Neural nets

Designing the fanciest usable differentiable loss surface

October 14, 2016 — September 9, 2020

machine learning
neural nets
optimization
Figure 1: Bjorn Stenger’s brief history of machine learning.

Modern computational neural network methods reascend the hype phase transition. a.k.a deep learning or double plus fancy brainbots or please give the department have a bigger GPU budget it’s not to play video games I swear.

I don’t intend to write an introduction to deep learning here; that ground has been tilled already.

But here are some handy links to resources I frequently use and a bit of under-discussed background.

1 What?

To be specific, deep learning is

  • a library of incremental improvements in areas such as Stochastic Gradient Descent, approximation theory, graphical models, and signal processing research, plus some handy advancements in SIMD architectures that, taken together, surprisingly elicit the kind of results from machine learning that everyone was hoping we’d get by at least 20 years ago, yet without requiring us to develop substantially more clever grad students to do so, or,
  • the state-of-the-art in artificial kitten recognition.
  • a metatstatizing buzzword

It’s a frothy (some might say foamy-mouthed) research bubble right now, with such cuteness at the extrema as, e.g. Inceptionising inceptionism (Andrychowicz et al. 2016) which learns to learn neural networks using neural networks. (well, it sort of does that, but is a long way from a bootstrapping general AI) Stay tuned for more of this.

There is not much to do with “neurons” left in the paradigm at this stage. What there is, is a bundle of clever tricks for training deep constrained hierarchical predictors and classifiers on modern computer hardware. Something closer to a convenient technology stack than a single “theory”.

Some network methods hew closer to behaviour of real neurons, although not that close; simulating actual brains is a different discipline with only intermittent and indirect connection.

Subtopics of interest to me:

2 Why bother?

There are many answers.

2.1 The ultimate regression algorithm

…until the next ultimate regression algorithm.

It turns out that this particular learning model (class of learning models) and training technologies is surprisingly good at getting every better models out of ever more data. Why burn three grad students on a perfect tractable and specific regression algorithm when you can use one algorithm to solve a whole bunch of regression problems, and which improves with the number of computers and the amount of data you have? How much of a relief is it to capital to decouple its effectiveness from the uncertainty and obstreperousness of human labour?

2.2 Cool maths

Function approximations, interesting manifold inference. Weird product measure things, e.g. (Montufar 2014).

Even the stuff I’d assumed was trivial, like backpropagation, has a few wrinkles in practice. See Michael Nielson’s chapter and Chrisopher Olah’s visual summary.

Yes, this is a regular paper mill. Not only are there probably new insights to be had, but also you can recycle any old machine learning insight, replace a layer in a network with that and poof — new paper.

2.3 Insight into the mind

🏗 Maybe.

There claims to be communication between real neurology and neural networks in computer vision, but elsewhere neural networks are driven by their similarities to other things, such as being differentiable relaxations of traditional models, (differentiable stack machines!) or of being license to fit hierarchical models without regard for statistical niceties.

There might be some kind of occasional “stylised fact”-type relationship.

For some works which lean harder in to this, try neural neural networks.

2.4 Trippy art projects

See generative art and neural networks

3 Hip keywords for NN models

Not necessarily mutually exclusive; some design patterns you can use.

There are many summaries floating around. Some that I looked at are Tomasz Malisiewicz’s summary of Deep Learning Trends @ ICLR 2016, or the Neural network zoo or Simon Brugman’s deep learning papers.

Some of these are descriptions of topologies, others of training tricks or whatever. Recurrent and convolutional are two types of topologies you might have in your ANN. But there are so many other possible ones: “Grid”, “highway”, “Turing” others…

Many are mentioned in passing in David McAllester’s Cognitive Architectures post.

3.1 Probabilistic/variational

See probabilistic Neural Networks.

3.2 Convolutional

See the convnets entry.

3.3 Generative Adversarial Networks

Train two networks to beat each other.

3.4 Recurrent neural networks

Feedback neural networks structures to have with memory and a notion of time and “current” versus “past” state. See recurrent neural networks.

3.5 Transfer learning

I have seen two versions of this term.

One starts from the idea that if you have, say, a network that solves, some particular computer vision problem well, possibly you can use them to solve another one without starting from scratch on another computer vision problem. This is the Recycling someone else’s features framing. I don’t know why this has a special term — I think it’s so that you can claim to do “end-to-end” learning, but then actually do what everyone else as done forever and works totally OK, which is to re-use other people’s work like real scientists.

The other version is you would like to do domain adaptation, which is to say, to learn on one dataset but still make good predictions on a different dataset.

These two things can clearly be related if you squint hard. Using ‘transfer learning’ in this second sense irritates me slightly because it already has so many names: I would describe that problem as external validity, instead of domain adaptation but other names spotted in the wild include dataset shift, covariate shift, data fusion and there are probably more. This is a fundamental problem in statistics, and the philosophy of science generally, and has been for a long time.

3.6 Attention mechanism

See Attention mechanism.

3.7 Spike-based

Most simulated neural networks are based on a continuous activation potential and discrete time, unlike spiking biological ones, which are driven by discrete events in continuous time. There are a great many other differences (to real biology). What difference does this in particular make? I suspect it means that time is handled different.

3.8 Kernel networks

Kernel trick + ANN = kernel ANNs.

(Stay tuned for reframing more things as deep learning.)

Is this what convex networks (Bengio et al. 2005) are?

Francis Bach:

Figure 2: I’m sure the brain totes does this

AFAICT these all boil down to rebadged extensions of Gaussian processes but maybe I’m missing something?

3.9 Autoencoding

🏗 Making a sparse encoding of something by demanding your network reproduces the after passing the network activations through a narrow bottleneck. Many flavours.

4 Optimisation methods

Backpropagation plus stochastic gradient descent rules at the moment.

Does anything else get performance at this scale? What other techniques can be extracted from variational inference or MC sampling, or particle filters, since there is no clear reason that shoving any of these in as intermediate layers in the network is any less well-posed than a classical backprop layer? Although it does require more nous from the enthusiastic grad student.

5 Preventing overfitting

See regularising deep learning.

6 Activations for neural networks

See activation functions

7 Implementing

See implementing neural nets.

8 References

Amari, Shunichi. 1967. A Theory of Adaptive Pattern Classifiers.” IEEE Transactions on Electronic Computers.
Amari, Shun-ichi. 1998. Natural Gradient Works Efficiently in Learning.” Neural Computation.
Andrychowicz, Denil, Gomez, et al. 2016. Learning to Learn by Gradient Descent by Gradient Descent.” arXiv:1606.04474 [Cs].
Arel, Rose, and Karnowski. 2010. Deep Machine Learning - A New Frontier in Artificial Intelligence Research [Research Frontier].” IEEE Computational Intelligence Magazine.
Arora, Ge, Ma, et al. 2015. Simple, Efficient, and Neural Algorithms for Sparse Coding.” In Proceedings of The 28th Conference on Learning Theory.
Bach. 2014. Breaking the Curse of Dimensionality with Convex Neural Networks.” arXiv:1412.8690 [Cs, Math, Stat].
Baldassi, Borgs, Chayes, et al. 2016. Unreasonable Effectiveness of Learning Neural Networks: From Accessible States and Robust Ensembles to Basic Algorithmic Schemes.” Proceedings of the National Academy of Sciences.
Barron. 1993. Universal Approximation Bounds for Superpositions of a Sigmoidal Function.” IEEE Transactions on Information Theory.
Baydin, Pearlmutter, and Siskind. 2016. Tricks from Deep Learning.” arXiv:1611.03777 [Cs, Stat].
Bengio. 2009. Learning Deep Architectures for AI.
Bengio, Courville, and Vincent. 2013. Representation Learning: A Review and New Perspectives.” IEEE Transactions on Pattern Analysis and Machine Intelligence.
Bengio, and LeCun. 2007. Scaling Learning Algorithms Towards AI.” Large-Scale Kernel Machines.
Bengio, Roux, Vincent, et al. 2005. Convex Neural Networks.” In Advances in Neural Information Processing Systems.
Boser. 1991. An Analog Neural Network Processor with Programmable Topology.” J. Solid State Circuits.
Brock, Lim, Ritchie, et al. 2017. FreezeOut: Accelerate Training by Progressively Freezing Layers.” arXiv:1706.04983 [Cs, Stat].
Cadieu. 2014. Deep Neural Networks Rival the Representation of Primate It Cortex for Core Visual Object Recognition.” PLoS Comp. Biol.
Chen, Goodfellow, and Shlens. 2015. Net2Net: Accelerating Learning via Knowledge Transfer.” arXiv:1511.05641 [Cs].
Choromanska, Henaff, Mathieu, et al. 2015. The Loss Surfaces of Multilayer Networks.” In Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics.
Cho, van Merriënboer, Bahdanau, et al. 2014. On the Properties of Neural Machine Translation: Encoder-Decoder Approaches.” arXiv Preprint arXiv:1409.1259.
Ciodaro. 2012. Online Particle Detection with Neural Networks Based on Topological Calorimetry Information.” J. Phys. Conf. Series.
Ciresan. 2012. Multi-Column Deep Neural Network for Traffic Sign Classification.” Neural Networks.
Cybenko. 1989. Approximation by Superpositions of a Sigmoidal Function.” Mathematics of Control, Signals and Systems.
Dahl. 2012. Context-Dependent Pre-Trained Deep Neural Networks for Large Vocabulary Speech Recognition.” IEEE Transactions on Audio, Speech and Language Processing.
Dauphin, Pascanu, Gulcehre, et al. 2014. Identifying and Attacking the Saddle Point Problem in High-Dimensional Non-Convex Optimization.” In Advances in Neural Information Processing Systems 27.
Dieleman, and Schrauwen. 2014. End to End Learning for Music Audio.” In 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
Erhan, Bengio, Courville, et al. 2010. Why Does Unsupervised Pre-Training Help Deep Learning? Journal of Machine Learning Research.
Farabet. 2013. Learning Hierarchical Features for Scene Labeling.” IEEE Transactions on Pattern Analysis and Machine Intelligence.
Fukumizu, and Amari. 2000. Local Minima and Plateaus in Hierarchical Structures of Multilayer Perceptrons.” Neural Networks.
Fukushima, and Miyake. 1982. Neocognitron: A New Algorithm for Pattern Recognition Tolerant of Deformations and Shifts in Position.” Pattern Recognition.
Gal, and Ghahramani. 2016. A Theoretically Grounded Application of Dropout in Recurrent Neural Networks.” In arXiv:1512.05287 [Stat].
Garcia. 2004. Convolutional Face Finder: A Neural Architecture for Fast and Robust Face Detection.” IEEE Transactions on Pattern Analysis and Machine Intelligence.
Gatys, Ecker, and Bethge. 2015. A Neural Algorithm of Artistic Style.” arXiv:1508.06576 [Cs, q-Bio].
Giryes, Raja, Sapiro, and Bronstein. 2014. On the Stability of Deep Networks.” arXiv:1412.5896 [Cs, Math, Stat].
Giryes, R., Sapiro, and Bronstein. 2016. Deep Neural Networks with Random Gaussian Weights: A Universal Classification Strategy? IEEE Transactions on Signal Processing.
Globerson, and Livni. 2016. Learning Infinite-Layer Networks: Beyond the Kernel Trick.” arXiv:1606.05316 [Cs].
Goodfellow, Ian, Pouget-Abadie, Mirza, et al. 2014. Generative Adversarial Nets.” In Advances in Neural Information Processing Systems 27. NIPS’14.
Goodfellow, Ian J., Shlens, and Szegedy. 2014. Explaining and Harnessing Adversarial Examples.” arXiv:1412.6572 [Cs, Stat].
Goodfellow, Ian J., Vinyals, and Saxe. 2014. Qualitatively Characterizing Neural Network Optimization Problems.” arXiv:1412.6544 [Cs, Stat].
Hadsell, Chopra, and LeCun. 2006. Dimensionality Reduction by Learning an Invariant Mapping.” In 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.
Hasson, Nastase, and Goldstein. 2020. Direct Fit to Nature: An Evolutionary Perspective on Biological and Artificial Neural Networks.” Neuron.
Helmstaedter. 2013. Connectomic Reconstruction of the Inner Plexiform Layer in the Mouse Retina.” Nature.
He, Wang, and Hopcroft. 2016. A Powerful Generative Model Using Random Weights for the Deep Image Representation.” In Advances in Neural Information Processing Systems.
Hinton, G. E. 1995. The Wake-Sleep Algorithm for Unsupervised Neural Networks.” Science.
Hinton, Geoffrey E. 2007. To Recognize Shapes, First Learn to Generate Images.” In Progress in Brain Research. Computational Neuroscience: Theoretical Insights into Brain Function.
Hinton, Geoffrey. 2010. A Practical Guide to Training Restricted Boltzmann Machines.” In Neural Networks: Tricks of the Trade. Lecture Notes in Computer Science 7700.
Hinton, G., Deng, Yu, et al. 2012. Deep Neural Networks for Acoustic Modeling in Speech Recognition: The Shared Views of Four Research Groups.” IEEE Signal Processing Magazine.
Hinton, G, Osindero, and Teh. 2006. A Fast Learning Algorithm for Deep Belief Nets.” Neural Computation.
Hinton, Geoffrey E., and Salakhutdinov. 2006. Reducing the Dimensionality of Data with Neural Networks.” Science.
Hornik, Stinchcombe, and White. 1989. Multilayer Feedforward Networks Are Universal Approximators.” Neural Networks.
Huang, and Siew. 2005. Extreme Learning Machine with Randomly Assigned RBF Kernels.” International Journal of Information Technology.
Huang, Wang, and Lan. 2011. Extreme Learning Machines: A Survey.” International Journal of Machine Learning and Cybernetics.
Huang, Zhu, and Siew. 2004. Extreme Learning Machine: A New Learning Scheme of Feedforward Neural Networks.” In 2004 IEEE International Joint Conference on Neural Networks, 2004. Proceedings.
———. 2006. Extreme Learning Machine: Theory and Applications.” Neurocomputing, Neural Networks Selected Papers from the 7th Brazilian Symposium on Neural Networks (SBRN ’04) 7th Brazilian Symposium on Neural Networks,.
Hubel. 1962. Receptive Fields, Binocular Interaction, and Functional Architecture in the Cat’s Visual Cortex.” J. Physiol.
Hu, Pehlevan, and Chklovskii. 2014. A Hebbian/Anti-Hebbian Network for Online Sparse Dictionary Learning Derived from Symmetric Matrix Factorization.” In 2014 48th Asilomar Conference on Signals, Systems and Computers.
Jaderberg, Czarnecki, Osindero, et al. 2016. Decoupled Neural Interfaces Using Synthetic Gradients.” arXiv:1608.05343 [Cs].
Kaiser, and Sutskever. 2015. Neural GPUs Learn Algorithms.” arXiv:1511.08228 [Cs].
Kalchbrenner, Danihelka, and Graves. 2016. Grid Long Short-Term Memory.” arXiv:1507.01526 [Cs].
Kavukcuoglu, Ranzato, and LeCun. 2010. Fast Inference in Sparse Coding Algorithms with Applications to Object Recognition.” arXiv:1010.3467 [Cs].
Kingma, Salimans, Jozefowicz, et al. 2016. Improving Variational Inference with Inverse Autoregressive Flow.” In Advances in Neural Information Processing Systems 29.
Krizhevsky, Sutskever, and Hinton. 2012. Imagenet Classification with Deep Convolutional Neural Networks.” In Advances in Neural Information Processing Systems.
Kulkarni, Whitney, Kohli, et al. 2015. Deep Convolutional Inverse Graphics Network.” arXiv:1503.03167 [Cs].
Larsen, Sønderby, Larochelle, et al. 2015. Autoencoding Beyond Pixels Using a Learned Similarity Metric.” arXiv:1512.09300 [Cs, Stat].
Lawrence. 1997. Face Recognition: A Convolutional Neural-Network Approach.” IEEE Transactions on Neural Networks.
LeCun, Y. 1998. Gradient-Based Learning Applied to Document Recognition.” Proceedings of the IEEE.
LeCun, Yann, Bengio, and Hinton. 2015. Deep Learning.” Nature.
LeCun, Yann, Chopra, Hadsell, et al. 2006. A Tutorial on Energy-Based Learning.” In Predicting Structured Data.
Lee, Wee Sun, Bartlett, and Williamson. 1996. Efficient Agnostic Learning of Neural Networks with Bounded Fan-in.” IEEE Transactions on Information Theory.
Lee, Honglak, Grosse, Ranganath, et al. 2009. Convolutional Deep Belief Networks for Scalable Unsupervised Learning of Hierarchical Representations.” In Proceedings of the 26th Annual International Conference on Machine Learning. ICML ’09.
Leung. 2014. Deep Learning of the Tissue-Regulated Splicing Code.” Bioinformatics.
Liang, Tomczak, Johnson, et al. n.d. “BachBot: Deep Generative Modeling of Bach Chorales.”
Lin, and Tegmark. 2016a. Critical Behavior from Deep Dynamics: A Hidden Dimension in Natural Language.” arXiv:1606.06737 [Cond-Mat].
———. 2016b. Why Does Deep and Cheap Learning Work so Well? arXiv:1608.08225 [Cond-Mat, Stat].
Lipton. 2016a. Stuck in a What? Adventures in Weight Space.” arXiv:1602.07320 [Cs].
———. 2016b. The Mythos of Model Interpretability.” In arXiv:1606.03490 [Cs, Stat].
Lipton, Berkowitz, and Elkan. 2015. A Critical Review of Recurrent Neural Networks for Sequence Learning.” arXiv:1506.00019 [Cs].
Ma. 2015. Deep Neural Nets as a Method for Quantitative Structure-Activity Relationships.” J. Chem. Inf. Model.
Maclaurin, Duvenaud, and Adams. 2015. Gradient-Based Hyperparameter Optimization Through Reversible Learning.” In Proceedings of the 32nd International Conference on Machine Learning.
Mallat. 2012. Group Invariant Scattering.” Communications on Pure and Applied Mathematics.
———. 2016. Understanding Deep Convolutional Networks.” arXiv:1601.04920 [Cs, Stat].
Mehta, and Schwab. 2014. An Exact Mapping Between the Variational Renormalization Group and Deep Learning.”
Mikolov, Chen, Corrado, et al. 2013. Efficient Estimation of Word Representations in Vector Space.” arXiv:1301.3781 [Cs].
Mikolov, Le, and Sutskever. 2013. Exploiting Similarities Among Languages for Machine Translation.” arXiv:1309.4168 [Cs].
Mnih. 2015. Human-Level Control Through Deep Reinforcement Learning.” Nature.
Mohamed, Dahl, and Hinton. 2012. Acoustic Modeling Using Deep Belief Networks.” IEEE Transactions on Audio, Speech, and Language Processing.
Monner, and Reggia. 2012. A Generalized LSTM-Like Training Algorithm for Second-Order Recurrent Neural Networks.” Neural Networks.
Montufar. 2014. When Does a Mixture of Products Contain a Product of Mixtures? J. Discrete Math.
Mousavi, and Baraniuk. 2017. Learning to Invert: Signal Recovery via Deep Convolutional Networks.” In ICASSP.
Ning. 2005. Toward Automatic Phenotyping of Developing Embryos from Videos.” IEEE Transactions on Image Processing.
Nøkland. 2016. Direct Feedback Alignment Provides Learning in Deep Neural Networks.” In Advances In Neural Information Processing Systems.
Olshausen, B. A., and Field. 1996. Natural image statistics and efficient coding.” Network (Bristol, England).
Olshausen, Bruno A., and Field. 1996. Emergence of Simple-Cell Receptive Field Properties by Learning a Sparse Code for Natural Images.” Nature.
Olshausen, Bruno A, and Field. 2004. Sparse Coding of Sensory Inputs.” Current Opinion in Neurobiology.
Pan, Dong, and Guo. 2016. DropNeuron: Simplifying the Structure of Deep Neural Networks.” arXiv:1606.07326 [Cs, Stat].
Parisotto, and Salakhutdinov. 2017. Neural Map: Structured Memory for Deep Reinforcement Learning.” arXiv:1702.08360 [Cs].
Pascanu, Dauphin, Ganguli, et al. 2014. On the Saddle Point Problem for Non-Convex Optimization.” arXiv:1405.4604 [Cs].
Paul, and Venkatasubramanian. 2014. Why Does Deep Learning Work? - A Perspective from Group Theory.” arXiv:1412.6621 [Cs, Stat].
Pinkus. 1999. Approximation Theory of the MLP Model in Neural Networks.” Acta Numerica.
Radford, Metz, and Chintala. 2015. Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks.” In arXiv:1511.06434 [Cs].
Ranzato, M. 2013. Modeling Natural Images Using Gated MRFs.” IEEE Transactions on Pattern Analysis and Machine Intelligence.
Ranzato, Marc’aurelio, Boureau, and Cun. 2008. Sparse Feature Learning for Deep Belief Networks.” In Advances in Neural Information Processing Systems 20.
Rumelhart, Hinton, and Williams. 1986. Learning Representations by Back-Propagating Errors.” Nature.
Sagun, Guney, Arous, et al. 2014. Explorations on High Dimensional Landscapes.” arXiv:1412.6615 [Cs, Stat].
Salimans, and Kingma. 2016. Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks.” In Advances in Neural Information Processing Systems 29.
Scardapane, Comminiello, Hussain, et al. 2016. Group Sparse Regularization for Deep Neural Networks.” arXiv:1607.00485 [Cs, Stat].
Schmidhuber. 2022. Annotated History of Modern AI and Deep Learning.”
Shazeer, Mirhoseini, Maziarz, et al. 2017. Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer.” arXiv:1701.06538 [Cs, Stat].
Shwartz-Ziv, and Tishby. 2017. Opening the Black Box of Deep Neural Networks via Information.” arXiv:1703.00810 [Cs].
Smith, and Topin. 2017. Exploring Loss Function Topology with Cyclical Learning Rates.” arXiv:1702.04283 [Cs].
Springenberg, Dosovitskiy, Brox, et al. 2014. Striving for Simplicity: The All Convolutional Net.” In Proceedings of International Conference on Learning Representations (ICLR) 2015.
Starr. 1913. Organic and functional nervous diseases; a text-book of neurology.
Telgarsky. 2015. Representation Benefits of Deep Feedforward Networks.” arXiv:1509.08101 [Cs].
Turaga. 2010. Convolutional Networks Can Learn to Generate Affinity Graphs for Image Segmentation.” Neural Comput.
Urban, Geras, Kahou, et al. 2016. Do Deep Convolutional Nets Really Need to Be Deep (Or Even Convolutional)? arXiv:1603.05691 [Cs, Stat].
Oord, Aaron van den, Dieleman, Zen, et al. 2016. WaveNet: A Generative Model for Raw Audio.” In 9th ISCA Speech Synthesis Workshop.
Oord, Aäron van den, Kalchbrenner, and Kavukcuoglu. 2016. Pixel Recurrent Neural Networks.” arXiv:1601.06759 [Cs].
Oord, Aäron van den, Kalchbrenner, Vinyals, et al. 2016. Conditional Image Generation with PixelCNN Decoders.” arXiv:1606.05328 [Cs].
ver Steeg, and Galstyan. 2015. The Information Sieve.” arXiv:1507.02284 [Cs, Math, Stat].
Wiatowski, and Bölcskei. 2015. A Mathematical Theory of Deep Convolutional Neural Networks for Feature Extraction.” In Proceedings of IEEE International Symposium on Information Theory.
Wiatowski, Grohs, and Bölcskei. 2018. Energy Propagation in Deep Convolutional Neural Networks.” IEEE Transactions on Information Theory.
Xie, Liang, and Song. 2016. Diversity Leads to Generalization in Neural Networks.” arXiv:1611.03131 [Cs, Stat].
Yu, and Deng. 2011. Deep Learning and Its Applications to Signal and Information Processing [Exploratory DSP].” IEEE Signal Processing Magazine.
Zhang, Chiyuan, Bengio, Hardt, et al. 2017. Understanding Deep Learning Requires Rethinking Generalization.” In Proceedings of ICLR.
Zhang, Sixin, Choromanska, and LeCun. 2015. Deep Learning with Elastic Averaging SGD.” In Advances In Neural Information Processing Systems.