# \#sequence-memory

**URL:** https://discourse.numenta.org/tag/sequence-memory/2.md

[Latest](https://discourse.numenta.org/latest.md) · [Categories](https://discourse.numenta.org/categories.md) · [Tags](https://discourse.numenta.org/tags.md)

---

## [Exploring cognitive skill profiles through training task performance](https://discourse.numenta.org/t/exploring-cognitive-skill-profiles-through-training-task-performance/12373)

<div class="topic-metadata">

**Author:** [@Emir\_Baycan](https://discourse.numenta.org/u/Emir_Baycan)\
**Replies:** 0\
**Last updated:** [May 12, 2026, 3:27am UTC](https://discourse.numenta.org/t/exploring-cognitive-skill-profiles-through-training-task-performance/12373 "2026-05-12T03:27:04Z")

</div>

We’ve been building an online cognitive testing platform and recently expanded it with a separate training section focused on specific cognitive functions rather than only IQ scoring. One interesting pattern we kept not…

---

## [Using HTM for logical deduction/induction and prediction of sentences](https://discourse.numenta.org/t/using-htm-for-logical-deduction-induction-and-prediction-of-sentences/12195)

<div class="topic-metadata">

**Author:** [@Aidan\_Busby](https://discourse.numenta.org/u/Aidan_Busby)\
**Replies:** 8\
**Last updated:** [January 27, 2026, 2:35pm UTC](https://discourse.numenta.org/t/using-htm-for-logical-deduction-induction-and-prediction-of-sentences/12195 "2026-01-27T14:35:14Z")

</div>

Hello all, I have recently been experimenting with the HTM.core code to train an HTM network to learn the logic behind a corpus of sentences. Currently, they look like this: “If A is in X, where is A?” Here is my trai…

---

## [SDR Implimentation of neural network Layers](https://discourse.numenta.org/t/sdr-implimentation-of-neural-network-layers/11521)

<div class="topic-metadata">

**Author:** [@LALAK\_YADAV](https://discourse.numenta.org/u/LALAK_YADAV)\
**Replies:** 5\
**Last updated:** [September 24, 2024, 12:15pm UTC](https://discourse.numenta.org/t/sdr-implimentation-of-neural-network-layers/11521 "2024-09-24T12:15:45Z")

</div>

I’ve been trying to apply Sparse Distributed Representations (SDR) to simple neural network layers. Can anyone suggest a method for achieving this? I want to ensure that only 2-5% of the neurons are active in each layer. …

---

## [Missing cortical column concept in some papers](https://discourse.numenta.org/t/missing-cortical-column-concept-in-some-papers/11509)

<div class="topic-metadata">

**Author:** [@ivanislav](https://discourse.numenta.org/u/ivanislav)\
**Replies:** 1\
**Last updated:** [September 22, 2024, 7:24pm UTC](https://discourse.numenta.org/t/missing-cortical-column-concept-in-some-papers/11509 "2024-09-22T19:24:04Z")

</div>

In earlier papers (2016), cortical columns were an important concept and they were explicitly modeled. In particular, only one cell would be activated at a time (for learned sequences) within a cortical column. However, …

---

## [How do HTM learn a sequence of the same input?](https://discourse.numenta.org/t/how-do-htm-learn-a-sequence-of-the-same-input/11364)

<div class="topic-metadata">

**Author:** [@Hui\_Liu](https://discourse.numenta.org/u/Hui_Liu)\
**Replies:** 1\
**Last updated:** [May 22, 2024, 7:04pm UTC](https://discourse.numenta.org/t/how-do-htm-learn-a-sequence-of-the-same-input/11364 "2024-05-22T19:04:46Z")

</div>

So suppose I have 6 mini-columns, each has 6 cells. And my input is all the same. i.e, the sequence is just AAAAAA… And because each cell can’t be in both predicted and active states, we will have to use at least two ce…

---

## [\[Seeking helps\] Questions about 2016 Neuron Paper](https://discourse.numenta.org/t/seeking-helps-questions-about-2016-neuron-paper/11332)

<div class="topic-metadata">

**Author:** [@Hui\_Liu](https://discourse.numenta.org/u/Hui_Liu)\
**Replies:** 6\
**Last updated:** [May 8, 2024, 3:36am UTC](https://discourse.numenta.org/t/seeking-helps-questions-about-2016-neuron-paper/11332 "2024-05-08T03:36:23Z")

</div>

The 2016 Neuron Paper is very inspiring, but I still have a few questions after reading it and some posts in the forum, specifically: for the learning/update rule for cells in the active column (equation (6) in the pap…

---

## [Accuracy is not improving while trying to train htm model for binary classification](https://discourse.numenta.org/t/accuracy-is-not-improving-while-trying-to-train-htm-model-for-binary-classification/11127)

<div class="topic-metadata">

**Author:** [@sasikiran](https://discourse.numenta.org/u/sasikiran)\
**Replies:** 1\
**Last updated:** [December 20, 2023, 11:43pm UTC](https://discourse.numenta.org/t/accuracy-is-not-improving-while-trying-to-train-htm-model-for-binary-classification/11127 "2023-12-20T23:43:16Z")

</div>

Hi Team, I am new to HTM and trying to run HTM model for Anomaly detection use case (wanted to train it as binary classification), took the MNIST example as reference tried to build the model (only difference is being, …

---

## [HTM + Logic for sequence learning](https://discourse.numenta.org/t/htm-logic-for-sequence-learning/11096)

<div class="topic-metadata">

**Author:** [@bowen](https://discourse.numenta.org/u/bowen)\
**Replies:** 2\
**Last updated:** [November 16, 2023, 7:08pm UTC](https://discourse.numenta.org/t/htm-logic-for-sequence-learning/11096 "2023-11-16T19:08:30Z")

</div>

Hi all, I’m excited to share the work that combines HTM with Non-Axomatic Logic: https://arxiv.org/pdf/2308.12486.pdf The model is highly inspired by HTM, especially the mini-column structure. However, it takes a fully…

---

## [Growing Temporal Memory (GTM)](https://discourse.numenta.org/t/growing-temporal-memory-gtm/10857)

<div class="topic-metadata">

**Author:** [@cezar\_t](https://discourse.numenta.org/u/cezar_t)\
**Replies:** 15\
**Last updated:** [July 7, 2023, 6:54pm UTC](https://discourse.numenta.org/t/growing-temporal-memory-gtm/10857 "2023-07-07T18:54:26Z")

</div>

TLDR Here I propose a TM block which instead of having a fixed depth (column size) it starts with all columns shallow and each column grows (stacks new cells) as necessary, how TM learning rules could be changed in ord…

---

## [Why am I unable to get tmRegion to learn a short repeating sequence](https://discourse.numenta.org/t/why-am-i-unable-to-get-tmregion-to-learn-a-short-repeating-sequence/10801)

<div class="topic-metadata">

**Author:** [@buckithed](https://discourse.numenta.org/u/buckithed)\
**Replies:** 0\
**Last updated:** [May 30, 2023, 6:38am UTC](https://discourse.numenta.org/t/why-am-i-unable-to-get-tmregion-to-learn-a-short-repeating-sequence/10801 "2023-05-30T06:38:48Z")

</div>

I created a simple network using the htmcore’s regions in python. It counts from 0 to 10, but the columns continue to burst because it thinks that the sequence is never ending. How can I reset the network every start of …

---

## [I created a program to visualize a cell activation in real time using OpenGL](https://discourse.numenta.org/t/i-created-a-program-to-visualize-a-cell-activation-in-real-time-using-opengl/10777)

<div class="topic-metadata">

**Author:** [@buckithed](https://discourse.numenta.org/u/buckithed)\
**Replies:** 0\
**Last updated:** [May 25, 2023, 5:37am UTC](https://discourse.numenta.org/t/i-created-a-program-to-visualize-a-cell-activation-in-real-time-using-opengl/10777 "2023-05-25T05:37:51Z")

</div>

This is just a simple matrix to visualize a network from OpenGL thread in real time, since I saw no easy way to do so from HTM.core. Bottom 2 layers are scalar encoders. The rest are columns of 5. Yellow = active. Red = …

---

## [Why are the bits of a spatial pooler called columns?](https://discourse.numenta.org/t/why-are-the-bits-of-a-spatial-pooler-called-columns/10665)

<div class="topic-metadata">

**Author:** [@buckithed](https://discourse.numenta.org/u/buckithed)\
**Replies:** 2\
**Last updated:** [April 18, 2023, 6:46pm UTC](https://discourse.numenta.org/t/why-are-the-bits-of-a-spatial-pooler-called-columns/10665 "2023-04-18T18:46:16Z")

</div>

I think I understand pretty well how the spatial pooling works, but I still haven’t figured out why the output of the SDR is always referred to as columns. I understand why the temporal memory has columns. Is it named co…

---

## [Alternative sequence memory](https://discourse.numenta.org/t/alternative-sequence-memory/3898)

<div class="topic-metadata">

**Author:** [@sebjwallace](https://discourse.numenta.org/u/sebjwallace)\
**Replies:** 34\
**Last updated:** [February 3, 2023, 9:50pm UTC](https://discourse.numenta.org/t/alternative-sequence-memory/3898 "2023-02-03T21:50:52Z")

</div>

Note: I am not suggesting the following method is a biologically plausible mechanism How could a sequence of SDRs be stored in memory and be retrieved from any point? How could SDRs be stored in a sequence while maintai…

---

## [Building up longer term contextual representations with HTM](https://discourse.numenta.org/t/building-up-longer-term-contextual-representations-with-htm/10469)

<div class="topic-metadata">

**Author:** [@seanjhardy](https://discourse.numenta.org/u/seanjhardy)\
**Replies:** 9\
**Last updated:** [February 2, 2023, 7:33pm UTC](https://discourse.numenta.org/t/building-up-longer-term-contextual-representations-with-htm/10469 "2023-02-02T19:33:36Z")

</div>

Li et al. writes “higher levels of cortical organisation increasingly integrate information over longer timescales, similar to how higher layers in the visual cortex correspond to larger spatial patterns.” (https://www.p…

---

## [The Static Input Problem](https://discourse.numenta.org/t/the-static-input-problem/10447)

<div class="topic-metadata">

**Author:** [@dmac](https://discourse.numenta.org/u/dmac)\
**Replies:** 7\
**Last updated:** [January 27, 2023, 12:39am UTC](https://discourse.numenta.org/t/the-static-input-problem/10447 "2023-01-27T00:39:05Z")

</div>

This post describes a subtle flaw in the Temporal Memory algorithm and suggests a way to fix it. Background When the temporal memory (TM) encounters a sequence that it has never seen before, it assigns a unique set of c…

---

## [Preferred way to pass recognized sequences from Temporal Memory to a "higher HTM layer"?](https://discourse.numenta.org/t/preferred-way-to-pass-recognized-sequences-from-temporal-memory-to-a-higher-htm-layer/10137)

<div class="topic-metadata">

**Author:** [@Markus](https://discourse.numenta.org/u/Markus)\
**Replies:** 5\
**Last updated:** [November 8, 2022, 1:47pm UTC](https://discourse.numenta.org/t/preferred-way-to-pass-recognized-sequences-from-temporal-memory-to-a-higher-htm-layer/10137 "2022-11-08T13:47:48Z")

</div>

Say I have trained a Temporal Memory instance with letters (encoded as SDRs) to distinguish sequences like “GEC” or “DEF” and predict the letter C after GE or the letter F after input of DE, something like it was seen in…

---

## [Sequence learning, prediction, and replay in networks of spiking neurons](https://discourse.numenta.org/t/sequence-learning-prediction-and-replay-in-networks-of-spiking-neurons/9130)

<div class="topic-metadata">

**Author:** [@younes\_bouhadjar](https://discourse.numenta.org/u/younes_bouhadjar)\
**Replies:** 4\
**Last updated:** [August 3, 2022, 8:07pm UTC](https://discourse.numenta.org/t/sequence-learning-prediction-and-replay-in-networks-of-spiking-neurons/9130 "2022-08-03T20:07:40Z")

</div>

Our work on extending the temporal memory (TM) component of HTM is now available at \[2111.03456\] Sequence learning, prediction, and replay in networks of spiking neurons. In this work, we demonstrate that the principle m…

---

## [Top-down context for the sensorimotor layer (L4)](https://discourse.numenta.org/t/top-down-context-for-the-sensorimotor-layer-l4/9938)

<div class="topic-metadata">

**Author:** [@Dmitri.Chertov](https://discourse.numenta.org/u/Dmitri.Chertov)\
**Replies:** 2\
**Last updated:** [July 28, 2022, 1:41pm UTC](https://discourse.numenta.org/t/top-down-context-for-the-sensorimotor-layer-l4/9938 "2022-07-28T13:41:29Z")

</div>

I am studying Numenta’s papers and am currently reading “Untangling Sequences: Behavior vs. External Causes”. According to the article, a stable representation of an object is studied on L2/3 becomes the top-down contex…

---

## [Spatial pooler training and Temporal Memory](https://discourse.numenta.org/t/spatial-pooler-training-and-temporal-memory/9879)

<div class="topic-metadata">

**Author:** [@Adam\_Brooks](https://discourse.numenta.org/u/Adam_Brooks)\
**Replies:** 0\
**Last updated:** [June 29, 2022, 7:52pm UTC](https://discourse.numenta.org/t/spatial-pooler-training-and-temporal-memory/9879 "2022-06-29T19:52:54Z")

</div>

I have created a spatial pooler and temporal memory following the guide in BAMI. Can anyone comment on how best to train them? What I mean is should I train the spatial pooler for a set number of iterations or is it be…

---

## [Training of Basal dendrites](https://discourse.numenta.org/t/training-of-basal-dendrites/9843)

<div class="topic-metadata">

**Author:** [@Dmitri.Chertov](https://discourse.numenta.org/u/Dmitri.Chertov)\
**Replies:** 4\
**Last updated:** [June 22, 2022, 3:34pm UTC](https://discourse.numenta.org/t/training-of-basal-dendrites/9843 "2022-06-22T15:34:16Z")

</div>

I’m studying how Temporal Memory works and I don’t fully understand how exactly basal dendrites learning to predict the next input. If I understand correctly, in order for the basal dendrites of the winner cell at time …

---

## [Speed of Spatial Pooler and Temporal Memory](https://discourse.numenta.org/t/speed-of-spatial-pooler-and-temporal-memory/9675)

<div class="topic-metadata">

**Author:** [@TheFinn](https://discourse.numenta.org/u/TheFinn)\
**Replies:** 11\
**Last updated:** [May 10, 2022, 7:06pm UTC](https://discourse.numenta.org/t/speed-of-spatial-pooler-and-temporal-memory/9675 "2022-05-10T19:06:17Z")

</div>

I’ve hacked together a proof of concept script from the hot gym example using the HTM community fork of htm.core. My goal is to eventually use HTMs to look for anomalies in multiple streams of data. The first hurdle I …

---

## [My analysis on why Temporal Memory prediction doesn't work on sequential data](https://discourse.numenta.org/t/my-analysis-on-why-temporal-memory-prediction-doesnt-work-on-sequential-data/3141)

<div class="topic-metadata">

**Author:** [@oiegorov](https://discourse.numenta.org/u/oiegorov)\
**Replies:** 58\
**Last updated:** [February 2, 2020, 3:37pm UTC](https://discourse.numenta.org/t/my-analysis-on-why-temporal-memory-prediction-doesnt-work-on-sequential-data/3141 "2020-02-02T15:37:45Z")

</div>

First of all, I hope that my analysis is flawed, and someone points out what I’ve got wrong. I like a lot Jeff Hawkings’ theory on how the brains work, and really want to see HTM solving real-world problems! However, I t…

---

## [Are direction selective cells just normal HTM cells?](https://discourse.numenta.org/t/are-direction-selective-cells-just-normal-htm-cells/9020)

<div class="topic-metadata">

**Author:** [@JarvisGoBrr](https://discourse.numenta.org/u/JarvisGoBrr)\
**Replies:** 0\
**Last updated:** [October 3, 2021, 11:53am UTC](https://discourse.numenta.org/t/are-direction-selective-cells-just-normal-htm-cells/9020 "2021-10-03T11:53:16Z")

</div>

So, I heard there are some experiments where they probed a cat’s visual cortex and found lots of motion selective cells, I was wondering what it could mean and just come to the realization that the cells in the temporal …

---

## [HTM Anomaly Detection on image encodings problem](https://discourse.numenta.org/t/htm-anomaly-detection-on-image-encodings-problem/9001)

<div class="topic-metadata">

**Author:** [@vladim0105](https://discourse.numenta.org/u/vladim0105)\
**Replies:** 0\
**Last updated:** [September 30, 2021, 2:08pm UTC](https://discourse.numenta.org/t/htm-anomaly-detection-on-image-encodings-problem/9001 "2021-09-30T14:08:47Z")

</div>

Hi, I am performing some anomaly detection on image encodings, the idea being that images classified as “scorpions” will have similar latent representation to eachother in a CNN, while “cars” will have a different laten…

---

## [Different decay rates for neurons' predictive state](https://discourse.numenta.org/t/different-decay-rates-for-neurons-predictive-state/8890)

<div class="topic-metadata">

**Author:** [@LukeTPeterson](https://discourse.numenta.org/u/LukeTPeterson)\
**Replies:** 0\
**Last updated:** [August 22, 2021, 2:39am UTC](https://discourse.numenta.org/t/different-decay-rates-for-neurons-predictive-state/8890 "2021-08-22T02:39:33Z")

</div>

Hello, I was thinking through @mraptor 's thoughts on reinforcement learning here: Anyway, it occurred to me that a lot of mysteries\* are resolved and complexity eliminated if different neurons’ predictive states are …

---

## [How do you know how many temporal patterns can a TemporalMemory module store?](https://discourse.numenta.org/t/how-do-you-know-how-many-temporal-patterns-can-a-temporalmemory-module-store/8774)

<div class="topic-metadata">

**Author:** [@anunezr](https://discourse.numenta.org/u/anunezr)\
**Replies:** 0\
**Last updated:** [July 17, 2021, 10:23am UTC](https://discourse.numenta.org/t/how-do-you-know-how-many-temporal-patterns-can-a-temporalmemory-module-store/8774 "2021-07-17T10:23:06Z")

</div>

How can you calculate, for a particular htm.bindings.algorithms.TemporalMemory configuration, how many temporal patterns it can store? What parameters have an influence on that? How can you know how many of them are on…

---

## [Application of HTM to Video Processing Problems](https://discourse.numenta.org/t/application-of-htm-to-video-processing-problems/8623)

<div class="topic-metadata">

**Author:** [@aklandi](https://discourse.numenta.org/u/aklandi)\
**Replies:** 0\
**Last updated:** [June 2, 2021, 6:20pm UTC](https://discourse.numenta.org/t/application-of-htm-to-video-processing-problems/8623 "2021-06-02T18:20:27Z")

</div>

Hello! I am interested in sequence prediction as it pertains to video frames. Supposedly a solution to this problem is important for: (a) completing missing information – whether just missing or because of noise, or (b…

---

## [How are the cells per columns and length of sequence related?](https://discourse.numenta.org/t/how-are-the-cells-per-columns-and-length-of-sequence-related/6413)

<div class="topic-metadata">

**Author:** [@david.pfx](https://discourse.numenta.org/u/david.pfx)\
**Replies:** 33\
**Last updated:** [May 9, 2021, 1:26am UTC](https://discourse.numenta.org/t/how-are-the-cells-per-columns-and-length-of-sequence-related/6413 "2021-05-09T01:26:24Z")

</div>

The videos show consistently 4 cells per column, while the hotgym example appears to actually use 13. What effect does it have? Specifically, does the number of cells per column correlate with the length of sequence tha…

---

## [(How) Does HTM distinguish procedural memory and declarative memory?](https://discourse.numenta.org/t/how-does-htm-distinguish-procedural-memory-and-declarative-memory/8377)

<div class="topic-metadata">

**Author:** [@CLQuantizer](https://discourse.numenta.org/u/CLQuantizer)\
**Replies:** 0\
**Last updated:** [March 31, 2021, 12:51pm UTC](https://discourse.numenta.org/t/how-does-htm-distinguish-procedural-memory-and-declarative-memory/8377 "2021-03-31T12:51:59Z")

</div>

Here is the paper that put forward the distinction. Anderson, J. R., Reder, L. M., & Lebiere, C. (1996). Working memory: Activation limitations on retrieval. Cognitive Psychology, 30 (3), 221–256. https://doi.org/10.100…

---

## [Hindsight replay in reinforcement learning](https://discourse.numenta.org/t/hindsight-replay-in-reinforcement-learning/8249)

<div class="topic-metadata">

**Author:** [@INNFINITEMINDS](https://discourse.numenta.org/u/INNFINITEMINDS)\
**Replies:** 0\
**Last updated:** [February 20, 2021, 3:45am UTC](https://discourse.numenta.org/t/hindsight-replay-in-reinforcement-learning/8249 "2021-02-20T03:45:48Z")

</div>

What is the biological basis of Hindsight replay in reinforcement learning and the biological equivalent of action replay mode? (in respect to Neuron level/Neuron populations level /neuron network level) Thanks Arpit

[Next page](https://discourse.numenta.org/tag/sequence-memory/2.md?match_all_tags=true&page=1&tags%5B%5D=sequence-memory)