Meta tags:

Headings (most frequently used words):

antonio, gulli, and, my, on, volume, learning, 2015, programming, data, machine, interview, questions, december, hands, search, blog, solved, in, special, big, sunday, may, 2016, tuesday, subscribe, to, microsoft, elsevier, awards, demystifying, deep, collection, of, science, python, spark, coding, playground, tree, graph, bit, dynamic, design, patterns, collections, saturday, thursday, wednesday, november, 22, this, popular, posts, google, ask, com, highlander, university, ferrari, engine, shipit, patents, image, archive, list, explaining, lenet, like, code, series, experimental, sessions, with, convnets, 3rd, 7th, amazon, edition, advanced, ii, content, discoverability,

Text of the page (most frequently used words):
the (124), and (82), log (65), optimizer (47), for (46), print (37), convnet_lenet (34), adam (33), history (33), gulli (29), ago (28), antonio (28), #learning (25), with (21), share (21), this (20), add (20), years (19), are (18), from (18), num_conv (17), data (16), you (15), blog (15), that (14), search (14), keras (14), batch_size (14), one (13), import (13), google (12), deep (12), num_filters (12), all (11), first (11), which (11), have (11), comments (11), december (11), machine (11), activation (11), num_pool (11), dropout_rate (11), layer (11), validation_split (11), x_train (11), your (10), code (10), number (10), facebook (10), may (10), then (10), microsoft (9), here (9), how (9), can (9), time (9), september (9), used (9), dropout (9), shape (9), x_test (9), email (8), july (8), august (8), november (8), parameters (8), plt (8), network (8), num_epochs (8), num_hidden (8), has (7), python (7), test (7), set (7), posts (7), june (7), january (7), february (7), march (7), april (7), 2015 (7), pinterest (7), blogthis (7), pubblicato (7), codingplayground (7), examples (7), dense (7), convolutional (7), training (7), default (7), simple (6), not (6), algorithms (6), about (6), problem (6), questions (6), programming (6), october (6), hands (6), collection (6), start (6), interview (6), mnist (6), input (6), verbose (6), normalize (6), 128 (6), side (6), length (6), convolution (6), num_classes (6), into (5), wolfram (5), get (5), like (5), second (5), year (5), awards (5), results (5), many (5), atom (5), see (5), different (5), more (5), spark (5), volume (5), will (5), convnets (5), layers (5), convolution2d (5), sgd (5), rmsprop (5), fitlog (5), elapsed (5), values (5), minst (5), true (5), function (5), relu (5), model (5), cut (4), post (4), news (4), research (4), out (4), numbers (4), engine (4), new (4), architecture (4), step (4), also (4), each (4), they (4), help (4), apply (4), weeks (4), business (4), coding (4), 2026 (4), computer (4), learn (4), demystifying (4), series (4), lenet (4), elsevier (4), well (4), patterns (4), clustering (4), algorithm (4), next (4), let (4), some (4), accuracy (4), plot (4), args (4), experiment_file (4), size (4), convnet (4), maxpooling (4), square (4), rate (4), hidden (4), neurons (4), validation (4), output (4), shape_x (4), shape_y (4), y_train (4), y_test (4), float32 (4), 255 (4), operations (4), 000 (3), alpha (3), our (3), high (3), performance (3), software (3), library (3), there (3), very (3), journalists (3), world (3), produced (3), built (3), fitbit (3), last (3), seo (3), but (3), identifying (3), subjective (3), attributes (3), entities (3), flask (3), known (3), advanced (3), problems (3), need (3), understand (3), ask (3), during (3), focs (3), introduced (3), max (3), something (3), design (3), directly (3), lot (3), following (3), science (3), day (3), subscribe (3), explaining (3), 2016 (3), image (3), com (3), pca (3), classical (3), linear (3), discuss (3), where (3), boost (3), means (3), big (3), solved (3), when (3), super (3), after (3), block (3), show (3), experiments (3), flatten (3), maxpooling2d (3), np_utils (3), inspect (3), experiment (3), print_graph (3), experiment_label (3), filters (3), regularization (3), representation (3), pooling (3), connections (3), outputs (3), softmax (3), value (3), categorical_crossentropy (3), epochs (3), 256 (3), initial (3), week (2), twitter (2), got (2), best (2), free (2), might (2), morning (2), positions (2), mini (2), team (2), just (2), find (2), place (2), latticecrypto (2), implements (2), don (2), smart (2), watch (2), quick (2), building (2), use (2), technology (2), across (2), today (2), want (2), society (2), dumb (2), terminals (2), john (2), battelle (2), their (2), seems (2), good (2), decade (2), asked (2), favorite (2), trumpet (2), put (2), much (2), perhaps (2), interest (2), web (2), application (2), was (2), misinformation (2), solve (2), book (2), weather (2), impact (2), official (2), over (2), past (2), within (2), art (2), award (2), paper (2), tiktok (2), gaming (2), techcrunch (2), sovereignty (2), state (2), defined (2), while (2), 2025 (2), months (2), think (2), traffic (2), chatgpt (2), question (2), down (2), why (2), needs (2), messaging (2), mcp (2), agent (2), whatsapp (2), several (2), quite (2), gandalf (2), days (2), currently (2), does (2), complexity (2), improve (2), coffee (2), home (2), latest (2), list (2), patents (2), shipit (2), ferrari (2), university (2), highlander (2), take (2), gradient (2), full (2), gamma (2), factory (2), dbscan (2), implement (2), wikipedia (2), should (2), either (2), board (2), robot (2), solution (2), moving (2), nearest (2), neighbour (2), tree (2), adaboost (2), idea (2), weak (2), http (2), sunday (2), tuesday (2), than (2), graph (2), special (2), experimental (2), hard (2), wrote (2), humans (2), simpler (2), taught (2), capacity (2), solving (2), handwritten (2), characters (2), previous (2), effective (2), posting (2), way (2), because (2), been (2), theano (2), tensorflow (2), both (2), computation (2), hyper (2), nets (2), random (2), sequential (2), utils (2), regularizers (2), callbacks (2), earlystopping (2), def (2), png (2), fig (2), val_accuracy (2), format (2), normlize (2), decay (2), momentum (2), nesterov (2), img_channels (2), reshape (2), gpu (2), astype (2), samples (2), ohe (2), to_categorical (2), num_conf (2), rectified (2), pool_size (2), summary (2), early (2), early_stopping (2), compile (2), loss (2), fit (2), frame (2), 512 (2), 1024 (2), n_hidden (2), lecun (2), stochastic (2), converted (2), proper (2), applied (2), maxpool (2), considered (2), playground (2), opinions, mine, theme, powered, blogger, turning, comedy, club, seats, house, price, saying, enjoying, invitation, comedyfest, byob, once, deeply, humbly, quickly, looking, including, jobs, gone, eventually, final, approximately, six, half, launch, started, prior, portable, lattice, based, cryptographic, release, downloads, leave, juicy, revenue, engaging, climb, five, tips, loyal, ecommerce, customer, base, around, interactive, explorer, international, center, icfj, appeared, god, innovation, banali, searchblog, discontinued, trackers, opportunity, back, reflect, wearing, activity, badges, matt, cutts, gadgets, someone, musical, instrument, hear, challenge, raise, head, look, beyond, horizons, slice, pizza, personal, recording, harvard, people, who, read, putting, others, current, 124, stats, biased, coin, ugc, recently, granted, patent, haven, seen, sea, framework, created, armin, ronacher, its, lightweight, efficient, nature, designed, sta, projects, source, beginners, geeksforgeeks, disinformation, biggest, internet, mis, geeking, greg, posted, lizao, larry, engineer, rob, carver, scientist, accurate, forecasts, direct, generative, quantify, uncertainty, forecasting, ibm, interviewed, thousand, programmers, fair, interns, most, basic, technical, interviews, zen, recognizes, papers, conference, earlier, theory, gmv, pets, conversion, lift, studies, phone, retail, brands, drive, stronger, unveils, advertising, tools, land, engines, marketing, online, india, undergone, dramatic, transformation, brightest, stars, evolution, classic, game, rummy, nabob, royal, destination, ultimate, card, europe, mattered, since, invention, nation, borders, laws, taxes, without, reclaiming, control, digital, gigaom, three, tested, completely, changed, organic, opened, what, aio, structural, engineers, physical, stamp, plans, approve, falls, due, industry, regulation, petesearch, connects, platform, create, accounts, build, templates, send, announcing, meta, manage, developer, reasons, attended, few, conferences, workshops, however, made, exception, held, notes, process, algebra, diary, nephew, school, china, developed, strong, mat, computational, cup, ways, perfect, routine, inside, updates, announced, 2008, 382, 2009, 387, 2010, 361, 2011, 241, 2012, 120, 2013, 2014, 184, experi, lik, archive, principal, component, analisys, method, reduce, dimensionality, involves, calcu, dimensional, reduction, eigen, restarted, old, behavior, algos, structures, published, every, regression, descend, creational, abstract, builder, prototype, object, pool, singleton, struct, easy, quoting, basically, point, densit, ideally, areas, memory, layout, programs, move, right, path, steps, chose, rectangular, typical, recursive, tentatively, queen, doesn, violate, conditions, column, continue, submatrix, queens, chessboard, describes, pseudo, computing, already, kdtree, implementatio, intriguing, classifiers, edit, 013, rece, popular, family, older, www, slideshare, net, antoniogulli, demos, content, discoverability, mining, practical, handson, wednesday, 150, useful, nail, job, bit, dynamic, collections, edition, thursday, 3rd, 7th, amazon, explain, enough, recognizing, hand, written, times, thought, mate, piece, possibly, could, welcome, kingdom, certain, tasks, say, mean, bunch, them, kinda, cool, dataset, luckily, handy, guru, invented, broke, any, result, domains, turing, follow, level, lucky, cambrian, explosion, libraries, players, giving, contribution, academic, testing, lasagne, neon, decided, clean, minimalist, plus, runs, top, switch, backend, transparently, supports, cpus, gpus, third, describing, dozens, exploring, space, inferring, rules, thumbs, fine, tuning, stay, tuned, contexts, human, numpy, matplotlib, pyplot, seed, 1111, reproducibility, datasets, models, core, activity_l2, visualize_util, optimizers, save, join, 02d, sec, replace, figure, figsize, val_acc, title, ylabel, xlabel, iteration, text, savefig, classifying, 28x28, classes, digits, digit, channels, load, split, load_data, convert, pixerl, dividing, max_value, train, represent, class, required, define, input_shape, outfut, w_regularizer, stopping, monitor, val_loss, patience, loss_function, parametric, metrics, nb_epoch, evaluate, currentframe, getargvalues, printing, pretty, arguments, return, 2048, 4096, false, 100, yan, yann, exdb, sessions, saturday, now, modules, together, methods, such, descent, utility, change, accepting, multiple, beginning, images, loaded, contains, 60000, 10000, only, allowed, computations, optionally, normalized, labels, original, hot, encoding, prerequisite, classification, letnet, operation, produces, num_filter, uses, passed, through, similar, subsequent, avoiding, overfitting, additional, module, drops, repeated, identical, notice, automatically, computes, dimension, transformed, blocks, convent, two, aggregated, single, neuron, wonder, adopting, particular, another, complex, indeed, composed, depends, specific, domain, necessarily, theoretical, motivations, optimal, composition, suggestion, check, achieved, iterate, adding, until, gains, observed, cost, execution, increasing, too, know, kind, magic, important, aspect, even, relatively, outperforms, traditional, techniques, compiled, using, metric, besides, stop, criterion, adopted, regularizer, elasticnet, among, percentage, applying, acts, form, regularizations, introducing, feature, maps, batch, weights, updated, whether, divided, pixel, commentary, bigdata, stl, perl,

Text of the page (random words):
utomatically computes the dimension of the data moving and transformed across different blocks after the proper convent layers two dense layers have been introduced the first one is a layer with n_hidden neurons and the second one has num_classes outputs which are aggregated into a single neuron with softmax activation you might wonder why we are adopting this particular architecture and not another one perhaps simpler or more complex well indeed the way in which convnet and maxpool operations are composed depends a lot on the specific domain and there are not necessarily theoretical motivations explaining the optimal composition the suggestion is to start with something very simple then check the achieved performance and then iterate by adding more layers until gains are observed and the cost of execution is not increasing too much i know it seems a kind of magic but the important aspect to understand is that even a relatively simple network like this one outperforms traditional machine learning techniques the model is then compiled by using categorical_crossentropy as loss function and accuracy as metric besides that an early stop criterion is also adopted pubblicato da codingplayground a 11 50 am no comments email this blogthis share to x share to facebook share to pinterest saturday may 7 2016 demystifying deep learning series hands on experimental sessions with convnets this is the first of a series of hands on series where i ll explain deep learning step by step and with a lot of experimental results let s start from a classical but hard enough problem recognizing hand written numbers how many times have you thought is that a 4 or a 9 when your best mate wrote a number on a piece of paper well if that s hard for humans how possibly could it be simpler for a computer to learn welcome in the kingdom of deep learning where certain tasks can be taught to computer with super humans capacity and when i say taught i mean it here we don t code algorithms for solving problems no here we code algorithms for learning how to solve a problem then we take a bunch of examples and the computer will learn from them kinda of cool no so let s start first we need a dataset with handwritten characters and luckily we have one handy that s mnist http yann lecun com exdb mnist which is produced by yan lecun the guru of deep learning currently at facebook he invented something known as convnets which broke any previous result in learning in so many different application domains i think he will get the turing award one day convnets are simple and effective as we will see in follow up posting second we need some high level library for coding deep learning in a simple and effective way here we are super lucky because in the last year there has been a cambrian explosion of deep learning libraries with all the big players giving a contribution from google to facebook to microsoft to the academic world after testing many theano google s tensorflow lasagne block neon i decided to go for keras because it is clean and minimalist plus it runs on the top of theano and tensorflow which are the state of the art today and you can switch the backend transparently keras supports both cpus and gpus computation third let s show directly some code which i wrote and can get to an accuracy of 98 import numpy as np import matplotlib pyplot as plt import time np random seed 1111 for reproducibility from keras datasets import mnist from keras models import sequential from keras layers core import dense dropout activation flatten from keras layers convolutional import convolution2d maxpooling2d from keras utils import np_utils from keras regularizers import l2 activity_l2 from keras utils visualize_util import plot from keras optimizers import sgd adam rmsprop from keras callbacks import earlystopping import inspect save the graph produced by the experiment def print_graph training log fitlog elapsed time elapsed input parameters for the experiment args input values for the experiment values experiment_label n join s s i values i for i in args experiment_file experiment_label time 02d elapsed sec experiment_file experiment_file replace n png fig plt figure figsize 6 3 plt plot fitlog history val_acc plt title val_accuracy plt ylabel val_accuracy plt xlabel iteration fig text 7 15 experiment_label size 6 plt savefig experiment_file format png a lenet like convnet for classifying minst handwritten characters 28x28 def convnet_lenet verbose 1 normlize normalize true network parameters batch_size 128 num_epochs 20 number of convolutional filters num_filters 32 side length of maxpooling square num_pool 2 side length of convolution square num_conv 3 dropout rate for regularization dropout_rate 0 5 hidden number of neurons first layer num_hidden 128 validation data validation_split 0 2 20 optimizer used optimizer sgd lr 0 01 decay 1e 6 momentum 0 9 nesterov true output classes number of minst digits num_classes 10 shape of an minst digit image shape_x shape_y 28 28 channels on minst img_channels 1 load the minst data split in training and test data x_train y_train x_test y_test mnist load_data x_train x_train reshape x_train shape 0 1 shape_x shape_y x_test x_test reshape x_test shape 0 1 shape_x shape_y convert in float32 representation for gpu computation x_train x_train astype float32 x_test x_test astype float32 if normalize normalize each pixerl by dividing by max_value 255 x_train 255 x_test 255 print x_train shape x_train shape print x_train shape 0 train samples print x_test shape 0 test samples keras needs to represent each output class into ohe representation y_train np_utils to_categorical y_train num_classes y_test np_utils to_categorical y_test num_classes nn sequential first layer of convnets pooling dropout apply a num_conv x num_conf convolution with num_filters output for the first layer it is also required to define the input shape activation function is rectified linear nn add convolution2d num_filters num_conv num_conv input_shape img_channels shape_x shape_y nn add activation relu nn add convolution2d num_filters num_conv num_conv nn add activation relu nn add maxpooling2d pool_size num_pool num_pool nn add dropout dropout_rate second layer of convnets pooling dropout apply a num_conv x num_conf convolution with num_filters output nn add convolution2d num_filters num_conv num_conv nn add activation relu nn add convolution2d num_filters num_conv num_conv nn add activation relu nn add maxpooling2d pool_size num_pool num_pool nn add dropout dropout_rate flatten the shape for dense connections nn add flatten first hidden layer of dense network nn add dense num_hidden nn add activation relu nn add dropout dropout_rate outfut layer with num_classes outputs activation is softmax regularization is l2 nn add dense num_classes w_regularizer l2 0 01 nn add activation softmax summary nn summary plot the model plot nn set an early stopping value early_stopping earlystopping monitor val_loss patience 2 compile the model loss_function is categorical_crossentropy optimizer is parametric nn compile loss categorical_crossentropy optimizer optimizer metrics accuracy start time time fit the model with validation data fitlog nn fit x_train y_train batch_size batch_size nb_epoch num_epochs verbose verbose validation_split validation_split callbacks early_stopping elapsed time time start test the network results nn evaluate x_test y_test verbose verbose print accuracy results 1 just to get the list of input parameters and their value frame inspect currentframe args _ _ values inspect getargvalues frame used for printing pretty arguments print_graph fitlog elapsed args values return fitlog 2 epochs log convnet_lenet optimizer adam num_epochs 2 print log history 20 epochs log convnet_lenet optimizer adam num_epochs 20 print log history default optimizer sgd log convnet_lenet num_epochs 20 print log history default optimizer rmsprop log convnet_lenet optimizer rmsprop num_epochs 20 print log history default optimizer log convnet_lenet optimizer adam dropout_rate 0 print log history default optimizer log convnet_lenet optimizer adam dropout_rate 0 1 print log history default optimizer log convnet_lenet optimizer adam dropout_rate 0 2 print log history default optimizer log convnet_lenet optimizer adam dropout_rate 0 4 print log history default optimizer log convnet_lenet optimizer adam batch_size 64 print log history log convnet_lenet optimizer adam batch_size 128 print log history log convnet_lenet optimizer adam batch_size 256 print log history log convnet_lenet optimizer adam batch_size 512 print log history log convnet_lenet optimizer adam batch_size 1024 print log history log convnet_lenet optimizer adam batch_size 2048 print log history log convnet_lenet optimizer adam batch_size 4096 print log history log convnet_lenet optimizer adam validation_split 0 8 print log history log convnet_lenet optimizer adam validation_split 0 6 print log history log convnet_lenet optimizer adam validation_split 0 4 print log history log convnet_lenet optimizer adam validation_split 0 2 print log history log convnet_lenet optimizer adam validation_split 0 2 normalize false print log history log convnet_lenet optimizer adam validation_split 0 2 num_filters 64 print log history log convnet_lenet optimizer adam num_filters 128 print log history log convnet_lenet optimizer adam num_filters 256 print log history x log convnet_lenet optimizer adam num_pool 4 x print log history log convnet_lenet optimizer adam num_pool 8 print log history log convnet_lenet optimizer adam num_conv 4 print log history x log convnet_lenet optimizer adam num_conv 8 x print log history log convnet_lenet optimizer adam num_hidden 32 print log history log convnet_lenet optimizer adam num_hidden 64 print log history log convnet_lenet optimizer adam num_hidden 256 print log history log convnet_lenet optimizer adam num_hidden 512 print log history log convnet_lenet optimizer adam num_hidden 1024 print log history verbose 1 normlize normalize true network parameters batch_size 128 num_epochs 100 number of convolutional filters num_filters 32 side length of maxpooling square num_pool 2 side length of convolution square num_conv 3 dropout rate for regularization dropout_rate 0 5 hidden number of neurons first layer n_hidden 128 validation data validation_split 0 2 20 optimizer used optimizer sgd lr 0 01 decay 1e 6 momentum 0 9 nesterov true plt show next posting is about describing the code then you will see dozens of experiments for exploring the hyper parameters space and inferring some rules of thumbs for fine tuning our deep learning nets stay tuned during the next months we will see more than 20 nets for deep learning in different contexts and show super human capacity pubblicato da codingplayground a 1 24 pm no comments email this blogthis share to x share to facebook share to pinterest tuesday december 8 2015 3rd and 7th on amazon pubblicato da codingplayground a 12 52 am no comments email this blogthis share to x share to facebook share to pinterest thursday december 3 2015 special edition programming interview questions solved in c tree graph bit dynamic programming and design patterns special collections on programming volume 1 a collection of more than 150 interview questions in c useful to nail your next job interview pubblicato da codingplayground a 11 20 pm no comments email this blogthis share to x share to facebook share to pinterest wednesday december 2 2015 a collection of advanced data science and machine learning interview questions solved in python and spark ii volume 7 hands on big data and machine programming interview questions volume 7 advanced machine learning and handson examples on spark and python pubblicato da codingplayground a 11 17 pm no comments email this blogthis share to x share to facebook share to pinterest tuesday december 1 2015 a collection of data science interview questions solved in python and spark volume 6 hands on big data and machine learning volume 6 a new book for machine learning and data mining with practical hands on examples in spark and python pubblicato da codingplayground a 11 15 pm no comments email this blogthis share to x share to facebook share to pinterest sunday november 22 2015 elsevier machine learning content discoverability http www slideshare net antoniogulli 2015 machine learning elsevier demos pubblicato da codingplayground a 10 38 pm no comments email this blogthis share to x share to facebook share to pinterest older posts home subscribe to posts atom antonio gulli gulli family subscribe to posts atom posts all comments atom all comments search this blog popular posts k means in c k means is a classical clustering algorithm here you have a c code for k means clustering edit 12 05 013 see also my more rece adaboost improve your weak performance adaboost is one of my favorite machine learning algorithm the idea is quite intriguing you start from a set of weak classifiers and learn nearest neighbour on kd tree in c and boost wikipedia describes the pseudo code for computing the nearest neighbour nn on an already built kdtree here you have a boost implementatio place n queens on a chessboard typical recursive solution where we tentatively put a queen if this doesn t violate conditions in column i then continue in submatrix a robot is moving in a rectangular board it can move either down or right and the board is n x m how many path does the robot have solution steps are n m and we can chose n discuss memory layout for c programs ideally you should discuss all the different areas that are used dbscan clustering algorithm dbscan is a well known clustering algorithm which is easy to implement quoting wikipedia basically a point q is directly densit design patterns c full collection of gamma s patterns full collection of gamma s patterns in c creational abstract factory builder factory prototype object pool singleton struct learning linear regression with gradient descend last week i restarted an old and good behavior see a collection of algos and data structures published here every day i take an well k pca dimensional reduction in eigen pca principal component analisys is a classical machine learning method to reduce the dimensionality of a problem pca involves the calcu antonio gulli google google antonio gulli microsoft antonio gulli microsoft antonio gulli ask com antonio gulli ask com antonio gulli highlander antonio gulli highlander antonio gulli university antonio gulli university antonio gulli elsevier antonio gulli elsevier antonio gulli my ferrari antonio gulli my ferrari antonio gulli my search engine antonio gulli my search engine antonio gulli my shipit microsoft antonio gulli my shipit microsoft antonio gulli my patents antonio gulli my patents antonio gulli my awards antonio gulli my awards antonio gulli my awards antonio gulli my awards antonio gulli image search antonio gulli image search blog archive 2016 2 may 2 demystifying deep le...