behavior and the fact that machines also have computational constraints, we propose \underline{CO}nditional \underline{CO}ordinate GAN (COCO-GAN) of which the generator generates images by parts based on their spatial coordinates as the condition. On the other hand, the discriminator learns to justify realism across multiple assembled patches by global coherence, local appearance, and edge-crossing continuity. Despite the full images are never generated during training, we show that COCO-GAN can produce \textbf{state-of-the-art-quality} full images during inference. We further demonstrate a variety of novel applications enabled by teaching the network to be aware of coordinates. First, we perform extrapolation to the learned coordinate manifold and generate off-the-boundary patches. Combining with the originally generated full image, COCO-GAN can produce images that are larger than training samples, which we called "beyond-boundary generation". We then showcase panorama generation within a cylindrical coordinate system that inherently preserves horizontally cyclic topology. On the computation side, COCO-GAN has a built-in divide-and-conquer paradigm that reduces memory requisition during training and inference, provides high-parallelism, and can generate parts of images on-demand.
bcr vidcast 107: AI governance, what are AI and ML, and the future is not here yet - Better Communication Results
Vikram Mahidhar reminds us all that AI is only as good as the humans supervising it and programming it. The biases and artefacts that come out of the processing are reflective of the biases programmed in at the beginning. A program trained to recognise totalled car bodies for insurance purposes, for example, will need close supervision of its decision-making outputs, for regulatory and consumer confidence and acceptance of the decision. There is a call and a growth in a new class of AI--one that is explainable, and that builds trust by providing evidence. Vikram also reminds us that a governance strategy is key to engendering trust in our organisation, processes and people.
Exploiting Event Log Data-Attributes in RNN Based Prediction
Hinkka, Markku, Lehto, Teemu, Heljanko, Keijo
In predictive process analytics, current and historical process data in event logs are used to predict future. E.g., to predict the next activity or how long a process will still require to complete. Recurrent neural networks (RNN) and its subclasses have been demonstrated to be well suited for creating prediction models. Thus far, event attributes have not been fully utilized in these models. The biggest challenge in exploiting them in prediction models is the potentially large amount of event attributes and attribute values. We present a novel clustering technique which allows for trade-offs between prediction accuracy and the time needed for model training and prediction. As an additional finding, we also found that this clustering method combined with having raw event attribute values provides even better prediction accuracy at the cost of additional time required for training and prediction. We also built a highly configurable test framework that can be used to efficiently evaluate different prediction approaches and parameterizations.
Random Projection in Neural Episodic Control
Nishio, Daichi, Yamane, Satoshi
End-to-end deep reinforcement learning has enabled agents to learn with little preprocessing by humans. However, it is still difficult to learn stably and efficiently because the learning method usually uses a nonlinear function approximation. Neural Episodic Control (NEC), which has been proposed in order to improve sample efficiency, is able to learn stably by estimating action values using a non-parametric method. In this paper, we propose an architecture that incorporates random projection into NEC to train with more stability. In addition, we verify the effectiveness of our architecture by Atari's five games. The main idea is to reduce the number of parameters that have to learn by replacing neural networks with random projection in order to reduce dimensions while keeping the learning end-to-end.
P\'olygamma Data Augmentation to address Non-conjugacy in the Bayesian Estimation of Mixed Multinomial Logit Models
Bansal, Prateek, Krueger, Rico, Bierlaire, Michel, Daziano, Ricardo A., Rashidi, Taha H.
The standard Gibbs sampler of Mixed Multinomial Logit (MMNL) models involves sampling from conditional densities of utility parameters using Metropolis-Hastings (MH) algorithm due to unavailability of conjugate prior for logit kernel. To address this non-conjugacy concern, we propose the application of P\'olygamma data augmentation (PG-DA) technique for the MMNL estimation. The posterior estimates of the augmented and the default Gibbs sampler are similar for two-alternative scenario (binary choice), but we encounter empirical identification issues in the case of more alternatives ($J \geq 3$).
How does it feel to be watched at work all the time?
Is workplace surveillance about improving productivity or simply a way to control staff and weed out poor performers? Courtney Hagen Ford, 34, left her job working as a bank teller because she found the surveillance she was under was "dehumanising". Her employer logged her keystrokes and used software to monitor how many of the customers she helped went on to take out loans and fee-paying accounts. "The sales pressure was relentless," she recalls. She decided selling fast food would be better, but ironically, left the bank to do a doctorate in surveillance technology.